Skip to content

[JIT] Enable EGPRs in JIT by adding REX2 encoding to the backend. - #106557

Merged
BruceForstall merged 41 commits into
dotnet:mainfrom
Ruihan-Yin:apx-rex2-july
Dec 17, 2024
Merged

[JIT] Enable EGPRs in JIT by adding REX2 encoding to the backend.#106557
BruceForstall merged 41 commits into
dotnet:mainfrom
Ruihan-Yin:apx-rex2-july

Conversation

@Ruihan-Yin

@Ruihan-YinRuihan-Yin commented Aug 16, 2024

Copy link
Copy Markdown
Member

Overview

This PR is the follow-up PR after #104637, which added the initial CPUID and XSAVE updates for APX.

This PR adds REX2 encoding functionality for legacy instructions which enables the use of EGPR for add, sub, etc. Note that this PR focuses on REX2 encoding only: a follow up PR will enable EGPR support via the register allocator.

Specification

REX2 is a 2-byte prefix with a leading byte of 0xD5, detailed format below:
rex2

Similar to REX prefix, it provides the extended bits for the MODRM.REG field, REX2.R4/R3, and MODRM.R/M field, REX2.B4/B3, and the index register in SIB byte, REX2.X4/X3, those bits will act as the higher 5th/4th bits and combine with the field in MODRM and SIB byte as a 5-bit binary to access up to 32 registers.

REX2 prefix is generally available for legacy-map-0 and legacy-map-1 instructions, say 1-byte opcode or 2-byte opcode with escape byte 0x0F, with some exceptions.

Like VEX/EVEX, REX2 is considered as the last prefix before the main opcode, so it can not co-exist with REX/VEX/EVEX.

Design

The bulk of the changes occur in the backend emitter.

As there is no existing hardware that has APX support yet, we had some hacks to bypass the CPUID checks. In this PR, DOTNET_JitStressRex2Encoding will force all the eligible instructions to be encoded in REX2, regardless the presence of EGPRs in the operand. We had another switch DOTNET_JitBypassAPXCheck, with which will only bypass the APX CPUID check but JIT will encode REX2 only if needed, this is more useful when the LSRA changes come.

Note: REX2 can be used to address the lower 16 vector registers (XMM0~XMM15). But in this PR, we are not planning to add the support for this part now for simplicity, and the EGPRs functionality for SIMD instructions can be achieved with EVEX, we are open to discuss this part and tweak the design in the follow-up PRs.

Testing

We followed a multi-step testing plan to verify the encoding correctness and the semantic correctness.

Testing results will be presented below.

1. Emitter unit tests

In codgenxarch.cpp, similar to genAmd64EmitterUnitTestsSse2, we used the JitLateDisasm feature to insert instructions to encode as unit tests for emitter, and LateDisasm will invoke LLVM to disasm the code stream, this gave us the chance to cross validate the disassembly from JIT and LLVM. The output of this step is to verify the emit paths are generating "correct" code that would not trigger #UD or have wrong semantics.

Note that we are using a custom coredistools.dll which uses a recent LLVM that supports APX decoding.

2. SuperPMI

In this step, we would run the SuperPMI pipeline to get the asmdiffs with REX2 on and off, the inputs are all the MCH files. This step will give us the chance to check if there is any assertion failure or internal error within JIT and since the pipeline will invoke coredistools.dll as well, so we can verify the encoding correctness in a larger scope.

To ensure the new changes will not hit the existing code path in terms of throughput, we ran tpdiff with base JIT to be the main branch where changes are based on, and diff JIT to be the one with all the REX2 changes.

3. JIT unit tests

The 2 steps mentioned above are mainly verifying the encoding correctness of the generated binary code. Then the last will examine the semantic correctness of the generated code, say since we are simply forcing all the compatible instructions to be encoded in REX2, so the original semantics should not change, so we expect exactly the same output with REX2 on/off.

We used the existing CoreCLR unit test set: JIT and run it in the Intel SDE emulator.

Follow-up plans

This PR is only intended to provide the REX2 encoding functionality to the JIT backend, in terms of how to properly use it, we are preparing another PR that includes the updates on LSRA such that JIT will be able to allocate EGPRs only when needed, and generate optimal code.

@ghostghost added the area-CodeGen-coreclr CLR JIT compiler in src/coreclr/src/jit and related components such as SuperPMI label Aug 16, 2024
@Ruihan-Yin

Ruihan-Yin commented Aug 16, 2024

Copy link
Copy Markdown
MemberAuthor

Testing results

1. Emitter unit tests

1-11-21-31-4

2. SuperPMI

2.1 AsmDiffs - REX2 off (No diffs expected)

Diffs are based on 2,830,588 contexts (1,185,269 MinOpts, 1,645,319 FullOpts).

MISSED contexts: base: 0 (0.00%), diff: 11 (0.00%)

Overall (-100 bytes)
CollectionBase size (bytes)Diff size (bytes)PerfScore in Diffs
coreclr_tests.run.windows.x64.checked.mch409,086,766-82-1.35%
libraries.pmi.windows.x64.checked.mch63,022,393+30.00%
smoke_tests.nativeaot.windows.x64.checked.mch5,023,568-210.00%
MinOpts (-40 bytes)
CollectionBase size (bytes)Diff size (bytes)PerfScore in Diffs
coreclr_tests.run.windows.x64.checked.mch287,081,075-40-2.31%
FullOpts (-60 bytes)
CollectionBase size (bytes)Diff size (bytes)PerfScore in Diffs
coreclr_tests.run.windows.x64.checked.mch122,005,691-420.00%
libraries.pmi.windows.x64.checked.mch62,909,432+30.00%
smoke_tests.nativeaot.windows.x64.checked.mch5,022,597-210.00%
Example diffs
coreclr_tests.run.windows.x64.checked.mch
-3 (-50.00%) : 509186.dasm - System.Runtime.Intrinsics.X86.Popcnt+X64:get_IsSupported():ubyte (FullOpts)
@@ -14,13 +14,13 @@
G_M37565_IG01: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref, nogc <-- Prolog IG
;; size=0 bbWeight=1 PerfScore 0.00
G_M37565_IG02: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref
- mov eax, 1- ;; size=5 bbWeight=1 PerfScore 0.25+ xor eax, eax+ ;; size=2 bbWeight=1 PerfScore 0.25
G_M37565_IG03: ; bbWeight=1, epilog, nogc, extend
ret ;; size=1 bbWeight=1 PerfScore 1.00
-; Total bytes of code 6, prolog size 0, PerfScore 1.25, instruction count 2, allocated bytes for code 6 (MethodHash=4c306d42) for method System.Runtime.Intrinsics.X86.Popcnt+X64:get_IsSupported():ubyte (FullOpts)+; Total bytes of code 3, prolog size 0, PerfScore 1.25, instruction count 2, allocated bytes for code 3 (MethodHash=4c306d42) for method System.Runtime.Intrinsics.X86.Popcnt+X64:get_IsSupported():ubyte (FullOpts)
; ============================================================
Unwind Info:
-3 (-50.00%) : 579101.dasm - Runtime_34587:get_PopcntX64IsSupported():ubyte (FullOpts)
@@ -14,13 +14,13 @@
G_M19947_IG01: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref, nogc <-- Prolog IG
;; size=0 bbWeight=1 PerfScore 0.00
G_M19947_IG02: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref
- mov eax, 1- ;; size=5 bbWeight=1 PerfScore 0.25+ xor eax, eax+ ;; size=2 bbWeight=1 PerfScore 0.25
G_M19947_IG03: ; bbWeight=1, epilog, nogc, extend
ret ;; size=1 bbWeight=1 PerfScore 1.00
-; Total bytes of code 6, prolog size 0, PerfScore 1.25, instruction count 2, allocated bytes for code 6 (MethodHash=801bb214) for method Runtime_34587:get_PopcntX64IsSupported():ubyte (FullOpts)+; Total bytes of code 3, prolog size 0, PerfScore 1.25, instruction count 2, allocated bytes for code 3 (MethodHash=801bb214) for method Runtime_34587:get_PopcntX64IsSupported():ubyte (FullOpts)
; ============================================================
Unwind Info:
-32 (-35.16%) : 510272.dasm - IntelHardwareIntrinsicTest.Program:TestEntryPoint():int (FullOpts)
@@ -12,8 +12,8 @@
;* V01 loc1 [V01 ] ( 0, 0 ) ref -> zero-ref class-hnd <<unknown class>>
; V02 OutArgs [V02 ] ( 1, 1 ) struct (32) [rsp+0x00] do-not-enreg[XS] addr-exposed "OutgoingArgSpace"
;* V03 tmp1 [V03 ] ( 0, 0 ) int -> zero-ref -; V04 tmp2 [V04,T01] ( 2, 0 ) ref -> rdx class-hnd single-def "impSpillSpecialSideEff" <<unknown class>>-; V05 tmp3 [V05,T02] ( 2, 0 ) int -> [rbp-0x04] do-not-enreg[M] EH-live+;* V04 tmp2 [V04 ] ( 0, 0 ) ref -> zero-ref class-hnd "impSpillSpecialSideEff" <<unknown class>>+;* V05 tmp3 [V05 ] ( 0, 0 ) int -> zero-ref 
; V06 PSPSym [V06,T00] ( 1, 1 ) long -> [rbp-0x10] do-not-enreg[V] "PSPSym"
;
; Lcl frame size = 48
@@ -30,41 +30,30 @@ G_M30609_IG02: ; bbWeight=0, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref
int3 ;; size=6 bbWeight=0 PerfScore 0.00
G_M30609_IG03: ; bbWeight=0, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref
- mov eax, dword ptr [rbp-0x04]- ;; size=3 bbWeight=0 PerfScore 0.00+ xor eax, eax+ ;; size=2 bbWeight=0 PerfScore 0.00
G_M30609_IG04: ; bbWeight=0, epilog, nogc, extend
add rsp, 48
pop rbp
ret ;; size=6 bbWeight=0 PerfScore 0.00
-G_M30609_IG05: ; bbWeight=0, gcrefRegs=0004 {rdx}, byrefRegs=0000 {}, byref, funclet prolog, nogc- ; gcrRegs +[rdx]+G_M30609_IG05: ; bbWeight=0, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref, funclet prolog, nogc
push rbp
sub rsp, 48
mov rbp, qword ptr [rcx+0x20]
mov qword ptr [rsp+0x20], rbp
lea rbp, [rbp+0x30]
;; size=18 bbWeight=0 PerfScore 0.00
-G_M30609_IG06: ; bbWeight=0, gcVars=0000000000000000 {}, gcrefRegs=0004 {rdx}, byrefRegs=0000 {}, gcvars, byref- mov rcx, 0xD1FFAB1E ; <unknown class>- call CORINFO_HELP_ISINSTANCEOFCLASS- ; gcrRegs -[rdx] +[rax]- ; gcr arg pop 0- xor ecx, ecx- mov edx, 100- test rax, rax- cmovne ecx, edx- mov dword ptr [rbp-0x04], ecx+G_M30609_IG06: ; bbWeight=0, gcVars=0000000000000000 {}, gcrefRegs=0000 {}, byrefRegs=0000 {}, gcvars, byref
lea rax, G_M30609_IG03
- ; gcrRegs -[rax]- ;; size=38 bbWeight=0 PerfScore 0.00+ ;; size=7 bbWeight=0 PerfScore 0.00
G_M30609_IG07: ; bbWeight=0, funclet epilog, nogc, extend
add rsp, 48
pop rbp
ret ;; size=6 bbWeight=0 PerfScore 0.00
-; Total bytes of code 91, prolog size 14, PerfScore 0.00, instruction count 26, allocated bytes for code 91 (MethodHash=3926886e) for method IntelHardwareIntrinsicTest.Program:TestEntryPoint():int (FullOpts)+; Total bytes of code 59, prolog size 14, PerfScore 0.00, instruction count 19, allocated bytes for code 59 (MethodHash=3926886e) for method IntelHardwareIntrinsicTest.Program:TestEntryPoint():int (FullOpts)
; ============================================================
Unwind Info:
-1 (-0.04%) : 510371.dasm - IntelHardwareIntrinsicTest.General.Program:IsSupported() (FullOpts)
@@ -915,13 +915,13 @@ G_M58490_IG59: ; bbWeight=0.50, gcrefRegs=0041 {rax rsi}, byrefRegs=0000
; gcrRegs +[rcx]
call [System.Convert:ToBoolean(System.Object):ubyte]
; gcrRegs -[rax rcx]
- cmp eax, 1+ test eax, eax
jne G_M58490_IG66
xor rcx, rcx
; gcrRegs +[rcx]
mov gword ptr [rsp+0x20], rcx
mov dword ptr [rsp+0x28], 3
- ;; size=59 bbWeight=0.50 PerfScore 7.88+ ;; size=58 bbWeight=0.50 PerfScore 7.88
G_M58490_IG60: ; bbWeight=0.50, gcrefRegs=0040 {rsi}, byrefRegs=0000 {}, byref
; gcrRegs -[rcx]
mov gword ptr [rsp+0x30], rcx
@@ -1041,7 +1041,7 @@ G_M58490_IG70: ; bbWeight=0, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref
int3 ;; size=28 bbWeight=0 PerfScore 0.00
-; Total bytes of code 2296, prolog size 27, PerfScore 264.21, instruction count 525, allocated bytes for code 2296 (MethodHash=a9531b85) for method IntelHardwareIntrinsicTest.General.Program:IsSupported() (FullOpts)+; Total bytes of code 2295, prolog size 27, PerfScore 264.21, instruction count 525, allocated bytes for code 2295 (MethodHash=a9531b85) for method IntelHardwareIntrinsicTest.General.Program:IsSupported() (FullOpts)
; ============================================================
Unwind Info:
-3 (-0.03%) : 579023.dasm - Runtime_34587:TestEntryPoint():int (FullOpts)
@@ -2680,7 +2680,7 @@ G_M52152_IG121: ; bbWeight=0.50, gcrefRegs=0008 {rbx}, byrefRegs=0040 {rs
G_M52152_IG122: ; bbWeight=1.00, gcrefRegs=0008 {rbx}, byrefRegs=0000 {}, byref
; byrRegs -[rsi]
lea rcx, [rsp+0x50]
- mov edx, 1+ xor edx, edx
call [<unknown method>]
; gcr arg pop 0
lea rcx, [rsp+0x50]
@@ -2703,7 +2703,7 @@ G_M52152_IG122: ; bbWeight=1.00, gcrefRegs=0008 {rbx}, byrefRegs=0000 {},
mov gword ptr [rsp+0x58], rax
test rax, rax
je G_M52152_IG205
- ;; size=71 bbWeight=1.00 PerfScore 17.50+ ;; size=68 bbWeight=1.00 PerfScore 17.50
G_M52152_IG123: ; bbWeight=0.50, gcrefRegs=0001 {rax}, byrefRegs=0000 {}, byref
lea rsi, bword ptr [rax+0x10]
; byrRegs +[rsi]
@@ -4142,7 +4142,7 @@ G_M52152_IG207: ; bbWeight=0, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref
int3 ;; size=7 bbWeight=0 PerfScore 0.00
-; Total bytes of code 9099, prolog size 37, PerfScore 1583.63, instruction count 1959, allocated bytes for code 9099 (MethodHash=30563447) for method Runtime_34587:TestEntryPoint():int (FullOpts)+; Total bytes of code 9096, prolog size 37, PerfScore 1583.63, instruction count 1959, allocated bytes for code 9096 (MethodHash=30563447) for method Runtime_34587:TestEntryPoint():int (FullOpts)
; ============================================================
Unwind Info:
+4 (+1.99%) : 205245.dasm - IntelHardwareIntrinsicTest.Program:TestEntryPoint():int (MinOpts)
@@ -39,14 +39,13 @@ G_M30609_IG03: ; bbWeight=0.50, gcrefRegs=0000 {}, byrefRegs=0000 {}, byr
;; size=5 bbWeight=0.50 PerfScore 0.50
G_M30609_IG04: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref, isz
nop - xor eax, eax- mov dword ptr [rbp-0x18], eax+ mov dword ptr [rbp-0x18], 1
cmp dword ptr [rbp-0x18], 0
jne SHORT G_M30609_IG05
xor eax, eax
mov dword ptr [rbp-0x1C], eax
jmp SHORT G_M30609_IG06
- ;; size=19 bbWeight=1 PerfScore 7.75+ ;; size=21 bbWeight=1 PerfScore 7.50
G_M30609_IG05: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref
mov dword ptr [rbp-0x1C], 100
;; size=7 bbWeight=1 PerfScore 1.00
@@ -87,14 +86,12 @@ G_M30609_IG11: ; bbWeight=1, gcVars=0000000000000000 {}, gcrefRegs=0004 {
; gcrRegs +[rax]
mov gword ptr [rbp-0x10], rax
nop - xor eax, eax- ; gcrRegs -[rax]- mov dword ptr [rbp-0x2C], eax+ mov dword ptr [rbp-0x2C], 1
cmp dword ptr [rbp-0x2C], 0
jne SHORT G_M30609_IG13
- ;; size=24 bbWeight=1 PerfScore 7.50+ ;; size=26 bbWeight=1 PerfScore 7.25
G_M30609_IG12: ; bbWeight=0, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref, isz
- ; gcrRegs -[rdx]+ ; gcrRegs -[rax rdx]
mov rdx, gword ptr [rbp-0x10]
; gcrRegs +[rdx]
mov rcx, 0xD1FFAB1E ; <unknown class>
@@ -124,7 +121,7 @@ G_M30609_IG15: ; bbWeight=0, funclet epilog, nogc, extend
ret ;; size=6 bbWeight=0 PerfScore 0.00
-; Total bytes of code 201, prolog size 28, PerfScore 43.58, instruction count 63, allocated bytes for code 201 (MethodHash=3926886e) for method IntelHardwareIntrinsicTest.Program:TestEntryPoint():int (MinOpts)+; Total bytes of code 205, prolog size 28, PerfScore 43.08, instruction count 61, allocated bytes for code 205 (MethodHash=3926886e) for method IntelHardwareIntrinsicTest.Program:TestEntryPoint():int (MinOpts)
; ============================================================
Unwind Info:
libraries.pmi.windows.x64.checked.mch
-3 (-50.00%) : 29280.dasm - System.Runtime.Intrinsics.X86.Popcnt+X64:get_IsSupported():ubyte (FullOpts)
@@ -14,13 +14,13 @@
G_M37565_IG01: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref, nogc <-- Prolog IG
;; size=0 bbWeight=1 PerfScore 0.00
G_M37565_IG02: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref
- mov eax, 1- ;; size=5 bbWeight=1 PerfScore 0.25+ xor eax, eax+ ;; size=2 bbWeight=1 PerfScore 0.25
G_M37565_IG03: ; bbWeight=1, epilog, nogc, extend
ret ;; size=1 bbWeight=1 PerfScore 1.00
-; Total bytes of code 6, prolog size 0, PerfScore 1.25, instruction count 2, allocated bytes for code 6 (MethodHash=4c306d42) for method System.Runtime.Intrinsics.X86.Popcnt+X64:get_IsSupported():ubyte (FullOpts)+; Total bytes of code 3, prolog size 0, PerfScore 1.25, instruction count 2, allocated bytes for code 3 (MethodHash=4c306d42) for method System.Runtime.Intrinsics.X86.Popcnt+X64:get_IsSupported():ubyte (FullOpts)
; ============================================================
Unwind Info:
+3 (+100.00%) : 29884.dasm - System.Runtime.Intrinsics.X86.X86Serialize+X64:get_IsSupported():ubyte (FullOpts)
@@ -14,13 +14,13 @@
G_M34763_IG01: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref, nogc <-- Prolog IG
;; size=0 bbWeight=1 PerfScore 0.00
G_M34763_IG02: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref
- xor eax, eax- ;; size=2 bbWeight=1 PerfScore 0.25+ mov eax, 1+ ;; size=5 bbWeight=1 PerfScore 0.25
G_M34763_IG03: ; bbWeight=1, epilog, nogc, extend
ret ;; size=1 bbWeight=1 PerfScore 1.00
-; Total bytes of code 3, prolog size 0, PerfScore 1.25, instruction count 2, allocated bytes for code 3 (MethodHash=bcd77834) for method System.Runtime.Intrinsics.X86.X86Serialize+X64:get_IsSupported():ubyte (FullOpts)+; Total bytes of code 6, prolog size 0, PerfScore 1.25, instruction count 2, allocated bytes for code 6 (MethodHash=bcd77834) for method System.Runtime.Intrinsics.X86.X86Serialize+X64:get_IsSupported():ubyte (FullOpts)
; ============================================================
Unwind Info:
+3 (+100.00%) : 29207.dasm - System.Runtime.Intrinsics.X86.AvxVnni+X64:get_IsSupported():ubyte (FullOpts)
@@ -14,13 +14,13 @@
G_M31227_IG01: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref, nogc <-- Prolog IG
;; size=0 bbWeight=1 PerfScore 0.00
G_M31227_IG02: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref
- xor eax, eax- ;; size=2 bbWeight=1 PerfScore 0.25+ mov eax, 1+ ;; size=5 bbWeight=1 PerfScore 0.25
G_M31227_IG03: ; bbWeight=1, epilog, nogc, extend
ret ;; size=1 bbWeight=1 PerfScore 1.00
-; Total bytes of code 3, prolog size 0, PerfScore 1.25, instruction count 2, allocated bytes for code 3 (MethodHash=e6278604) for method System.Runtime.Intrinsics.X86.AvxVnni+X64:get_IsSupported():ubyte (FullOpts)+; Total bytes of code 6, prolog size 0, PerfScore 1.25, instruction count 2, allocated bytes for code 6 (MethodHash=e6278604) for method System.Runtime.Intrinsics.X86.AvxVnni+X64:get_IsSupported():ubyte (FullOpts)
; ============================================================
Unwind Info:
smoke_tests.nativeaot.windows.x64.checked.mch
-3 (-50.00%) : 14296.dasm - Program:X86SerializeX64IsSupported():ubyte (FullOpts)
@@ -14,13 +14,13 @@
G_M13406_IG01: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref, nogc <-- Prolog IG
;; size=0 bbWeight=1 PerfScore 0.00
G_M13406_IG02: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref
- mov eax, 1- ;; size=5 bbWeight=1 PerfScore 0.25+ xor eax, eax+ ;; size=2 bbWeight=1 PerfScore 0.25
G_M13406_IG03: ; bbWeight=1, epilog, nogc, extend
ret ;; size=1 bbWeight=1 PerfScore 1.00
-; Total bytes of code 6, prolog size 0, PerfScore 1.25, instruction count 2, allocated bytes for code 6 (MethodHash=00a5cba1) for method Program:X86SerializeX64IsSupported():ubyte (FullOpts)+; Total bytes of code 3, prolog size 0, PerfScore 1.25, instruction count 2, allocated bytes for code 3 (MethodHash=00a5cba1) for method Program:X86SerializeX64IsSupported():ubyte (FullOpts)
; ============================================================
Unwind Info:
-3 (-50.00%) : 21532.dasm - Program:AesX64IsSupported():ubyte (FullOpts)
@@ -14,13 +14,13 @@
G_M55817_IG01: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref, nogc <-- Prolog IG
;; size=0 bbWeight=1 PerfScore 0.00
G_M55817_IG02: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref
- mov eax, 1- ;; size=5 bbWeight=1 PerfScore 0.25+ xor eax, eax+ ;; size=2 bbWeight=1 PerfScore 0.25
G_M55817_IG03: ; bbWeight=1, epilog, nogc, extend
ret ;; size=1 bbWeight=1 PerfScore 1.00
-; Total bytes of code 6, prolog size 0, PerfScore 1.25, instruction count 2, allocated bytes for code 6 (MethodHash=c5da25f6) for method Program:AesX64IsSupported():ubyte (FullOpts)+; Total bytes of code 3, prolog size 0, PerfScore 1.25, instruction count 2, allocated bytes for code 3 (MethodHash=c5da25f6) for method Program:AesX64IsSupported():ubyte (FullOpts)
; ============================================================
Unwind Info:
-3 (-50.00%) : 19229.dasm - Program:X86SerializeX64IsSupported():ubyte (FullOpts)
@@ -14,13 +14,13 @@
G_M13406_IG01: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref, nogc <-- Prolog IG
;; size=0 bbWeight=1 PerfScore 0.00
G_M13406_IG02: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref
- mov eax, 1- ;; size=5 bbWeight=1 PerfScore 0.25+ xor eax, eax+ ;; size=2 bbWeight=1 PerfScore 0.25
G_M13406_IG03: ; bbWeight=1, epilog, nogc, extend
ret ;; size=1 bbWeight=1 PerfScore 1.00
-; Total bytes of code 6, prolog size 0, PerfScore 1.25, instruction count 2, allocated bytes for code 6 (MethodHash=00a5cba1) for method Program:X86SerializeX64IsSupported():ubyte (FullOpts)+; Total bytes of code 3, prolog size 0, PerfScore 1.25, instruction count 2, allocated bytes for code 3 (MethodHash=00a5cba1) for method Program:X86SerializeX64IsSupported():ubyte (FullOpts)
; ============================================================
Unwind Info:
+3 (+100.00%) : 19199.dasm - Program:AvxVnniX64IsSupported():ubyte (FullOpts)
@@ -14,13 +14,13 @@
G_M60430_IG01: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref, nogc <-- Prolog IG
;; size=0 bbWeight=1 PerfScore 0.00
G_M60430_IG02: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref
- xor eax, eax- ;; size=2 bbWeight=1 PerfScore 0.25+ mov eax, 1+ ;; size=5 bbWeight=1 PerfScore 0.25
G_M60430_IG03: ; bbWeight=1, epilog, nogc, extend
ret ;; size=1 bbWeight=1 PerfScore 1.00
-; Total bytes of code 3, prolog size 0, PerfScore 1.25, instruction count 2, allocated bytes for code 3 (MethodHash=e20b13f1) for method Program:AvxVnniX64IsSupported():ubyte (FullOpts)+; Total bytes of code 6, prolog size 0, PerfScore 1.25, instruction count 2, allocated bytes for code 6 (MethodHash=e20b13f1) for method Program:AvxVnniX64IsSupported():ubyte (FullOpts)
; ============================================================
Unwind Info:
+3 (+100.00%) : 21515.dasm - Program:FmaX64IsSupported():ubyte (FullOpts)
@@ -14,13 +14,13 @@
G_M2260_IG01: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref, nogc <-- Prolog IG
;; size=0 bbWeight=1 PerfScore 0.00
G_M2260_IG02: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref
- xor eax, eax- ;; size=2 bbWeight=1 PerfScore 0.25+ mov eax, 1+ ;; size=5 bbWeight=1 PerfScore 0.25
G_M2260_IG03: ; bbWeight=1, epilog, nogc, extend
ret ;; size=1 bbWeight=1 PerfScore 1.00
-; Total bytes of code 3, prolog size 0, PerfScore 1.25, instruction count 2, allocated bytes for code 3 (MethodHash=36a7f72b) for method Program:FmaX64IsSupported():ubyte (FullOpts)+; Total bytes of code 6, prolog size 0, PerfScore 1.25, instruction count 2, allocated bytes for code 6 (MethodHash=36a7f72b) for method Program:FmaX64IsSupported():ubyte (FullOpts)
; ============================================================
Unwind Info:
+3 (+100.00%) : 21526.dasm - Program:Avx2X64IsSupported():ubyte (FullOpts)
@@ -14,13 +14,13 @@
G_M13187_IG01: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref, nogc <-- Prolog IG
;; size=0 bbWeight=1 PerfScore 0.00
G_M13187_IG02: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref
- xor eax, eax- ;; size=2 bbWeight=1 PerfScore 0.25+ mov eax, 1+ ;; size=5 bbWeight=1 PerfScore 0.25
G_M13187_IG03: ; bbWeight=1, epilog, nogc, extend
ret ;; size=1 bbWeight=1 PerfScore 1.00
-; Total bytes of code 3, prolog size 0, PerfScore 1.25, instruction count 2, allocated bytes for code 3 (MethodHash=f683cc7c) for method Program:Avx2X64IsSupported():ubyte (FullOpts)+; Total bytes of code 6, prolog size 0, PerfScore 1.25, instruction count 2, allocated bytes for code 6 (MethodHash=f683cc7c) for method Program:Avx2X64IsSupported():ubyte (FullOpts)
; ============================================================
Unwind Info:
Details

Size improvements/regressions per collection

CollectionContexts with diffsImprovementsRegressionsSame sizeImprovements (bytes)Regressions (bytes)
aspnet.run.windows.x64.checked.mch0000-0+0
benchmarks.run.windows.x64.checked.mch0000-0+0
benchmarks.run_pgo.windows.x64.checked.mch0000-0+0
benchmarks.run_tiered.windows.x64.checked.mch0000-0+0
coreclr_tests.run.windows.x64.checked.mch121110-86+4
libraries.crossgen2.windows.x64.checked.mch0000-0+0
libraries.pmi.windows.x64.checked.mch3120-3+6
libraries_tests.run.windows.x64.Release.mch0000-0+0
libraries_tests_no_tiered_compilation.run.windows.x64.Release.mch0000-0+0
realworld.run.windows.x64.checked.mch0000-0+0
smoke_tests.nativeaot.windows.x64.checked.mch171250-36+15
322480-125+25

PerfScore improvements/regressions per collection

CollectionContexts with diffsImprovementsRegressionsSame PerfScoreImprovements (PerfScore)Regressions (PerfScore)PerfScore Overall in FullOpts
aspnet.run.windows.x64.checked.mch00000.00%0.00%0.0000%
benchmarks.run.windows.x64.checked.mch00000.00%0.00%0.0000%
benchmarks.run_pgo.windows.x64.checked.mch00000.00%0.00%0.0000%
benchmarks.run_tiered.windows.x64.checked.mch00000.00%0.00%0.0000%
coreclr_tests.run.windows.x64.checked.mch12219-7.87%+0.03%0.0000%
libraries.crossgen2.windows.x64.checked.mch00000.00%0.00%0.0000%
libraries.pmi.windows.x64.checked.mch30030.00%0.00%0.0000%
libraries_tests.run.windows.x64.Release.mch00000.00%0.00%0.0000%
libraries_tests_no_tiered_compilation.run.windows.x64.Release.mch00000.00%0.00%0.0000%
realworld.run.windows.x64.checked.mch00000.00%0.00%0.0000%
smoke_tests.nativeaot.windows.x64.checked.mch1700170.00%0.00%0.0000%

Context information

CollectionDiffed contextsMinOptsFullOptsMissed, baseMissed, diff
aspnet.run.windows.x64.checked.mch141,22477,32463,9000 (0.00%)0 (0.00%)
benchmarks.run.windows.x64.checked.mch38,352638,3460 (0.00%)0 (0.00%)
benchmarks.run_pgo.windows.x64.checked.mch120,28068,10352,1770 (0.00%)0 (0.00%)
benchmarks.run_tiered.windows.x64.checked.mch76,87656,35820,5180 (0.00%)0 (0.00%)
coreclr_tests.run.windows.x64.checked.mch642,813393,776249,0370 (0.00%)5 (0.00%)
libraries.crossgen2.windows.x64.checked.mch276,88915276,8740 (0.00%)2 (0.00%)
libraries.pmi.windows.x64.checked.mch316,0106316,0040 (0.00%)1 (0.00%)
libraries_tests.run.windows.x64.Release.mch814,679567,674247,0050 (0.00%)0 (0.00%)
libraries_tests_no_tiered_compilation.run.windows.x64.Release.mch343,89521,994321,9010 (0.00%)0 (0.00%)
realworld.run.windows.x64.checked.mch28,368328,3650 (0.00%)0 (0.00%)
smoke_tests.nativeaot.windows.x64.checked.mch31,2021031,1920 (0.00%)3 (0.01%)
2,830,5881,185,2691,645,3190 (0.00%)11 (0.00%)

jit-analyze output

Comment:
SuperPMI pipeline with REX2 off:
Theoretically, it should be clean compared with the base corerun, the diff found here is because of the changes in the ISA definition, as it can be noticed, all the diffs are either from xor eax, eax to mov eax, 1, or in the reverse way, this is essentially indicating runtime is reporting discrepant ISA availability, and this is expected to be resolved when the public CPUID PR gets merged.

2.2 AsmDiffs - REX2 on

SuperPMI pipeline:

Diffs are based on 2,830,588 contexts (1,185,269 MinOpts, 1,645,319 FullOpts).

MISSED contexts: base: 0 (0.00%), diff: 11 (0.00%)

Diff JIT options: JitStressRex2Encoding=1

Overall (+243,564,575 bytes)
CollectionBase size (bytes)Diff size (bytes)PerfScore in Diffs
aspnet.run.windows.x64.checked.mch49,406,065+10,392,1790.00%
benchmarks.run.windows.x64.checked.mch12,230,572+3,013,3990.00%
benchmarks.run_pgo.windows.x64.checked.mch40,192,955+8,962,4740.00%
benchmarks.run_tiered.windows.x64.checked.mch17,606,620+4,199,7460.00%
coreclr_tests.run.windows.x64.checked.mch409,086,766+84,314,011-0.00%
libraries.crossgen2.windows.x64.checked.mch45,250,222+11,739,1390.00%
libraries.pmi.windows.x64.checked.mch63,022,393+15,125,7250.00%
libraries_tests.run.windows.x64.Release.mch336,307,360+69,356,3940.00%
libraries_tests_no_tiered_compilation.run.windows.x64.Release.mch147,986,092+32,521,9410.00%
realworld.run.windows.x64.checked.mch11,552,911+2,545,1260.00%
smoke_tests.nativeaot.windows.x64.checked.mch5,023,568+1,394,4410.00%
MinOpts (+113,554,608 bytes)
CollectionBase size (bytes)Diff size (bytes)PerfScore in Diffs
aspnet.run.windows.x64.checked.mch23,379,337+4,569,5100.00%
benchmarks.run.windows.x64.checked.mch588+1630.00%
benchmarks.run_pgo.windows.x64.checked.mch18,796,230+4,022,2390.00%
benchmarks.run_tiered.windows.x64.checked.mch13,707,415+3,160,9670.00%
coreclr_tests.run.windows.x64.checked.mch287,081,075+59,383,662-0.00%
libraries.crossgen2.windows.x64.checked.mch1,705+4420.00%
libraries.pmi.windows.x64.checked.mch112,961+15,3580.00%
libraries_tests.run.windows.x64.Release.mch203,705,533+39,847,5110.00%
libraries_tests_no_tiered_compilation.run.windows.x64.Release.mch10,696,900+2,483,9620.00%
realworld.run.windows.x64.checked.mch412,968+70,5900.00%
smoke_tests.nativeaot.windows.x64.checked.mch971+2040.00%
FullOpts (+130,009,967 bytes)
CollectionBase size (bytes)Diff size (bytes)PerfScore in Diffs
aspnet.run.windows.x64.checked.mch26,026,728+5,822,6690.00%
benchmarks.run.windows.x64.checked.mch12,229,984+3,013,2360.00%
benchmarks.run_pgo.windows.x64.checked.mch21,396,725+4,940,2350.00%
benchmarks.run_tiered.windows.x64.checked.mch3,899,205+1,038,7790.00%
coreclr_tests.run.windows.x64.checked.mch122,005,691+24,930,3490.00%
libraries.crossgen2.windows.x64.checked.mch45,248,517+11,738,6970.00%
libraries.pmi.windows.x64.checked.mch62,909,432+15,110,3670.00%
libraries_tests.run.windows.x64.Release.mch132,601,827+29,508,8830.00%
libraries_tests_no_tiered_compilation.run.windows.x64.Release.mch137,289,192+30,037,9790.00%
realworld.run.windows.x64.checked.mch11,139,943+2,474,5360.00%
smoke_tests.nativeaot.windows.x64.checked.mch5,022,597+1,394,2370.00%
Example diffs
aspnet.run.windows.x64.checked.mch
+2 (+0.97%) : 32079.dasm - Perfolizer.Mathematics.Distributions.StudentDistribution:InverseTwoTailStudent(double,double):double (Tier1)
@@ -29,7 +29,7 @@ G_M40993_IG01: ; bbWeight=0.25, gcrefRegs=0000 {}, byrefRegs=0000 {}, byr
vmovaps xmmword ptr [rsp+0x20], xmm12
vmovaps xmm6, xmm0
vmovaps xmm7, xmm1
- ;; size=60 bbWeight=0.25 PerfScore 3.69+ ;; size=61 bbWeight=0.25 PerfScore 3.69
G_M40993_IG02: ; bbWeight=0.25, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref
vxorps xmm8, xmm8, xmm8
vmovsd xmm9, qword ptr [reloc @RWD00]
@@ -72,13 +72,13 @@ G_M40993_IG08: ; bbWeight=1, epilog, nogc, extend
vmovaps xmm12, xmmword ptr [rsp+0x20]
add rsp, 152
ret - ;; size=53 bbWeight=1 PerfScore 29.25+ ;; size=54 bbWeight=1 PerfScore 29.25
RWD00 dq	408F400000000000h	; 1000
RWD08 dq	3FE0000000000000h	; 0.5
RWD16 dq	3E112E0BE826D695h	; 1e-09
-; Total bytes of code 207, prolog size 52, PerfScore 120.27, instruction count 38, allocated bytes for code 207 (MethodHash=c7795fde) for method Perfolizer.Mathematics.Distributions.StudentDistribution:InverseTwoTailStudent(double,double):double (Tier1)+; Total bytes of code 209, prolog size 53, PerfScore 120.27, instruction count 38, allocated bytes for code 209 (MethodHash=c7795fde) for method Perfolizer.Mathematics.Distributions.StudentDistribution:InverseTwoTailStudent(double,double):double (Tier1)
; ============================================================
Unwind Info:
@@ -86,24 +86,24 @@ Unwind Info:
>> End offset : 0xd1ffab1e (not in unwind data)
Version : 1
Flags : 0x00
- SizeOfProlog : 0x34+ SizeOfProlog : 0x35
CountOfUnwindCodes: 16
FrameRegister : none (0)
FrameOffset : N/A (no FrameRegister) (Value=0)
UnwindCodes :
- CodeOffset: 0x34 UnwindOp: UWOP_SAVE_XMM128 (8) OpInfo: XMM12 (12)+ CodeOffset: 0x35 UnwindOp: UWOP_SAVE_XMM128 (8) OpInfo: XMM12 (12)
Scaled Small Offset: 2 * 16 = 32 = 0x00020
- CodeOffset: 0x2E UnwindOp: UWOP_SAVE_XMM128 (8) OpInfo: XMM11 (11)+ CodeOffset: 0x2F UnwindOp: UWOP_SAVE_XMM128 (8) OpInfo: XMM11 (11)
Scaled Small Offset: 3 * 16 = 48 = 0x00030
- CodeOffset: 0x28 UnwindOp: UWOP_SAVE_XMM128 (8) OpInfo: XMM10 (10)+ CodeOffset: 0x29 UnwindOp: UWOP_SAVE_XMM128 (8) OpInfo: XMM10 (10)
Scaled Small Offset: 4 * 16 = 64 = 0x00040
- CodeOffset: 0x22 UnwindOp: UWOP_SAVE_XMM128 (8) OpInfo: XMM9 (9)+ CodeOffset: 0x23 UnwindOp: UWOP_SAVE_XMM128 (8) OpInfo: XMM9 (9)
Scaled Small Offset: 5 * 16 = 80 = 0x00050
- CodeOffset: 0x1C UnwindOp: UWOP_SAVE_XMM128 (8) OpInfo: XMM8 (8)+ CodeOffset: 0x1D UnwindOp: UWOP_SAVE_XMM128 (8) OpInfo: XMM8 (8)
Scaled Small Offset: 6 * 16 = 96 = 0x00060
- CodeOffset: 0x16 UnwindOp: UWOP_SAVE_XMM128 (8) OpInfo: XMM7 (7)+ CodeOffset: 0x17 UnwindOp: UWOP_SAVE_XMM128 (8) OpInfo: XMM7 (7)
Scaled Small Offset: 7 * 16 = 112 = 0x00070
- CodeOffset: 0x10 UnwindOp: UWOP_SAVE_XMM128 (8) OpInfo: XMM6 (6)+ CodeOffset: 0x11 UnwindOp: UWOP_SAVE_XMM128 (8) OpInfo: XMM6 (6)
Scaled Small Offset: 8 * 16 = 128 = 0x00080
- CodeOffset: 0x07 UnwindOp: UWOP_ALLOC_LARGE (1) OpInfo: 0 - Scaled small + CodeOffset: 0x08 UnwindOp: UWOP_ALLOC_LARGE (1) OpInfo: 0 - Scaled small 
Size: 19 * 8 = 152 = 0x00098
+2 (+0.97%) : 66029.dasm - Perfolizer.Mathematics.Distributions.StudentDistribution:InverseTwoTailStudent(double,double):double (Tier1)
@@ -30,7 +30,7 @@ G_M40993_IG01: ; bbWeight=0.98, gcrefRegs=0000 {}, byrefRegs=0000 {}, byr
vmovaps xmmword ptr [rsp+0x20], xmm12
vmovaps xmm6, xmm0
vmovaps xmm7, xmm1
- ;; size=60 bbWeight=0.98 PerfScore 14.39+ ;; size=61 bbWeight=0.98 PerfScore 14.39
G_M40993_IG02: ; bbWeight=0.98, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref
vxorps xmm8, xmm8, xmm8
vmovsd xmm9, qword ptr [reloc @RWD00]
@@ -69,7 +69,7 @@ G_M40993_IG07: ; bbWeight=1.00, epilog, nogc, extend
vmovaps xmm12, xmmword ptr [rsp+0x20]
add rsp, 152
ret - ;; size=53 bbWeight=1.00 PerfScore 29.25+ ;; size=54 bbWeight=1.00 PerfScore 29.25
G_M40993_IG08: ; bbWeight=15.73, gcVars=0000000000000000 {}, gcrefRegs=0000 {}, byrefRegs=0000 {}, gcvars, byref, isz
vmovaps xmm8, xmm12
jmp SHORT G_M40993_IG05
@@ -79,7 +79,7 @@ RWD08 dq	3FE0000000000000h	; 0.5
RWD16 dq	3E112E0BE826D695h	; 1e-09
-; Total bytes of code 207, prolog size 52, PerfScore 840.20, instruction count 38, allocated bytes for code 207 (MethodHash=c7795fde) for method Perfolizer.Mathematics.Distributions.StudentDistribution:InverseTwoTailStudent(double,double):double (Tier1)+; Total bytes of code 209, prolog size 53, PerfScore 840.20, instruction count 38, allocated bytes for code 209 (MethodHash=c7795fde) for method Perfolizer.Mathematics.Distributions.StudentDistribution:InverseTwoTailStudent(double,double):double (Tier1)
; ============================================================
Unwind Info:
@@ -87,24 +87,24 @@ Unwind Info:
>> End offset : 0xd1ffab1e (not in unwind data)
Version : 1
Flags : 0x00
- SizeOfProlog : 0x34+ SizeOfProlog : 0x35
CountOfUnwindCodes: 16
FrameRegister : none (0)
FrameOffset : N/A (no FrameRegister) (Value=0)
UnwindCodes :
- CodeOffset: 0x34 UnwindOp: UWOP_SAVE_XMM128 (8) OpInfo: XMM12 (12)+ CodeOffset: 0x35 UnwindOp: UWOP_SAVE_XMM128 (8) OpInfo: XMM12 (12)
Scaled Small Offset: 2 * 16 = 32 = 0x00020
- CodeOffset: 0x2E UnwindOp: UWOP_SAVE_XMM128 (8) OpInfo: XMM11 (11)+ CodeOffset: 0x2F UnwindOp: UWOP_SAVE_XMM128 (8) OpInfo: XMM11 (11)
Scaled Small Offset: 3 * 16 = 48 = 0x00030
- CodeOffset: 0x28 UnwindOp: UWOP_SAVE_XMM128 (8) OpInfo: XMM10 (10)+ CodeOffset: 0x29 UnwindOp: UWOP_SAVE_XMM128 (8) OpInfo: XMM10 (10)
Scaled Small Offset: 4 * 16 = 64 = 0x00040
- CodeOffset: 0x22 UnwindOp: UWOP_SAVE_XMM128 (8) OpInfo: XMM9 (9)+ CodeOffset: 0x23 UnwindOp: UWOP_SAVE_XMM128 (8) OpInfo: XMM9 (9)
Scaled Small Offset: 5 * 16 = 80 = 0x00050
- CodeOffset: 0x1C UnwindOp: UWOP_SAVE_XMM128 (8) OpInfo: XMM8 (8)+ CodeOffset: 0x1D UnwindOp: UWOP_SAVE_XMM128 (8) OpInfo: XMM8 (8)
Scaled Small Offset: 6 * 16 = 96 = 0x00060
- CodeOffset: 0x16 UnwindOp: UWOP_SAVE_XMM128 (8) OpInfo: XMM7 (7)+ CodeOffset: 0x17 UnwindOp: UWOP_SAVE_XMM128 (8) OpInfo: XMM7 (7)
Scaled Small Offset: 7 * 16 = 112 = 0x00070
- CodeOffset: 0x10 UnwindOp: UWOP_SAVE_XMM128 (8) OpInfo: XMM6 (6)+ CodeOffset: 0x11 UnwindOp: UWOP_SAVE_XMM128 (8) OpInfo: XMM6 (6)
Scaled Small Offset: 8 * 16 = 128 = 0x00080
- CodeOffset: 0x07 UnwindOp: UWOP_ALLOC_LARGE (1) OpInfo: 0 - Scaled small + CodeOffset: 0x08 UnwindOp: UWOP_ALLOC_LARGE (1) OpInfo: 0 - Scaled small 
Size: 19 * 8 = 152 = 0x00098
+2 (+0.97%) : 60922.dasm - Perfolizer.Mathematics.Distributions.StudentDistribution:InverseTwoTailStudent(double,double):double (FullOpts)
@@ -29,7 +29,7 @@ G_M40993_IG01: ; bbWeight=0.25, gcrefRegs=0000 {}, byrefRegs=0000 {}, byr
vmovaps xmmword ptr [rsp+0x20], xmm12
vmovaps xmm6, xmm0
vmovaps xmm7, xmm1
- ;; size=60 bbWeight=0.25 PerfScore 3.69+ ;; size=61 bbWeight=0.25 PerfScore 3.69
G_M40993_IG02: ; bbWeight=0.25, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref
vxorps xmm8, xmm8, xmm8
vmovsd xmm9, qword ptr [reloc @RWD00]
@@ -72,13 +72,13 @@ G_M40993_IG08: ; bbWeight=1, epilog, nogc, extend
vmovaps xmm12, xmmword ptr [rsp+0x20]
add rsp, 152
ret - ;; size=53 bbWeight=1 PerfScore 29.25+ ;; size=54 bbWeight=1 PerfScore 29.25
RWD00 dq	408F400000000000h	; 1000
RWD08 dq	3FE0000000000000h	; 0.5
RWD16 dq	3E112E0BE826D695h	; 1e-09
-; Total bytes of code 207, prolog size 52, PerfScore 120.27, instruction count 38, allocated bytes for code 207 (MethodHash=c7795fde) for method Perfolizer.Mathematics.Distributions.StudentDistribution:InverseTwoTailStudent(double,double):double (FullOpts)+; Total bytes of code 209, prolog size 53, PerfScore 120.27, instruction count 38, allocated bytes for code 209 (MethodHash=c7795fde) for method Perfolizer.Mathematics.Distributions.StudentDistribution:InverseTwoTailStudent(double,double):double (FullOpts)
; ============================================================
Unwind Info:
@@ -86,24 +86,24 @@ Unwind Info:
>> End offset : 0xd1ffab1e (not in unwind data)
Version : 1
Flags : 0x00
- SizeOfProlog : 0x34+ SizeOfProlog : 0x35
CountOfUnwindCodes: 16
FrameRegister : none (0)
FrameOffset : N/A (no FrameRegister) (Value=0)
UnwindCodes :
- CodeOffset: 0x34 UnwindOp: UWOP_SAVE_XMM128 (8) OpInfo: XMM12 (12)+ CodeOffset: 0x35 UnwindOp: UWOP_SAVE_XMM128 (8) OpInfo: XMM12 (12)
Scaled Small Offset: 2 * 16 = 32 = 0x00020
- CodeOffset: 0x2E UnwindOp: UWOP_SAVE_XMM128 (8) OpInfo: XMM11 (11)+ CodeOffset: 0x2F UnwindOp: UWOP_SAVE_XMM128 (8) OpInfo: XMM11 (11)
Scaled Small Offset: 3 * 16 = 48 = 0x00030
- CodeOffset: 0x28 UnwindOp: UWOP_SAVE_XMM128 (8) OpInfo: XMM10 (10)+ CodeOffset: 0x29 UnwindOp: UWOP_SAVE_XMM128 (8) OpInfo: XMM10 (10)
Scaled Small Offset: 4 * 16 = 64 = 0x00040
- CodeOffset: 0x22 UnwindOp: UWOP_SAVE_XMM128 (8) OpInfo: XMM9 (9)+ CodeOffset: 0x23 UnwindOp: UWOP_SAVE_XMM128 (8) OpInfo: XMM9 (9)
Scaled Small Offset: 5 * 16 = 80 = 0x00050
- CodeOffset: 0x1C UnwindOp: UWOP_SAVE_XMM128 (8) OpInfo: XMM8 (8)+ CodeOffset: 0x1D UnwindOp: UWOP_SAVE_XMM128 (8) OpInfo: XMM8 (8)
Scaled Small Offset: 6 * 16 = 96 = 0x00060
- CodeOffset: 0x16 UnwindOp: UWOP_SAVE_XMM128 (8) OpInfo: XMM7 (7)+ CodeOffset: 0x17 UnwindOp: UWOP_SAVE_XMM128 (8) OpInfo: XMM7 (7)
Scaled Small Offset: 7 * 16 = 112 = 0x00070
- CodeOffset: 0x10 UnwindOp: UWOP_SAVE_XMM128 (8) OpInfo: XMM6 (6)+ CodeOffset: 0x11 UnwindOp: UWOP_SAVE_XMM128 (8) OpInfo: XMM6 (6)
Scaled Small Offset: 8 * 16 = 128 = 0x00080
- CodeOffset: 0x07 UnwindOp: UWOP_ALLOC_LARGE (1) OpInfo: 0 - Scaled small + CodeOffset: 0x08 UnwindOp: UWOP_ALLOC_LARGE (1) OpInfo: 0 - Scaled small 
Size: 19 * 8 = 152 = 0x00098
+7 (+87.50%) : 23755.dasm - System.Runtime.Intrinsics.X86.X86Serialize:get_IsSupported():ubyte (Tier0)
@@ -12,16 +12,16 @@
G_M10906_IG01: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref, nogc <-- Prolog IG
push rbp
mov rbp, rsp
- ;; size=4 bbWeight=1 PerfScore 1.25+ ;; size=7 bbWeight=1 PerfScore 1.25
G_M10906_IG02: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref
xor eax, eax
- ;; size=2 bbWeight=1 PerfScore 0.25+ ;; size=4 bbWeight=1 PerfScore 0.25
G_M10906_IG03: ; bbWeight=1, epilog, nogc, extend
pop rbp
ret - ;; size=2 bbWeight=1 PerfScore 1.50+ ;; size=4 bbWeight=1 PerfScore 1.50-; Total bytes of code 8, prolog size 4, PerfScore 3.00, instruction count 5, allocated bytes for code 8 (MethodHash=da05d565) for method System.Runtime.Intrinsics.X86.X86Serialize:get_IsSupported():ubyte (Tier0)+; Total bytes of code 15, prolog size 7, PerfScore 3.00, instruction count 5, allocated bytes for code 15 (MethodHash=da05d565) for method System.Runtime.Intrinsics.X86.X86Serialize:get_IsSupported():ubyte (Tier0)
; ============================================================
Unwind Info:
@@ -29,9 +29,9 @@ Unwind Info:
>> End offset : 0xd1ffab1e (not in unwind data)
Version : 1
Flags : 0x00
- SizeOfProlog : 0x01+ SizeOfProlog : 0x03
CountOfUnwindCodes: 1
FrameRegister : none (0)
FrameOffset : N/A (no FrameRegister) (Value=0)
UnwindCodes :
- CodeOffset: 0x01 UnwindOp: UWOP_PUSH_NONVOL (0) OpInfo: rbp (5)+ CodeOffset: 0x03 UnwindOp: UWOP_PUSH_NONVOL (0) OpInfo: rbp (5)
+7 (+87.50%) : 9608.dasm - System.Runtime.CompilerServices.RuntimeHelpers:IsReferenceOrContainsReferences[System.Text.Json.JsonDocument+DbRow]():ubyte (Tier0)
@@ -12,16 +12,16 @@
G_M31768_IG01: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref, nogc <-- Prolog IG
push rbp
mov rbp, rsp
- ;; size=4 bbWeight=1 PerfScore 1.25+ ;; size=7 bbWeight=1 PerfScore 1.25
G_M31768_IG02: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref
xor eax, eax
- ;; size=2 bbWeight=1 PerfScore 0.25+ ;; size=4 bbWeight=1 PerfScore 0.25
G_M31768_IG03: ; bbWeight=1, epilog, nogc, extend
pop rbp
ret - ;; size=2 bbWeight=1 PerfScore 1.50+ ;; size=4 bbWeight=1 PerfScore 1.50-; Total bytes of code 8, prolog size 4, PerfScore 3.00, instruction count 5, allocated bytes for code 8 (MethodHash=2e9583e7) for method System.Runtime.CompilerServices.RuntimeHelpers:IsReferenceOrContainsReferences[System.Text.Json.JsonDocument+DbRow]():ubyte (Tier0)+; Total bytes of code 15, prolog size 7, PerfScore 3.00, instruction count 5, allocated bytes for code 15 (MethodHash=2e9583e7) for method System.Runtime.CompilerServices.RuntimeHelpers:IsReferenceOrContainsReferences[System.Text.Json.JsonDocument+DbRow]():ubyte (Tier0)
; ============================================================
Unwind Info:
@@ -29,9 +29,9 @@ Unwind Info:
>> End offset : 0xd1ffab1e (not in unwind data)
Version : 1
Flags : 0x00
- SizeOfProlog : 0x01+ SizeOfProlog : 0x03
CountOfUnwindCodes: 1
FrameRegister : none (0)
FrameOffset : N/A (no FrameRegister) (Value=0)
UnwindCodes :
- CodeOffset: 0x01 UnwindOp: UWOP_PUSH_NONVOL (0) OpInfo: rbp (5)+ CodeOffset: 0x03 UnwindOp: UWOP_PUSH_NONVOL (0) OpInfo: rbp (5)
+7 (+87.50%) : 861.dasm - System.Buffers.IndexOfAnyAsciiSearcher+ContainsAnyResultMapper`1[short]:get_NotFound():ubyte (Tier0)
@@ -12,16 +12,16 @@
G_M16088_IG01: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref, nogc <-- Prolog IG
push rbp
mov rbp, rsp
- ;; size=4 bbWeight=1 PerfScore 1.25+ ;; size=7 bbWeight=1 PerfScore 1.25
G_M16088_IG02: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref
xor eax, eax
- ;; size=2 bbWeight=1 PerfScore 0.25+ ;; size=4 bbWeight=1 PerfScore 0.25
G_M16088_IG03: ; bbWeight=1, epilog, nogc, extend
pop rbp
ret - ;; size=2 bbWeight=1 PerfScore 1.50+ ;; size=4 bbWeight=1 PerfScore 1.50-; Total bytes of code 8, prolog size 4, PerfScore 3.00, instruction count 5, allocated bytes for code 8 (MethodHash=da12c127) for method System.Buffers.IndexOfAnyAsciiSearcher+ContainsAnyResultMapper`1[short]:get_NotFound():ubyte (Tier0)+; Total bytes of code 15, prolog size 7, PerfScore 3.00, instruction count 5, allocated bytes for code 15 (MethodHash=da12c127) for method System.Buffers.IndexOfAnyAsciiSearcher+ContainsAnyResultMapper`1[short]:get_NotFound():ubyte (Tier0)
; ============================================================
Unwind Info:
@@ -29,9 +29,9 @@ Unwind Info:
>> End offset : 0xd1ffab1e (not in unwind data)
Version : 1
Flags : 0x00
- SizeOfProlog : 0x01+ SizeOfProlog : 0x03
CountOfUnwindCodes: 1
FrameRegister : none (0)
FrameOffset : N/A (no FrameRegister) (Value=0)
UnwindCodes :
- CodeOffset: 0x01 UnwindOp: UWOP_PUSH_NONVOL (0) OpInfo: rbp (5)+ CodeOffset: 0x03 UnwindOp: UWOP_PUSH_NONVOL (0) OpInfo: rbp (5)
benchmarks.run.windows.x64.checked.mch
+1 (+0.14%) : 30038.dasm - System.Numerics.Matrix4x4+Impl:Transform(byref,byref):System.Numerics.Matrix4x4+Impl (FullOpts)
@@ -199,14 +199,14 @@ G_M8955_IG03: ; bbWeight=1, extend
vmovups xmmword ptr [rcx+0x30], xmm0
mov rax, rcx
; byrRegs +[rax]
- ;; size=354 bbWeight=1 PerfScore 161.25+ ;; size=355 bbWeight=1 PerfScore 161.25
G_M8955_IG04: ; bbWeight=1, epilog, nogc, extend
ret ;; size=1 bbWeight=1 PerfScore 1.00
RWD00 dd	3F800000h ; 1
-; Total bytes of code 693, prolog size 0, PerfScore 345.25, instruction count 118, allocated bytes for code 693 (MethodHash=5906dd04) for method System.Numerics.Matrix4x4+Impl:Transform(byref,byref):System.Numerics.Matrix4x4+Impl (FullOpts)+; Total bytes of code 694, prolog size 0, PerfScore 345.25, instruction count 118, allocated bytes for code 694 (MethodHash=5906dd04) for method System.Numerics.Matrix4x4+Impl:Transform(byref,byref):System.Numerics.Matrix4x4+Impl (FullOpts)
; ============================================================
Unwind Info:
+1 (+0.20%) : 21351.dasm - System.Numerics.Quaternion:CreateFromRotationMatrix(System.Numerics.Matrix4x4):System.Numerics.Quaternion (FullOpts)
@@ -140,7 +140,7 @@ G_M15800_IG07: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0002 {rcx}, byr
vmovups xmmword ptr [rcx], xmm4
mov rax, rcx
; byrRegs +[rax]
- ;; size=7 bbWeight=1 PerfScore 2.25+ ;; size=8 bbWeight=1 PerfScore 2.25
G_M15800_IG08: ; bbWeight=1, epilog, nogc, extend
ret ;; size=1 bbWeight=1 PerfScore 1.00
@@ -148,7 +148,7 @@ RWD00 dd	3F800000h ; 1
RWD04 dd	3F000000h ; 0.5
-; Total bytes of code 494, prolog size 0, PerfScore 173.42, instruction count 93, allocated bytes for code 494 (MethodHash=096ec247) for method System.Numerics.Quaternion:CreateFromRotationMatrix(System.Numerics.Matrix4x4):System.Numerics.Quaternion (FullOpts)+; Total bytes of code 495, prolog size 0, PerfScore 173.42, instruction count 93, allocated bytes for code 495 (MethodHash=096ec247) for method System.Numerics.Quaternion:CreateFromRotationMatrix(System.Numerics.Matrix4x4):System.Numerics.Quaternion (FullOpts)
; ============================================================
Unwind Info:
+1 (+0.23%) : 33197.dasm - System.Runtime.Intrinsics.VectorMath:ExpSingle[System.Runtime.Intrinsics.Vector512`1[float],System.Runtime.Intrinsics.Vector512`1[uint],System.Runtime.Intrinsics.Vector512`1[double],System.Runtime.Intrinsics.Vector512`1[ulong]](System.Runtime.Intrinsics.Vector512`1[float]):System.Runtime.Intrinsics.Vector512`1[float] (FullOpts)
@@ -140,7 +140,7 @@ G_M41960_IG04: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0002 {rcx}, byr
vmovups zmmword ptr [rcx], zmm1
mov rax, rcx
; byrRegs +[rax]
- ;; size=9 bbWeight=1 PerfScore 2.25+ ;; size=10 bbWeight=1 PerfScore 2.25
G_M41960_IG05: ; bbWeight=1, epilog, nogc, extend
vzeroupper ret @@ -163,7 +163,7 @@ RWD588 dd	00000000h, 00000000h, 00000000h, 00000000h, 00000000h, 00000000h
RWD640 dq	7F8000007F800000h, 7F8000007F800000h, 7F8000007F800000h, 7F8000007F800000h, 7F8000007F800000h, 7F8000007F800000h, 7F8000007F800000h, 7F8000007F800000h
-; Total bytes of code 438, prolog size 0, PerfScore 166.17, instruction count 66, allocated bytes for code 441 (MethodHash=326b5c17) for method System.Runtime.Intrinsics.VectorMath:ExpSingle[System.Runtime.Intrinsics.Vector512`1[float],System.Runtime.Intrinsics.Vector512`1[uint],System.Runtime.Intrinsics.Vector512`1[double],System.Runtime.Intrinsics.Vector512`1[ulong]](System.Runtime.Intrinsics.Vector512`1[float]):System.Runtime.Intrinsics.Vector512`1[float] (FullOpts)+; Total bytes of code 439, prolog size 0, PerfScore 166.17, instruction count 66, allocated bytes for code 442 (MethodHash=326b5c17) for method System.Runtime.Intrinsics.VectorMath:ExpSingle[System.Runtime.Intrinsics.Vector512`1[float],System.Runtime.Intrinsics.Vector512`1[uint],System.Runtime.Intrinsics.Vector512`1[double],System.Runtime.Intrinsics.Vector512`1[ulong]](System.Runtime.Intrinsics.Vector512`1[float]):System.Runtime.Intrinsics.Vector512`1[float] (FullOpts)
; ============================================================
Unwind Info:
+4 (+80.00%) : 32809.dasm - System.Linq.Tests.Perf_Enumerable+<>c:b__25_1(int):int:this (FullOpts)
@@ -18,12 +18,12 @@ G_M1177_IG01: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref,
G_M1177_IG02: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref
mov eax, edx
neg eax
- ;; size=4 bbWeight=1 PerfScore 0.50+ ;; size=8 bbWeight=1 PerfScore 0.50
G_M1177_IG03: ; bbWeight=1, epilog, nogc, extend
ret ;; size=1 bbWeight=1 PerfScore 1.00
-; Total bytes of code 5, prolog size 0, PerfScore 1.50, instruction count 3, allocated bytes for code 5 (MethodHash=bea0fb66) for method System.Linq.Tests.Perf_Enumerable+<>c:<OrderByThenBy>b__25_1(int):int:this (FullOpts)+; Total bytes of code 9, prolog size 0, PerfScore 1.50, instruction count 3, allocated bytes for code 9 (MethodHash=bea0fb66) for method System.Linq.Tests.Perf_Enumerable+<>c:<OrderByThenBy>b__25_1(int):int:this (FullOpts)
; ============================================================
Unwind Info:
+4 (+80.00%) : 38257.dasm - Microsoft.Extensions.Primitives.StringSegmentBenchmark:Equals_Object_Invalid():ubyte:this (FullOpts)
@@ -98,12 +98,12 @@ G_M24192_IG02: ; bbWeight=1, gcrefRegs=0002 {rcx}, byrefRegs=0000 {}, byr
; gcrRegs +[rcx]
cmp byte ptr [rcx], cl
xor eax, eax
- ;; size=4 bbWeight=1 PerfScore 3.25+ ;; size=8 bbWeight=1 PerfScore 3.25
G_M24192_IG03: ; bbWeight=1, epilog, nogc, extend
ret ;; size=1 bbWeight=1 PerfScore 1.00
-; Total bytes of code 5, prolog size 0, PerfScore 4.25, instruction count 3, allocated bytes for code 5 (MethodHash=7f17a17f) for method Microsoft.Extensions.Primitives.StringSegmentBenchmark:Equals_Object_Invalid():ubyte:this (FullOpts)+; Total bytes of code 9, prolog size 0, PerfScore 4.25, instruction count 3, allocated bytes for code 9 (MethodHash=7f17a17f) for method Microsoft.Extensions.Primitives.StringSegmentBenchmark:Equals_Object_Invalid():ubyte:this (FullOpts)
; ============================================================
Unwind Info:
+4 (+80.00%) : 22418.dasm - System.Xml.XmlDictionaryReader:TryGetArrayLength(byref):ubyte:this (FullOpts)
@@ -19,13 +19,13 @@ G_M47878_IG02: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0004 {rdx}, byr
; byrRegs +[rdx]
xor eax, eax
mov dword ptr [rdx], eax
- ;; size=4 bbWeight=1 PerfScore 1.25+ ;; size=8 bbWeight=1 PerfScore 1.25
G_M47878_IG03: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref, epilog, nogc
; byrRegs -[rdx]
ret ;; size=1 bbWeight=1 PerfScore 1.00
-; Total bytes of code 5, prolog size 0, PerfScore 2.25, instruction count 3, allocated bytes for code 5 (MethodHash=068744f9) for method System.Xml.XmlDictionaryReader:TryGetArrayLength(byref):ubyte:this (FullOpts)+; Total bytes of code 9, prolog size 0, PerfScore 2.25, instruction count 3, allocated bytes for code 9 (MethodHash=068744f9) for method System.Xml.XmlDictionaryReader:TryGetArrayLength(byref):ubyte:this (FullOpts)
; ============================================================
Unwind Info:
benchmarks.run_pgo.windows.x64.checked.mch
+19 (+0.74%) : 95347.dasm - System.Runtime.Intrinsics.X86.Sse2:ShiftRightLogical128BitLane(System.Runtime.Intrinsics.Vector128`1[ulong],ubyte):System.Runtime.Intrinsics.Vector128`1[ulong] (Tier0)
@@ -18,7 +18,7 @@ G_M9309_IG01: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref,
mov bword ptr [rbp+0x10], rcx
mov bword ptr [rbp+0x18], rdx
mov dword ptr [rbp+0x20], r8d
- ;; size=16 bbWeight=1 PerfScore 4.25+ ;; size=22 bbWeight=1 PerfScore 4.25
G_M9309_IG02: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref
mov rax, bword ptr [rbp+0x18]
; byrRegs +[rax]
@@ -31,7 +31,7 @@ G_M9309_IG02: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref
lea rcx, G_M9309_IG02
add rdx, rcx
jmp rdx
- ;; size=36 bbWeight=1 PerfScore 12.00+ ;; size=45 bbWeight=1 PerfScore 12.00
G_M9309_IG03: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref
vpsrldq xmm0, xmm0, 0
jmp G_M9309_IG259
@@ -1061,11 +1061,11 @@ G_M9309_IG259: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref
; byrRegs +[rax]
vmovups xmmword ptr [rax], xmm0
mov rax, bword ptr [rbp+0x10]
- ;; size=12 bbWeight=1 PerfScore 4.00+ ;; size=14 bbWeight=1 PerfScore 4.00
G_M9309_IG260: ; bbWeight=1, epilog, nogc, extend
pop rbp
ret - ;; size=2 bbWeight=1 PerfScore 1.50+ ;; size=4 bbWeight=1 PerfScore 1.50
RWD00 dd	G_M9309_IG03 - G_M9309_IG02
dd	G_M9309_IG04 - G_M9309_IG02
dd	G_M9309_IG05 - G_M9309_IG02
@@ -1324,7 +1324,7 @@ RWD00 dd	G_M9309_IG03 - G_M9309_IG02
dd	G_M9309_IG258 - G_M9309_IG02
-; Total bytes of code 2572, prolog size 4, PerfScore 789.75, instruction count 531, allocated bytes for code 2572 (MethodHash=c6eadba2) for method System.Runtime.Intrinsics.X86.Sse2:ShiftRightLogical128BitLane(System.Runtime.Intrinsics.Vector128`1[ulong],ubyte):System.Runtime.Intrinsics.Vector128`1[ulong] (Tier0)+; Total bytes of code 2591, prolog size 7, PerfScore 789.75, instruction count 531, allocated bytes for code 2591 (MethodHash=c6eadba2) for method System.Runtime.Intrinsics.X86.Sse2:ShiftRightLogical128BitLane(System.Runtime.Intrinsics.Vector128`1[ulong],ubyte):System.Runtime.Intrinsics.Vector128`1[ulong] (Tier0)
; ============================================================
Unwind Info:
@@ -1332,9 +1332,9 @@ Unwind Info:
>> End offset : 0xd1ffab1e (not in unwind data)
Version : 1
Flags : 0x00
- SizeOfProlog : 0x01+ SizeOfProlog : 0x03
CountOfUnwindCodes: 1
FrameRegister : none (0)
FrameOffset : N/A (no FrameRegister) (Value=0)
UnwindCodes :
- CodeOffset: 0x01 UnwindOp: UWOP_PUSH_NONVOL (0) OpInfo: rbp (5)+ CodeOffset: 0x03 UnwindOp: UWOP_PUSH_NONVOL (0) OpInfo: rbp (5)
+1 (+0.85%) : 74356.dasm - Benchmarks.SIMD.RayTracer.Sphere:Normal(Benchmarks.SIMD.RayTracer.Vector):Benchmarks.SIMD.RayTracer.Vector:this (Tier1)
@@ -75,7 +75,7 @@ G_M56301_IG04: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0004 {rdx}, byr
vextractps dword ptr [rdx+0x08], xmm0, 2
mov rax, rdx
; byrRegs +[rax]
- ;; size=31 bbWeight=1 PerfScore 18.25+ ;; size=32 bbWeight=1 PerfScore 18.25
G_M56301_IG05: ; bbWeight=1, epilog, nogc, extend
ret ;; size=1 bbWeight=1 PerfScore 1.00
@@ -88,7 +88,7 @@ RWD00 dd	3F800000h ; 1
RWD04 dd	7F800000h ; inf
-; Total bytes of code 118, prolog size 0, PerfScore 86.58, instruction count 26, allocated bytes for code 118 (MethodHash=3e632412) for method Benchmarks.SIMD.RayTracer.Sphere:Normal(Benchmarks.SIMD.RayTracer.Vector):Benchmarks.SIMD.RayTracer.Vector:this (Tier1)+; Total bytes of code 119, prolog size 0, PerfScore 86.58, instruction count 26, allocated bytes for code 119 (MethodHash=3e632412) for method Benchmarks.SIMD.RayTracer.Sphere:Normal(Benchmarks.SIMD.RayTracer.Vector):Benchmarks.SIMD.RayTracer.Vector:this (Tier1)
; ============================================================
Unwind Info:
+1 (+0.99%) : 74345.dasm - Benchmarks.SIMD.RayTracer.Vector:Norm(Benchmarks.SIMD.RayTracer.Vector):Benchmarks.SIMD.RayTracer.Vector (Tier1)
@@ -60,7 +60,7 @@ G_M16924_IG04: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0002 {rcx}, byr
vextractps dword ptr [rcx+0x08], xmm0, 2
mov rax, rcx
; byrRegs +[rax]
- ;; size=31 bbWeight=1 PerfScore 18.25+ ;; size=32 bbWeight=1 PerfScore 18.25
G_M16924_IG05: ; bbWeight=1, epilog, nogc, extend
ret ;; size=1 bbWeight=1 PerfScore 1.00
@@ -73,7 +73,7 @@ RWD00 dd	3F800000h ; 1
RWD04 dd	7F800000h ; inf
-; Total bytes of code 101, prolog size 0, PerfScore 76.58, instruction count 23, allocated bytes for code 101 (MethodHash=5eedbde3) for method Benchmarks.SIMD.RayTracer.Vector:Norm(Benchmarks.SIMD.RayTracer.Vector):Benchmarks.SIMD.RayTracer.Vector (Tier1)+; Total bytes of code 102, prolog size 0, PerfScore 76.58, instruction count 23, allocated bytes for code 102 (MethodHash=5eedbde3) for method Benchmarks.SIMD.RayTracer.Vector:Norm(Benchmarks.SIMD.RayTracer.Vector):Benchmarks.SIMD.RayTracer.Vector (Tier1)
; ============================================================
Unwind Info:
+7 (+87.50%) : 1151.dasm - System.OperatingSystem:IsBrowser():ubyte (Tier0)
@@ -12,16 +12,16 @@
G_M61665_IG01: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref, nogc <-- Prolog IG
push rbp
mov rbp, rsp
- ;; size=4 bbWeight=1 PerfScore 1.25+ ;; size=7 bbWeight=1 PerfScore 1.25
G_M61665_IG02: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref
xor eax, eax
- ;; size=2 bbWeight=1 PerfScore 0.25+ ;; size=4 bbWeight=1 PerfScore 0.25
G_M61665_IG03: ; bbWeight=1, epilog, nogc, extend
pop rbp
ret - ;; size=2 bbWeight=1 PerfScore 1.50+ ;; size=4 bbWeight=1 PerfScore 1.50-; Total bytes of code 8, prolog size 4, PerfScore 3.00, instruction count 5, allocated bytes for code 8 (MethodHash=f0b70f1e) for method System.OperatingSystem:IsBrowser():ubyte (Tier0)+; Total bytes of code 15, prolog size 7, PerfScore 3.00, instruction count 5, allocated bytes for code 15 (MethodHash=f0b70f1e) for method System.OperatingSystem:IsBrowser():ubyte (Tier0)
; ============================================================
Unwind Info:
@@ -29,9 +29,9 @@ Unwind Info:
>> End offset : 0xd1ffab1e (not in unwind data)
Version : 1
Flags : 0x00
- SizeOfProlog : 0x01+ SizeOfProlog : 0x03
CountOfUnwindCodes: 1
FrameRegister : none (0)
FrameOffset : N/A (no FrameRegister) (Value=0)
UnwindCodes :
- CodeOffset: 0x01 UnwindOp: UWOP_PUSH_NONVOL (0) OpInfo: rbp (5)+ CodeOffset: 0x03 UnwindOp: UWOP_PUSH_NONVOL (0) OpInfo: rbp (5)
+7 (+87.50%) : 76175.dasm - Microsoft.CodeAnalysis.Collections.Internal.RoslynUnsafe:NullRef[int]():byref (Tier0)
@@ -12,16 +12,16 @@
G_M47256_IG01: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref, nogc <-- Prolog IG
push rbp
mov rbp, rsp
- ;; size=4 bbWeight=1 PerfScore 1.25+ ;; size=7 bbWeight=1 PerfScore 1.25
G_M47256_IG02: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref
xor eax, eax
- ;; size=2 bbWeight=1 PerfScore 0.25+ ;; size=4 bbWeight=1 PerfScore 0.25
G_M47256_IG03: ; bbWeight=1, epilog, nogc, extend
pop rbp
ret - ;; size=2 bbWeight=1 PerfScore 1.50+ ;; size=4 bbWeight=1 PerfScore 1.50-; Total bytes of code 8, prolog size 4, PerfScore 3.00, instruction count 5, allocated bytes for code 8 (MethodHash=8e824767) for method Microsoft.CodeAnalysis.Collections.Internal.RoslynUnsafe:NullRef[int]():byref (Tier0)+; Total bytes of code 15, prolog size 7, PerfScore 3.00, instruction count 5, allocated bytes for code 15 (MethodHash=8e824767) for method Microsoft.CodeAnalysis.Collections.Internal.RoslynUnsafe:NullRef[int]():byref (Tier0)
; ============================================================
Unwind Info:
@@ -29,9 +29,9 @@ Unwind Info:
>> End offset : 0xd1ffab1e (not in unwind data)
Version : 1
Flags : 0x00
- SizeOfProlog : 0x01+ SizeOfProlog : 0x03
CountOfUnwindCodes: 1
FrameRegister : none (0)
FrameOffset : N/A (no FrameRegister) (Value=0)
UnwindCodes :
- CodeOffset: 0x01 UnwindOp: UWOP_PUSH_NONVOL (0) OpInfo: rbp (5)+ CodeOffset: 0x03 UnwindOp: UWOP_PUSH_NONVOL (0) OpInfo: rbp (5)
+7 (+87.50%) : 27756.dasm - System.SByte:System.Numerics.INumberBase.get_Zero():byte (Tier0)
@@ -12,16 +12,16 @@
G_M14356_IG01: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref, nogc <-- Prolog IG
push rbp
mov rbp, rsp
- ;; size=4 bbWeight=1 PerfScore 1.25+ ;; size=7 bbWeight=1 PerfScore 1.25
G_M14356_IG02: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref
xor eax, eax
- ;; size=2 bbWeight=1 PerfScore 0.25+ ;; size=4 bbWeight=1 PerfScore 0.25
G_M14356_IG03: ; bbWeight=1, epilog, nogc, extend
pop rbp
ret - ;; size=2 bbWeight=1 PerfScore 1.50+ ;; size=4 bbWeight=1 PerfScore 1.50-; Total bytes of code 8, prolog size 4, PerfScore 3.00, instruction count 5, allocated bytes for code 8 (MethodHash=1d31c7eb) for method System.SByte:System.Numerics.INumberBase<System.SByte>.get_Zero():byte (Tier0)+; Total bytes of code 15, prolog size 7, PerfScore 3.00, instruction count 5, allocated bytes for code 15 (MethodHash=1d31c7eb) for method System.SByte:System.Numerics.INumberBase<System.SByte>.get_Zero():byte (Tier0)
; ============================================================
Unwind Info:
@@ -29,9 +29,9 @@ Unwind Info:
>> End offset : 0xd1ffab1e (not in unwind data)
Version : 1
Flags : 0x00
- SizeOfProlog : 0x01+ SizeOfProlog : 0x03
CountOfUnwindCodes: 1
FrameRegister : none (0)
FrameOffset : N/A (no FrameRegister) (Value=0)
UnwindCodes :
- CodeOffset: 0x01 UnwindOp: UWOP_PUSH_NONVOL (0) OpInfo: rbp (5)+ CodeOffset: 0x03 UnwindOp: UWOP_PUSH_NONVOL (0) OpInfo: rbp (5)
benchmarks.run_tiered.windows.x64.checked.mch
+1 (+0.26%) : 61218.dasm - System.Runtime.Intrinsics.VectorMath:ExpSingle[System.Runtime.Intrinsics.Vector256`1[float],System.Runtime.Intrinsics.Vector256`1[uint],System.Runtime.Intrinsics.Vector256`1[double],System.Runtime.Intrinsics.Vector256`1[ulong]](System.Runtime.Intrinsics.Vector256`1[float]):System.Runtime.Intrinsics.Vector256`1[float] (Tier1)
@@ -139,7 +139,7 @@ G_M61896_IG04: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0002 {rcx}, byr
vmovups ymmword ptr [rcx], ymm1
mov rax, rcx
; byrRegs +[rax]
- ;; size=7 bbWeight=1 PerfScore 2.25+ ;; size=8 bbWeight=1 PerfScore 2.25
G_M61896_IG05: ; bbWeight=1, epilog, nogc, extend
vzeroupper ret @@ -161,7 +161,7 @@ RWD352 dq	42B1721842B17218h, 42B1721842B17218h, 42B1721842B17218h, 42B1721842B1
RWD384 dq	7F8000007F800000h, 7F8000007F800000h, 7F8000007F800000h, 7F8000007F800000h
-; Total bytes of code 378, prolog size 0, PerfScore 188.17, instruction count 65, allocated bytes for code 380 (MethodHash=a8e80e37) for method System.Runtime.Intrinsics.VectorMath:ExpSingle[System.Runtime.Intrinsics.Vector256`1[float],System.Runtime.Intrinsics.Vector256`1[uint],System.Runtime.Intrinsics.Vector256`1[double],System.Runtime.Intrinsics.Vector256`1[ulong]](System.Runtime.Intrinsics.Vector256`1[float]):System.Runtime.Intrinsics.Vector256`1[float] (Tier1)+; Total bytes of code 379, prolog size 0, PerfScore 188.17, instruction count 65, allocated bytes for code 381 (MethodHash=a8e80e37) for method System.Runtime.Intrinsics.VectorMath:ExpSingle[System.Runtime.Intrinsics.Vector256`1[float],System.Runtime.Intrinsics.Vector256`1[uint],System.Runtime.Intrinsics.Vector256`1[double],System.Runtime.Intrinsics.Vector256`1[ulong]](System.Runtime.Intrinsics.Vector256`1[float]):System.Runtime.Intrinsics.Vector256`1[float] (Tier1)
; ============================================================
Unwind Info:
+19 (+0.74%) : 59614.dasm - System.Runtime.Intrinsics.X86.Sse2:ShiftRightLogical128BitLane(System.Runtime.Intrinsics.Vector128`1[ulong],ubyte):System.Runtime.Intrinsics.Vector128`1[ulong] (Tier0)
@@ -18,7 +18,7 @@ G_M9309_IG01: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref,
mov bword ptr [rbp+0x10], rcx
mov bword ptr [rbp+0x18], rdx
mov dword ptr [rbp+0x20], r8d
- ;; size=16 bbWeight=1 PerfScore 4.25+ ;; size=22 bbWeight=1 PerfScore 4.25
G_M9309_IG02: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref
mov rax, bword ptr [rbp+0x18]
; byrRegs +[rax]
@@ -31,7 +31,7 @@ G_M9309_IG02: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref
lea rcx, G_M9309_IG02
add rdx, rcx
jmp rdx
- ;; size=36 bbWeight=1 PerfScore 12.00+ ;; size=45 bbWeight=1 PerfScore 12.00
G_M9309_IG03: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref
vpsrldq xmm0, xmm0, 0
jmp G_M9309_IG259
@@ -1061,11 +1061,11 @@ G_M9309_IG259: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref
; byrRegs +[rax]
vmovups xmmword ptr [rax], xmm0
mov rax, bword ptr [rbp+0x10]
- ;; size=12 bbWeight=1 PerfScore 4.00+ ;; size=14 bbWeight=1 PerfScore 4.00
G_M9309_IG260: ; bbWeight=1, epilog, nogc, extend
pop rbp
ret - ;; size=2 bbWeight=1 PerfScore 1.50+ ;; size=4 bbWeight=1 PerfScore 1.50
RWD00 dd	G_M9309_IG03 - G_M9309_IG02
dd	G_M9309_IG04 - G_M9309_IG02
dd	G_M9309_IG05 - G_M9309_IG02
@@ -1324,7 +1324,7 @@ RWD00 dd	G_M9309_IG03 - G_M9309_IG02
dd	G_M9309_IG258 - G_M9309_IG02
-; Total bytes of code 2572, prolog size 4, PerfScore 789.75, instruction count 531, allocated bytes for code 2572 (MethodHash=c6eadba2) for method System.Runtime.Intrinsics.X86.Sse2:ShiftRightLogical128BitLane(System.Runtime.Intrinsics.Vector128`1[ulong],ubyte):System.Runtime.Intrinsics.Vector128`1[ulong] (Tier0)+; Total bytes of code 2591, prolog size 7, PerfScore 789.75, instruction count 531, allocated bytes for code 2591 (MethodHash=c6eadba2) for method System.Runtime.Intrinsics.X86.Sse2:ShiftRightLogical128BitLane(System.Runtime.Intrinsics.Vector128`1[ulong],ubyte):System.Runtime.Intrinsics.Vector128`1[ulong] (Tier0)
; ============================================================
Unwind Info:
@@ -1332,9 +1332,9 @@ Unwind Info:
>> End offset : 0xd1ffab1e (not in unwind data)
Version : 1
Flags : 0x00
- SizeOfProlog : 0x01+ SizeOfProlog : 0x03
CountOfUnwindCodes: 1
FrameRegister : none (0)
FrameOffset : N/A (no FrameRegister) (Value=0)
UnwindCodes :
- CodeOffset: 0x01 UnwindOp: UWOP_PUSH_NONVOL (0) OpInfo: rbp (5)+ CodeOffset: 0x03 UnwindOp: UWOP_PUSH_NONVOL (0) OpInfo: rbp (5)
+1 (+0.85%) : 56215.dasm - Benchmarks.SIMD.RayTracer.Sphere:Normal(Benchmarks.SIMD.RayTracer.Vector):Benchmarks.SIMD.RayTracer.Vector:this (Tier1)
@@ -77,7 +77,7 @@ G_M56301_IG05: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0004 {rdx}, byr
vextractps dword ptr [rdx+0x08], xmm0, 2
mov rax, rdx
; byrRegs +[rax]
- ;; size=31 bbWeight=1 PerfScore 18.25+ ;; size=32 bbWeight=1 PerfScore 18.25
G_M56301_IG06: ; bbWeight=1, epilog, nogc, extend
ret ;; size=1 bbWeight=1 PerfScore 1.00
@@ -85,7 +85,7 @@ RWD00 dd	3F800000h ; 1
RWD04 dd	7F800000h ; inf
-; Total bytes of code 118, prolog size 0, PerfScore 82.58, instruction count 26, allocated bytes for code 118 (MethodHash=3e632412) for method Benchmarks.SIMD.RayTracer.Sphere:Normal(Benchmarks.SIMD.RayTracer.Vector):Benchmarks.SIMD.RayTracer.Vector:this (Tier1)+; Total bytes of code 119, prolog size 0, PerfScore 82.58, instruction count 26, allocated bytes for code 119 (MethodHash=3e632412) for method Benchmarks.SIMD.RayTracer.Sphere:Normal(Benchmarks.SIMD.RayTracer.Vector):Benchmarks.SIMD.RayTracer.Vector:this (Tier1)
; ============================================================
Unwind Info:
+7 (+87.50%) : 12704.dasm - System.UInt16:System.Numerics.INumberBase.get_Zero():ushort (Tier0)
@@ -12,16 +12,16 @@
G_M3961_IG01: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref, nogc <-- Prolog IG
push rbp
mov rbp, rsp
- ;; size=4 bbWeight=1 PerfScore 1.25+ ;; size=7 bbWeight=1 PerfScore 1.25
G_M3961_IG02: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref
xor eax, eax
- ;; size=2 bbWeight=1 PerfScore 0.25+ ;; size=4 bbWeight=1 PerfScore 0.25
G_M3961_IG03: ; bbWeight=1, epilog, nogc, extend
pop rbp
ret - ;; size=2 bbWeight=1 PerfScore 1.50+ ;; size=4 bbWeight=1 PerfScore 1.50-; Total bytes of code 8, prolog size 4, PerfScore 3.00, instruction count 5, allocated bytes for code 8 (MethodHash=985cf086) for method System.UInt16:System.Numerics.INumberBase<System.UInt16>.get_Zero():ushort (Tier0)+; Total bytes of code 15, prolog size 7, PerfScore 3.00, instruction count 5, allocated bytes for code 15 (MethodHash=985cf086) for method System.UInt16:System.Numerics.INumberBase<System.UInt16>.get_Zero():ushort (Tier0)
; ============================================================
Unwind Info:
@@ -29,9 +29,9 @@ Unwind Info:
>> End offset : 0xd1ffab1e (not in unwind data)
Version : 1
Flags : 0x00
- SizeOfProlog : 0x01+ SizeOfProlog : 0x03
CountOfUnwindCodes: 1
FrameRegister : none (0)
FrameOffset : N/A (no FrameRegister) (Value=0)
UnwindCodes :
- CodeOffset: 0x01 UnwindOp: UWOP_PUSH_NONVOL (0) OpInfo: rbp (5)+ CodeOffset: 0x03 UnwindOp: UWOP_PUSH_NONVOL (0) OpInfo: rbp (5)
+7 (+87.50%) : 73294.dasm - System.Byte:System.Numerics.INumberBase.get_Zero():ubyte (Tier0)
@@ -12,16 +12,16 @@
G_M54785_IG01: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref, nogc <-- Prolog IG
push rbp
mov rbp, rsp
- ;; size=4 bbWeight=1 PerfScore 1.25+ ;; size=7 bbWeight=1 PerfScore 1.25
G_M54785_IG02: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref
xor eax, eax
- ;; size=2 bbWeight=1 PerfScore 0.25+ ;; size=4 bbWeight=1 PerfScore 0.25
G_M54785_IG03: ; bbWeight=1, epilog, nogc, extend
pop rbp
ret - ;; size=2 bbWeight=1 PerfScore 1.50+ ;; size=4 bbWeight=1 PerfScore 1.50-; Total bytes of code 8, prolog size 4, PerfScore 3.00, instruction count 5, allocated bytes for code 8 (MethodHash=e45a29fe) for method System.Byte:System.Numerics.INumberBase<System.Byte>.get_Zero():ubyte (Tier0)+; Total bytes of code 15, prolog size 7, PerfScore 3.00, instruction count 5, allocated bytes for code 15 (MethodHash=e45a29fe) for method System.Byte:System.Numerics.INumberBase<System.Byte>.get_Zero():ubyte (Tier0)
; ============================================================
Unwind Info:
@@ -29,9 +29,9 @@ Unwind Info:
>> End offset : 0xd1ffab1e (not in unwind data)
Version : 1
Flags : 0x00
- SizeOfProlog : 0x01+ SizeOfProlog : 0x03
CountOfUnwindCodes: 1
FrameRegister : none (0)
FrameOffset : N/A (no FrameRegister) (Value=0)
UnwindCodes :
- CodeOffset: 0x01 UnwindOp: UWOP_PUSH_NONVOL (0) OpInfo: rbp (5)+ CodeOffset: 0x03 UnwindOp: UWOP_PUSH_NONVOL (0) OpInfo: rbp (5)
+7 (+87.50%) : 15012.dasm - System.Runtime.CompilerServices.RuntimeHelpers:IsBitwiseEquatable[System.Reflection.Emit.OpCode]():ubyte (Tier0)
@@ -12,16 +12,16 @@
G_M969_IG01: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref, nogc <-- Prolog IG
push rbp
mov rbp, rsp
- ;; size=4 bbWeight=1 PerfScore 1.25+ ;; size=7 bbWeight=1 PerfScore 1.25
G_M969_IG02: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref
xor eax, eax
- ;; size=2 bbWeight=1 PerfScore 0.25+ ;; size=4 bbWeight=1 PerfScore 0.25
G_M969_IG03: ; bbWeight=1, epilog, nogc, extend
pop rbp
ret - ;; size=2 bbWeight=1 PerfScore 1.50+ ;; size=4 bbWeight=1 PerfScore 1.50-; Total bytes of code 8, prolog size 4, PerfScore 3.00, instruction count 5, allocated bytes for code 8 (MethodHash=f362fc36) for method System.Runtime.CompilerServices.RuntimeHelpers:IsBitwiseEquatable[System.Reflection.Emit.OpCode]():ubyte (Tier0)+; Total bytes of code 15, prolog size 7, PerfScore 3.00, instruction count 5, allocated bytes for code 15 (MethodHash=f362fc36) for method System.Runtime.CompilerServices.RuntimeHelpers:IsBitwiseEquatable[System.Reflection.Emit.OpCode]():ubyte (Tier0)
; ============================================================
Unwind Info:
@@ -29,9 +29,9 @@ Unwind Info:
>> End offset : 0xd1ffab1e (not in unwind data)
Version : 1
Flags : 0x00
- SizeOfProlog : 0x01+ SizeOfProlog : 0x03
CountOfUnwindCodes: 1
FrameRegister : none (0)
FrameOffset : N/A (no FrameRegister) (Value=0)
UnwindCodes :
- CodeOffset: 0x01 UnwindOp: UWOP_PUSH_NONVOL (0) OpInfo: rbp (5)+ CodeOffset: 0x03 UnwindOp: UWOP_PUSH_NONVOL (0) OpInfo: rbp (5)
coreclr_tests.run.windows.x64.checked.mch
-1 (-16.67%) : 509186.dasm - System.Runtime.Intrinsics.X86.Popcnt+X64:get_IsSupported():ubyte (FullOpts)
@@ -14,13 +14,13 @@
G_M37565_IG01: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref, nogc <-- Prolog IG
;; size=0 bbWeight=1 PerfScore 0.00
G_M37565_IG02: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref
- mov eax, 1- ;; size=5 bbWeight=1 PerfScore 0.25+ xor eax, eax+ ;; size=4 bbWeight=1 PerfScore 0.25
G_M37565_IG03: ; bbWeight=1, epilog, nogc, extend
ret ;; size=1 bbWeight=1 PerfScore 1.00
-; Total bytes of code 6, prolog size 0, PerfScore 1.25, instruction count 2, allocated bytes for code 6 (MethodHash=4c306d42) for method System.Runtime.Intrinsics.X86.Popcnt+X64:get_IsSupported():ubyte (FullOpts)+; Total bytes of code 5, prolog size 0, PerfScore 1.25, instruction count 2, allocated bytes for code 5 (MethodHash=4c306d42) for method System.Runtime.Intrinsics.X86.Popcnt+X64:get_IsSupported():ubyte (FullOpts)
; ============================================================
Unwind Info:
-1 (-16.67%) : 579101.dasm - Runtime_34587:get_PopcntX64IsSupported():ubyte (FullOpts)
@@ -14,13 +14,13 @@
G_M19947_IG01: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref, nogc <-- Prolog IG
;; size=0 bbWeight=1 PerfScore 0.00
G_M19947_IG02: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref
- mov eax, 1- ;; size=5 bbWeight=1 PerfScore 0.25+ xor eax, eax+ ;; size=4 bbWeight=1 PerfScore 0.25
G_M19947_IG03: ; bbWeight=1, epilog, nogc, extend
ret ;; size=1 bbWeight=1 PerfScore 1.00
-; Total bytes of code 6, prolog size 0, PerfScore 1.25, instruction count 2, allocated bytes for code 6 (MethodHash=801bb214) for method Runtime_34587:get_PopcntX64IsSupported():ubyte (FullOpts)+; Total bytes of code 5, prolog size 0, PerfScore 1.25, instruction count 2, allocated bytes for code 5 (MethodHash=801bb214) for method Runtime_34587:get_PopcntX64IsSupported():ubyte (FullOpts)
; ============================================================
Unwind Info:
-12 (-13.19%) : 510272.dasm - IntelHardwareIntrinsicTest.Program:TestEntryPoint():int (FullOpts)
@@ -12,8 +12,8 @@
;* V01 loc1 [V01 ] ( 0, 0 ) ref -> zero-ref class-hnd <<unknown class>>
; V02 OutArgs [V02 ] ( 1, 1 ) struct (32) [rsp+0x00] do-not-enreg[XS] addr-exposed "OutgoingArgSpace"
;* V03 tmp1 [V03 ] ( 0, 0 ) int -> zero-ref -; V04 tmp2 [V04,T01] ( 2, 0 ) ref -> rdx class-hnd single-def "impSpillSpecialSideEff" <<unknown class>>-; V05 tmp3 [V05,T02] ( 2, 0 ) int -> [rbp-0x04] do-not-enreg[M] EH-live+;* V04 tmp2 [V04 ] ( 0, 0 ) ref -> zero-ref class-hnd "impSpillSpecialSideEff" <<unknown class>>+;* V05 tmp3 [V05 ] ( 0, 0 ) int -> zero-ref 
; V06 PSPSym [V06,T00] ( 1, 1 ) long -> [rbp-0x10] do-not-enreg[V] "PSPSym"
;
; Lcl frame size = 48
@@ -23,48 +23,37 @@ G_M30609_IG01: ; bbWeight=0, gcVars=0000000000000000 {}, gcrefRegs=0000 {
sub rsp, 48
lea rbp, [rsp+0x30]
mov qword ptr [rbp-0x10], rsp
- ;; size=14 bbWeight=0 PerfScore 0.00+ ;; size=19 bbWeight=0 PerfScore 0.00
G_M30609_IG02: ; bbWeight=0, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref
call CORINFO_HELP_THROW_PLATFORM_NOT_SUPPORTED
; gcr arg pop 0
int3 ;; size=6 bbWeight=0 PerfScore 0.00
G_M30609_IG03: ; bbWeight=0, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref
- mov eax, dword ptr [rbp-0x04]- ;; size=3 bbWeight=0 PerfScore 0.00+ xor eax, eax+ ;; size=4 bbWeight=0 PerfScore 0.00
G_M30609_IG04: ; bbWeight=0, epilog, nogc, extend
add rsp, 48
pop rbp
ret - ;; size=6 bbWeight=0 PerfScore 0.00-G_M30609_IG05: ; bbWeight=0, gcrefRegs=0004 {rdx}, byrefRegs=0000 {}, byref, funclet prolog, nogc- ; gcrRegs +[rdx]+ ;; size=9 bbWeight=0 PerfScore 0.00+G_M30609_IG05: ; bbWeight=0, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref, funclet prolog, nogc
push rbp
sub rsp, 48
mov rbp, qword ptr [rcx+0x20]
mov qword ptr [rsp+0x20], rbp
lea rbp, [rbp+0x30]
- ;; size=18 bbWeight=0 PerfScore 0.00-G_M30609_IG06: ; bbWeight=0, gcVars=0000000000000000 {}, gcrefRegs=0004 {rdx}, byrefRegs=0000 {}, gcvars, byref- mov rcx, 0xD1FFAB1E ; <unknown class>- call CORINFO_HELP_ISINSTANCEOFCLASS- ; gcrRegs -[rdx] +[rax]- ; gcr arg pop 0- xor ecx, ecx- mov edx, 100- test rax, rax- cmovne ecx, edx- mov dword ptr [rbp-0x04], ecx+ ;; size=24 bbWeight=0 PerfScore 0.00+G_M30609_IG06: ; bbWeight=0, gcVars=0000000000000000 {}, gcrefRegs=0000 {}, byrefRegs=0000 {}, gcvars, byref
lea rax, G_M30609_IG03
- ; gcrRegs -[rax]- ;; size=38 bbWeight=0 PerfScore 0.00+ ;; size=8 bbWeight=0 PerfScore 0.00
G_M30609_IG07: ; bbWeight=0, funclet epilog, nogc, extend
add rsp, 48
pop rbp
ret - ;; size=6 bbWeight=0 PerfScore 0.00+ ;; size=9 bbWeight=0 PerfScore 0.00-; Total bytes of code 91, prolog size 14, PerfScore 0.00, instruction count 26, allocated bytes for code 91 (MethodHash=3926886e) for method IntelHardwareIntrinsicTest.Program:TestEntryPoint():int (FullOpts)+; Total bytes of code 79, prolog size 19, PerfScore 0.00, instruction count 19, allocated bytes for code 79 (MethodHash=3926886e) for method IntelHardwareIntrinsicTest.Program:TestEntryPoint():int (FullOpts)
; ============================================================
Unwind Info:
@@ -72,25 +61,25 @@ Unwind Info:
>> End offset : 0xd1ffab1e (not in unwind data)
Version : 1
Flags : 0x00
- SizeOfProlog : 0x05+ SizeOfProlog : 0x08
CountOfUnwindCodes: 2
FrameRegister : none (0)
FrameOffset : N/A (no FrameRegister) (Value=0)
UnwindCodes :
- CodeOffset: 0x05 UnwindOp: UWOP_ALLOC_SMALL (2) OpInfo: 5 * 8 + 8 = 48 = 0x30- CodeOffset: 0x01 UnwindOp: UWOP_PUSH_NONVOL (0) OpInfo: rbp (5)+ CodeOffset: 0x08 UnwindOp: UWOP_ALLOC_SMALL (2) OpInfo: 5 * 8 + 8 = 48 = 0x30+ CodeOffset: 0x03 UnwindOp: UWOP_PUSH_NONVOL (0) OpInfo: rbp (5)
Unwind Info:
>> Start offset : 0xd1ffab1e (not in unwind data)
>> End offset : 0xd1ffab1e (not in unwind data)
Version : 1
Flags : 0x00
- SizeOfProlog : 0x05+ SizeOfProlog : 0x08
CountOfUnwindCodes: 2
FrameRegister : none (0)
FrameOffset : N/A (no FrameRegister) (Value=0)
UnwindCodes :
- CodeOffset: 0x05 UnwindOp: UWOP_ALLOC_SMALL (2) OpInfo: 5 * 8 + 8 = 48 = 0x30- CodeOffset: 0x01 UnwindOp: UWOP_PUSH_NONVOL (0) OpInfo: rbp (5)+ CodeOffset: 0x08 UnwindOp: UWOP_ALLOC_SMALL (2) OpInfo: 5 * 8 + 8 = 48 = 0x30+ CodeOffset: 0x03 UnwindOp: UWOP_PUSH_NONVOL (0) OpInfo: rbp (5)
*************** EH table for IntelHardwareIntrinsicTest.Program:TestEntryPoint():int
1 EH table entries, 0 duplicate clauses, 0 cloned finallys, 1 total EH entries reported to VM
EH#0: try [G_M30609_IG02..G_M30609_IG03) handled by [G_M30609_IG05..END) (class: 100000B)
+7 (+87.50%) : 149326.dasm - System.Runtime.CompilerServices.RuntimeHelpers:IsReferenceOrContainsReferences[double]():ubyte (Tier0)
@@ -12,16 +12,16 @@
G_M50835_IG01: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref, nogc <-- Prolog IG
push rbp
mov rbp, rsp
- ;; size=4 bbWeight=1 PerfScore 1.25+ ;; size=7 bbWeight=1 PerfScore 1.25
G_M50835_IG02: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref
xor eax, eax
- ;; size=2 bbWeight=1 PerfScore 0.25+ ;; size=4 bbWeight=1 PerfScore 0.25
G_M50835_IG03: ; bbWeight=1, epilog, nogc, extend
pop rbp
ret - ;; size=2 bbWeight=1 PerfScore 1.50+ ;; size=4 bbWeight=1 PerfScore 1.50-; Total bytes of code 8, prolog size 4, PerfScore 3.00, instruction count 5, allocated bytes for code 8 (MethodHash=94b8396c) for method System.Runtime.CompilerServices.RuntimeHelpers:IsReferenceOrContainsReferences[double]():ubyte (Tier0)+; Total bytes of code 15, prolog size 7, PerfScore 3.00, instruction count 5, allocated bytes for code 15 (MethodHash=94b8396c) for method System.Runtime.CompilerServices.RuntimeHelpers:IsReferenceOrContainsReferences[double]():ubyte (Tier0)
; ============================================================
Unwind Info:
@@ -29,9 +29,9 @@ Unwind Info:
>> End offset : 0xd1ffab1e (not in unwind data)
Version : 1
Flags : 0x00
- SizeOfProlog : 0x01+ SizeOfProlog : 0x03
CountOfUnwindCodes: 1
FrameRegister : none (0)
FrameOffset : N/A (no FrameRegister) (Value=0)
UnwindCodes :
- CodeOffset: 0x01 UnwindOp: UWOP_PUSH_NONVOL (0) OpInfo: rbp (5)+ CodeOffset: 0x03 UnwindOp: UWOP_PUSH_NONVOL (0) OpInfo: rbp (5)
+7 (+87.50%) : 270935.dasm - Benchstone.BenchF.LLoops:Clock():int (Instrumented Tier0)
@@ -12,16 +12,16 @@
G_M63398_IG01: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref, nogc <-- Prolog IG
push rbp
mov rbp, rsp
- ;; size=4 bbWeight=1 PerfScore 1.25+ ;; size=7 bbWeight=1 PerfScore 1.25
G_M63398_IG02: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref
xor eax, eax
- ;; size=2 bbWeight=1 PerfScore 0.25+ ;; size=4 bbWeight=1 PerfScore 0.25
G_M63398_IG03: ; bbWeight=1, epilog, nogc, extend
pop rbp
ret - ;; size=2 bbWeight=1 PerfScore 1.50+ ;; size=4 bbWeight=1 PerfScore 1.50-; Total bytes of code 8, prolog size 4, PerfScore 3.00, instruction count 5, allocated bytes for code 8 (MethodHash=e4d00859) for method Benchstone.BenchF.LLoops:Clock():int (Instrumented Tier0)+; Total bytes of code 15, prolog size 7, PerfScore 3.00, instruction count 5, allocated bytes for code 15 (MethodHash=e4d00859) for method Benchstone.BenchF.LLoops:Clock():int (Instrumented Tier0)
; ============================================================
Unwind Info:
@@ -29,9 +29,9 @@ Unwind Info:
>> End offset : 0xd1ffab1e (not in unwind data)
Version : 1
Flags : 0x00
- SizeOfProlog : 0x01+ SizeOfProlog : 0x03
CountOfUnwindCodes: 1
FrameRegister : none (0)
FrameOffset : N/A (no FrameRegister) (Value=0)
UnwindCodes :
- CodeOffset: 0x01 UnwindOp: UWOP_PUSH_NONVOL (0) OpInfo: rbp (5)+ CodeOffset: 0x03 UnwindOp: UWOP_PUSH_NONVOL (0) OpInfo: rbp (5)
+10 (+90.91%) : 526916.dasm - BringUpTest_NotAndNeg:NotAndNeg(int,int):int (FullOpts)
@@ -21,12 +21,12 @@ G_M23640_IG02: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref
mov ecx, edx
not ecx
xor eax, ecx
- ;; size=10 bbWeight=1 PerfScore 1.25+ ;; size=20 bbWeight=1 PerfScore 1.25
G_M23640_IG03: ; bbWeight=1, epilog, nogc, extend
ret ;; size=1 bbWeight=1 PerfScore 1.00
-; Total bytes of code 11, prolog size 0, PerfScore 2.25, instruction count 6, allocated bytes for code 11 (MethodHash=81d2a3a7) for method BringUpTest_NotAndNeg:NotAndNeg(int,int):int (FullOpts)+; Total bytes of code 21, prolog size 0, PerfScore 2.25, instruction count 6, allocated bytes for code 21 (MethodHash=81d2a3a7) for method BringUpTest_NotAndNeg:NotAndNeg(int,int):int (FullOpts)
; ============================================================
Unwind Info:
libraries.crossgen2.windows.x64.checked.mch
+2 (+0.08%) : 30851.dasm - System.Globalization.CalendricalCalculationsHelper:SumLongSequenceOfPeriodicTerms(double):double (FullOpts)
@@ -174,7 +174,7 @@ G_M54838_IG01: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref,
movaps xmmword ptr [rsp+0x30], xmm9
movaps xmmword ptr [rsp+0x20], xmm10
movaps xmm6, xmm0
- ;; size=35 bbWeight=1 PerfScore 10.50+ ;; size=36 bbWeight=1 PerfScore 10.50
G_M54838_IG02: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref
movaps xmm0, xmm6
mulsd xmm0, qword ptr [reloc @RWD00]
@@ -643,7 +643,7 @@ G_M54838_IG08: ; bbWeight=1, epilog, nogc, extend
movaps xmm10, xmmword ptr [rsp+0x20]
add rsp, 120
ret - ;; size=33 bbWeight=1 PerfScore 21.25+ ;; size=34 bbWeight=1 PerfScore 21.25
RWD00 dq	3FEDB8A420DC189Ah	; 0.9287892
RWD08 dq	4070E8C71B478423h	; 270.54861
RWD16 dq	400921FB54442D18h	; 3.14159265
@@ -788,7 +788,7 @@ RWD1120	dq	40F5FD9C72B020C5h	; 90073.778
RWD1128	dq	4062433333333333h	; 146.1
-; Total bytes of code 2395, prolog size 32, PerfScore 1804.33, instruction count 413, allocated bytes for code 2395 (MethodHash=72e629c9) for method System.Globalization.CalendricalCalculationsHelper:SumLongSequenceOfPeriodicTerms(double):double (FullOpts)+; Total bytes of code 2397, prolog size 33, PerfScore 1804.33, instruction count 413, allocated bytes for code 2397 (MethodHash=72e629c9) for method System.Globalization.CalendricalCalculationsHelper:SumLongSequenceOfPeriodicTerms(double):double (FullOpts)
; ============================================================
Unwind Info:
@@ -796,19 +796,19 @@ Unwind Info:
>> End offset : 0xd1ffab1e (not in unwind data)
Version : 1
Flags : 0x00
- SizeOfProlog : 0x20+ SizeOfProlog : 0x21
CountOfUnwindCodes: 11
FrameRegister : none (0)
FrameOffset : N/A (no FrameRegister) (Value=0)
UnwindCodes :
- CodeOffset: 0x20 UnwindOp: UWOP_SAVE_XMM128 (8) OpInfo: XMM10 (10)+ CodeOffset: 0x21 UnwindOp: UWOP_SAVE_XMM128 (8) OpInfo: XMM10 (10)
Scaled Small Offset: 2 * 16 = 32 = 0x00020
- CodeOffset: 0x1A UnwindOp: UWOP_SAVE_XMM128 (8) OpInfo: XMM9 (9)+ CodeOffset: 0x1B UnwindOp: UWOP_SAVE_XMM128 (8) OpInfo: XMM9 (9)
Scaled Small Offset: 3 * 16 = 48 = 0x00030
- CodeOffset: 0x14 UnwindOp: UWOP_SAVE_XMM128 (8) OpInfo: XMM8 (8)+ CodeOffset: 0x15 UnwindOp: UWOP_SAVE_XMM128 (8) OpInfo: XMM8 (8)
Scaled Small Offset: 4 * 16 = 64 = 0x00040
- CodeOffset: 0x0E UnwindOp: UWOP_SAVE_XMM128 (8) OpInfo: XMM7 (7)+ CodeOffset: 0x0F UnwindOp: UWOP_SAVE_XMM128 (8) OpInfo: XMM7 (7)
Scaled Small Offset: 5 * 16 = 80 = 0x00050
- CodeOffset: 0x09 UnwindOp: UWOP_SAVE_XMM128 (8) OpInfo: XMM6 (6)+ CodeOffset: 0x0A UnwindOp: UWOP_SAVE_XMM128 (8) OpInfo: XMM6 (6)
Scaled Small Offset: 6 * 16 = 96 = 0x00060
- CodeOffset: 0x04 UnwindOp: UWOP_ALLOC_SMALL (2) OpInfo: 14 * 8 + 8 = 120 = 0x78+ CodeOffset: 0x05 UnwindOp: UWOP_ALLOC_SMALL (2) OpInfo: 14 * 8 + 8 = 120 = 0x78
+1 (+0.11%) : 15946.dasm - System.Runtime.Intrinsics.Vector512:Create(byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte):System.Runtime.Intrinsics.Vector512`1[byte] (FullOpts)
@@ -245,12 +245,12 @@ G_M31854_IG03: ; bbWeight=1, extend
movups xmmword ptr [rcx+0x30], xmm3
mov rax, rcx
; byrRegs +[rax]
- ;; size=436 bbWeight=1 PerfScore 175.25+ ;; size=437 bbWeight=1 PerfScore 175.25
G_M31854_IG04: ; bbWeight=1, epilog, nogc, extend
ret ;; size=1 bbWeight=1 PerfScore 1.00
-; Total bytes of code 923, prolog size 0, PerfScore 381.00, instruction count 134, allocated bytes for code 923 (MethodHash=6feb8391) for method System.Runtime.Intrinsics.Vector512:Create(byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte):System.Runtime.Intrinsics.Vector512`1[byte] (FullOpts)+; Total bytes of code 924, prolog size 0, PerfScore 381.00, instruction count 134, allocated bytes for code 924 (MethodHash=6feb8391) for method System.Runtime.Intrinsics.Vector512:Create(byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte):System.Runtime.Intrinsics.Vector512`1[byte] (FullOpts)
; ============================================================
Unwind Info:
+1 (+0.23%) : 15798.dasm - System.Runtime.Intrinsics.Vector256:Create(byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte):System.Runtime.Intrinsics.Vector256`1[byte] (FullOpts)
@@ -126,12 +126,12 @@ G_M51694_IG02: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0002 {rcx}, byr
movups xmmword ptr [rcx+0x10], xmm1
mov rax, rcx
; byrRegs +[rax]
- ;; size=438 bbWeight=1 PerfScore 186.00+ ;; size=439 bbWeight=1 PerfScore 186.00
G_M51694_IG03: ; bbWeight=1, epilog, nogc, extend
ret ;; size=1 bbWeight=1 PerfScore 1.00
-; Total bytes of code 439, prolog size 0, PerfScore 187.00, instruction count 68, allocated bytes for code 439 (MethodHash=4b683611) for method System.Runtime.Intrinsics.Vector256:Create(byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte):System.Runtime.Intrinsics.Vector256`1[byte] (FullOpts)+; Total bytes of code 440, prolog size 0, PerfScore 187.00, instruction count 68, allocated bytes for code 440 (MethodHash=4b683611) for method System.Runtime.Intrinsics.Vector256:Create(byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte,byte):System.Runtime.Intrinsics.Vector256`1[byte] (FullOpts)
; ============================================================
Unwind Info:
+6 (+85.71%) : 11664.dasm - System.UInt32:System.Numerics.IShiftOperators.op_LeftShift(uint,int):uint (FullOpts)
@@ -16,16 +16,16 @@
G_M41089_IG01: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref, nogc <-- Prolog IG
mov eax, ecx
- ;; size=2 bbWeight=1 PerfScore 0.25+ ;; size=4 bbWeight=1 PerfScore 0.25
G_M41089_IG02: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref
mov ecx, edx
shl eax, cl
- ;; size=4 bbWeight=1 PerfScore 2.25+ ;; size=8 bbWeight=1 PerfScore 2.25
G_M41089_IG03: ; bbWeight=1, epilog, nogc, extend
ret ;; size=1 bbWeight=1 PerfScore 1.00
-; Total bytes of code 7, prolog size 0, PerfScore 3.50, instruction count 4, allocated bytes for code 7 (MethodHash=9e7e5f7e) for method System.UInt32:System.Numerics.IShiftOperators<System.UInt32,System.Int32,System.UInt32>.op_LeftShift(uint,int):uint (FullOpts)+; Total bytes of code 13, prolog size 0, PerfScore 3.50, instruction count 4, allocated bytes for code 13 (MethodHash=9e7e5f7e) for method System.UInt32:System.Numerics.IShiftOperators<System.UInt32,System.Int32,System.UInt32>.op_LeftShift(uint,int):uint (FullOpts)
; ============================================================
Unwind Info:
+34 (+87.18%) : 170393.dasm - Microsoft.CodeAnalysis.CachingBase`1[System.__Canon]:AlignSize(int):int (FullOpts)
@@ -34,12 +34,12 @@ G_M65205_IG02: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref
sar eax, 16
or eax, edx
inc eax
- ;; size=38 bbWeight=1 PerfScore 5.50+ ;; size=72 bbWeight=1 PerfScore 5.50
G_M65205_IG03: ; bbWeight=1, epilog, nogc, extend
ret ;; size=1 bbWeight=1 PerfScore 1.00
-; Total bytes of code 39, prolog size 0, PerfScore 6.50, instruction count 18, allocated bytes for code 39 (MethodHash=8bf1014a) for method Microsoft.CodeAnalysis.CachingBase`1[System.__Canon]:AlignSize(int):int (FullOpts)+; Total bytes of code 73, prolog size 0, PerfScore 6.50, instruction count 18, allocated bytes for code 73 (MethodHash=8bf1014a) for method Microsoft.CodeAnalysis.CachingBase`1[System.__Canon]:AlignSize(int):int (FullOpts)
; ============================================================
Unwind Info:
+8 (+88.89%) : 126160.dasm - Microsoft.Diagnostics.Tracing.Ctf.IntHelpers:AlignDown(int,int):int (FullOpts)
@@ -21,12 +21,12 @@ G_M35496_IG02: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref
mov eax, edx
not eax
and eax, ecx
- ;; size=8 bbWeight=1 PerfScore 1.00+ ;; size=16 bbWeight=1 PerfScore 1.00
G_M35496_IG03: ; bbWeight=1, epilog, nogc, extend
ret ;; size=1 bbWeight=1 PerfScore 1.00
-; Total bytes of code 9, prolog size 0, PerfScore 2.00, instruction count 5, allocated bytes for code 9 (MethodHash=596f7557) for method Microsoft.Diagnostics.Tracing.Ctf.IntHelpers:AlignDown(int,int):int (FullOpts)+; Total bytes of code 17, prolog size 0, PerfScore 2.00, instruction count 5, allocated bytes for code 17 (MethodHash=596f7557) for method Microsoft.Diagnostics.Tracing.Ctf.IntHelpers:AlignDown(int,int):int (FullOpts)
; ============================================================
Unwind Info:
libraries.pmi.windows.x64.checked.mch
-1 (-16.67%) : 29280.dasm - System.Runtime.Intrinsics.X86.Popcnt+X64:get_IsSupported():ubyte (FullOpts)
@@ -14,13 +14,13 @@
G_M37565_IG01: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref, nogc <-- Prolog IG
;; size=0 bbWeight=1 PerfScore 0.00
G_M37565_IG02: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref
- mov eax, 1- ;; size=5 bbWeight=1 PerfScore 0.25+ xor eax, eax+ ;; size=4 bbWeight=1 PerfScore 0.25
G_M37565_IG03: ; bbWeight=1, epilog, nogc, extend
ret ;; size=1 bbWeight=1 PerfScore 1.00
-; Total bytes of code 6, prolog size 0, PerfScore 1.25, instruction count 2, allocated bytes for code 6 (MethodHash=4c306d42) for method System.Runtime.Intrinsics.X86.Popcnt+X64:get_IsSupported():ubyte (FullOpts)+; Total bytes of code 5, prolog size 0, PerfScore 1.25, instruction count 2, allocated bytes for code 5 (MethodHash=4c306d42) for method System.Runtime.Intrinsics.X86.Popcnt+X64:get_IsSupported():ubyte (FullOpts)
; ============================================================
Unwind Info:
+1 (+0.14%) : 22972.dasm - System.Numerics.Matrix4x4+Impl:Transform(byref,byref):System.Numerics.Matrix4x4+Impl (FullOpts)
@@ -199,14 +199,14 @@ G_M8955_IG03: ; bbWeight=1, extend
vmovups xmmword ptr [rcx+0x30], xmm0
mov rax, rcx
; byrRegs +[rax]
- ;; size=354 bbWeight=1 PerfScore 161.25+ ;; size=355 bbWeight=1 PerfScore 161.25
G_M8955_IG04: ; bbWeight=1, epilog, nogc, extend
ret ;; size=1 bbWeight=1 PerfScore 1.00
RWD00 dd	3F800000h ; 1
-; Total bytes of code 693, prolog size 0, PerfScore 345.25, instruction count 118, allocated bytes for code 693 (MethodHash=5906dd04) for method System.Numerics.Matrix4x4+Impl:Transform(byref,byref):System.Numerics.Matrix4x4+Impl (FullOpts)+; Total bytes of code 694, prolog size 0, PerfScore 345.25, instruction count 118, allocated bytes for code 694 (MethodHash=5906dd04) for method System.Numerics.Matrix4x4+Impl:Transform(byref,byref):System.Numerics.Matrix4x4+Impl (FullOpts)
; ============================================================
Unwind Info:
+5 (+0.16%) : 28844.dasm - System.Runtime.Intrinsics.X86.Avx512F:Shuffle(System.Runtime.Intrinsics.Vector512`1[float],System.Runtime.Intrinsics.Vector512`1[float],ubyte):System.Runtime.Intrinsics.Vector512`1[float] (FullOpts)
@@ -29,7 +29,7 @@ G_M43564_IG02: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0106 {rcx rdx r
; byrRegs -[rdx]
add r8, rdx
jmp r8
- ;; size=40 bbWeight=1 PerfScore 14.00+ ;; size=44 bbWeight=1 PerfScore 14.00
G_M43564_IG03: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0002 {rcx}, byref
vshufps zmm0, zmm0, zmm1, 0
jmp G_M43564_IG259
@@ -1058,7 +1058,7 @@ G_M43564_IG259: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0002 {rcx}, by
vmovups zmmword ptr [rcx], zmm0
mov rax, rcx
; byrRegs +[rax]
- ;; size=9 bbWeight=1 PerfScore 2.25+ ;; size=10 bbWeight=1 PerfScore 2.25
G_M43564_IG260: ; bbWeight=1, epilog, nogc, extend
vzeroupper ret @@ -1321,7 +1321,7 @@ RWD00 dd	G_M43564_IG03 - G_M43564_IG02
dd	G_M43564_IG258 - G_M43564_IG02
-; Total bytes of code 3083, prolog size 0, PerfScore 786.25, instruction count 524, allocated bytes for code 3083 (MethodHash=20a155d3) for method System.Runtime.Intrinsics.X86.Avx512F:Shuffle(System.Runtime.Intrinsics.Vector512`1[float],System.Runtime.Intrinsics.Vector512`1[float],ubyte):System.Runtime.Intrinsics.Vector512`1[float] (FullOpts)+; Total bytes of code 3088, prolog size 0, PerfScore 786.25, instruction count 524, allocated bytes for code 3088 (MethodHash=20a155d3) for method System.Runtime.Intrinsics.X86.Avx512F:Shuffle(System.Runtime.Intrinsics.Vector512`1[float],System.Runtime.Intrinsics.Vector512`1[float],ubyte):System.Runtime.Intrinsics.Vector512`1[float] (FullOpts)
; ============================================================
Unwind Info:
+34 (+87.18%) : 137491.dasm - Microsoft.CodeAnalysis.CachingBase`1[ubyte]:AlignSize(int):int (FullOpts)
@@ -32,12 +32,12 @@ G_M17100_IG02: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref
sar eax, 16
or eax, ecx
inc eax
- ;; size=38 bbWeight=1 PerfScore 5.50+ ;; size=72 bbWeight=1 PerfScore 5.50
G_M17100_IG03: ; bbWeight=1, epilog, nogc, extend
ret ;; size=1 bbWeight=1 PerfScore 1.00
-; Total bytes of code 39, prolog size 0, PerfScore 6.50, instruction count 18, allocated bytes for code 39 (MethodHash=31c8bd33) for method Microsoft.CodeAnalysis.CachingBase`1[ubyte]:AlignSize(int):int (FullOpts)+; Total bytes of code 73, prolog size 0, PerfScore 6.50, instruction count 18, allocated bytes for code 73 (MethodHash=31c8bd33) for method Microsoft.CodeAnalysis.CachingBase`1[ubyte]:AlignSize(int):int (FullOpts)
; ============================================================
Unwind Info:
+5 (+166.67%) : 29884.dasm - System.Runtime.Intrinsics.X86.X86Serialize+X64:get_IsSupported():ubyte (FullOpts)
@@ -14,13 +14,13 @@
G_M34763_IG01: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref, nogc <-- Prolog IG
;; size=0 bbWeight=1 PerfScore 0.00
G_M34763_IG02: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref
- xor eax, eax- ;; size=2 bbWeight=1 PerfScore 0.25+ mov eax, 1+ ;; size=7 bbWeight=1 PerfScore 0.25
G_M34763_IG03: ; bbWeight=1, epilog, nogc, extend
ret ;; size=1 bbWeight=1 PerfScore 1.00
-; Total bytes of code 3, prolog size 0, PerfScore 1.25, instruction count 2, allocated bytes for code 3 (MethodHash=bcd77834) for method System.Runtime.Intrinsics.X86.X86Serialize+X64:get_IsSupported():ubyte (FullOpts)+; Total bytes of code 8, prolog size 0, PerfScore 1.25, instruction count 2, allocated bytes for code 8 (MethodHash=bcd77834) for method System.Runtime.Intrinsics.X86.X86Serialize+X64:get_IsSupported():ubyte (FullOpts)
; ============================================================
Unwind Info:
+5 (+166.67%) : 29207.dasm - System.Runtime.Intrinsics.X86.AvxVnni+X64:get_IsSupported():ubyte (FullOpts)
@@ -14,13 +14,13 @@
G_M31227_IG01: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref, nogc <-- Prolog IG
;; size=0 bbWeight=1 PerfScore 0.00
G_M31227_IG02: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref
- xor eax, eax- ;; size=2 bbWeight=1 PerfScore 0.25+ mov eax, 1+ ;; size=7 bbWeight=1 PerfScore 0.25
G_M31227_IG03: ; bbWeight=1, epilog, nogc, extend
ret ;; size=1 bbWeight=1 PerfScore 1.00
-; Total bytes of code 3, prolog size 0, PerfScore 1.25, instruction count 2, allocated bytes for code 3 (MethodHash=e6278604) for method System.Runtime.Intrinsics.X86.AvxVnni+X64:get_IsSupported():ubyte (FullOpts)+; Total bytes of code 8, prolog size 0, PerfScore 1.25, instruction count 2, allocated bytes for code 8 (MethodHash=e6278604) for method System.Runtime.Intrinsics.X86.AvxVnni+X64:get_IsSupported():ubyte (FullOpts)
; ============================================================
Unwind Info:
libraries_tests.run.windows.x64.Release.mch
+2 (+0.08%) : 600803.dasm - System.Globalization.CalendricalCalculationsHelper:SumLongSequenceOfPeriodicTerms(double):double (Instrumented Tier1)
@@ -176,7 +176,7 @@ G_M54838_IG01: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref,
vmovaps xmmword ptr [rsp+0x40], xmm9
vmovaps xmmword ptr [rsp+0x30], xmm10
vmovaps xmm6, xmm0
- ;; size=41 bbWeight=1 PerfScore 10.50+ ;; size=42 bbWeight=1 PerfScore 10.50
G_M54838_IG02: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref
vmulsd xmm0, xmm6, qword ptr [reloc @RWD00]
vaddsd xmm0, xmm0, qword ptr [reloc @RWD08]
@@ -622,7 +622,7 @@ G_M54838_IG09: ; bbWeight=1, epilog, nogc, extend
vmovaps xmm10, xmmword ptr [rsp+0x30]
add rsp, 136
ret - ;; size=38 bbWeight=1 PerfScore 21.25+ ;; size=39 bbWeight=1 PerfScore 21.25
RWD00 dq	3FEDB8A420DC189Ah	; 0.9287892
RWD08 dq	4070E8C71B478423h	; 270.54861
RWD16 dq	400921FB54442D18h	; 3.14159265
@@ -767,7 +767,7 @@ RWD1120	dq	40F5FD9C72B020C5h	; 90073.778
RWD1128	dq	4062433333333333h	; 146.1
-; Total bytes of code 2372, prolog size 37, PerfScore 1766.08, instruction count 388, allocated bytes for code 2372 (MethodHash=72e629c9) for method System.Globalization.CalendricalCalculationsHelper:SumLongSequenceOfPeriodicTerms(double):double (Instrumented Tier1)+; Total bytes of code 2374, prolog size 38, PerfScore 1766.08, instruction count 388, allocated bytes for code 2374 (MethodHash=72e629c9) for method System.Globalization.CalendricalCalculationsHelper:SumLongSequenceOfPeriodicTerms(double):double (Instrumented Tier1)
; ============================================================
Unwind Info:
@@ -775,20 +775,20 @@ Unwind Info:
>> End offset : 0xd1ffab1e (not in unwind data)
Version : 1
Flags : 0x00
- SizeOfProlog : 0x25+ SizeOfProlog : 0x26
CountOfUnwindCodes: 12
FrameRegister : none (0)
FrameOffset : N/A (no FrameRegister) (Value=0)
UnwindCodes :
- CodeOffset: 0x25 UnwindOp: UWOP_SAVE_XMM128 (8) OpInfo: XMM10 (10)+ CodeOffset: 0x26 UnwindOp: UWOP_SAVE_XMM128 (8) OpInfo: XMM10 (10)
Scaled Small Offset: 3 * 16 = 48 = 0x00030
- CodeOffset: 0x1F UnwindOp: UWOP_SAVE_XMM128 (8) OpInfo: XMM9 (9)+ CodeOffset: 0x20 UnwindOp: UWOP_SAVE_XMM128 (8) OpInfo: XMM9 (9)
Scaled Small Offset: 4 * 16 = 64 = 0x00040
- CodeOffset: 0x19 UnwindOp: UWOP_SAVE_XMM128 (8) OpInfo: XMM8 (8)+ CodeOffset: 0x1A UnwindOp: UWOP_SAVE_XMM128 (8) OpInfo: XMM8 (8)
Scaled Small Offset: 5 * 16 = 80 = 0x00050
- CodeOffset: 0x13 UnwindOp: UWOP_SAVE_XMM128 (8) OpInfo: XMM7 (7)+ CodeOffset: 0x14 UnwindOp: UWOP_SAVE_XMM128 (8) OpInfo: XMM7 (7)
Scaled Small Offset: 6 * 16 = 96 = 0x00060
- CodeOffset: 0x0D UnwindOp: UWOP_SAVE_XMM128 (8) OpInfo: XMM6 (6)+ CodeOffset: 0x0E UnwindOp: UWOP_SAVE_XMM128 (8) OpInfo: XMM6 (6)
Scaled Small Offset: 7 * 16 = 112 = 0x00070
- CodeOffset: 0x07 UnwindOp: UWOP_ALLOC_LARGE (1) OpInfo: 0 - Scaled small + CodeOffset: 0x08 UnwindOp: UWOP_ALLOC_LARGE (1) OpInfo: 0 - Scaled small 
Size: 17 * 8 = 136 = 0x00088
+1 (+0.23%) : 497356.dasm - System.Runtime.Intrinsics.VectorMath:LogSingle[System.Runtime.Intrinsics.Vector256`1[float],System.Runtime.Intrinsics.Vector256`1[int],System.Runtime.Intrinsics.Vector256`1[uint]](System.Runtime.Intrinsics.Vector256`1[float]):System.Runtime.Intrinsics.Vector256`1[float] (Tier1)
@@ -138,7 +138,7 @@ G_M36528_IG04: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0006 {rcx rdx},
vmovups ymmword ptr [rcx], ymm2
mov rax, rcx
; byrRegs +[rax]
- ;; size=243 bbWeight=1 PerfScore 117.75+ ;; size=244 bbWeight=1 PerfScore 117.75
G_M36528_IG05: ; bbWeight=1, epilog, nogc, extend
vzeroupper ret @@ -167,7 +167,7 @@ RWD232 dd	BF000002h ; -0.5
RWD236 dd	3F317218h ; 0.693147
-; Total bytes of code 432, prolog size 0, PerfScore 173.03, instruction count 68, allocated bytes for code 432 (MethodHash=421e714f) for method System.Runtime.Intrinsics.VectorMath:LogSingle[System.Runtime.Intrinsics.Vector256`1[float],System.Runtime.Intrinsics.Vector256`1[int],System.Runtime.Intrinsics.Vector256`1[uint]](System.Runtime.Intrinsics.Vector256`1[float]):System.Runtime.Intrinsics.Vector256`1[float] (Tier1)+; Total bytes of code 433, prolog size 0, PerfScore 173.03, instruction count 68, allocated bytes for code 433 (MethodHash=421e714f) for method System.Runtime.Intrinsics.VectorMath:LogSingle[System.Runtime.Intrinsics.Vector256`1[float],System.Runtime.Intrinsics.Vector256`1[int],System.Runtime.Intrinsics.Vector256`1[uint]](System.Runtime.Intrinsics.Vector256`1[float]):System.Runtime.Intrinsics.Vector256`1[float] (Tier1)
; ============================================================
Unwind Info:
+1 (+0.23%) : 484509.dasm - System.Numerics.Tensors.TensorPrimitives+LogOperatorSingle:Invoke(System.Runtime.Intrinsics.Vector256`1[float]):System.Runtime.Intrinsics.Vector256`1[float] (Tier1)
@@ -108,7 +108,7 @@ G_M39372_IG04: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0006 {rcx rdx},
vmovups ymmword ptr [rcx], ymm2
mov rax, rcx
; byrRegs +[rax]
- ;; size=243 bbWeight=1 PerfScore 117.75+ ;; size=244 bbWeight=1 PerfScore 117.75
G_M39372_IG05: ; bbWeight=1, epilog, nogc, extend
vzeroupper ret @@ -137,7 +137,7 @@ RWD232 dd	BF000002h ; -0.5
RWD236 dd	3F317218h ; 0.693147
-; Total bytes of code 432, prolog size 0, PerfScore 161.72, instruction count 68, allocated bytes for code 432 (MethodHash=a2e86633) for method System.Numerics.Tensors.TensorPrimitives+LogOperatorSingle:Invoke(System.Runtime.Intrinsics.Vector256`1[float]):System.Runtime.Intrinsics.Vector256`1[float] (Tier1)+; Total bytes of code 433, prolog size 0, PerfScore 161.72, instruction count 68, allocated bytes for code 433 (MethodHash=a2e86633) for method System.Numerics.Tensors.TensorPrimitives+LogOperatorSingle:Invoke(System.Runtime.Intrinsics.Vector256`1[float]):System.Runtime.Intrinsics.Vector256`1[float] (Tier1)
; ============================================================
Unwind Info:
+7 (+87.50%) : 344948.dasm - ManagedTests.DynamicCSharp.Conformance.dynamic.declarations.backwardscompatible.dynamictypedeclared017.dynamictypedeclared017.A:MainMethod():int (Tier0)
@@ -12,16 +12,16 @@
G_M39926_IG01: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref, nogc <-- Prolog IG
push rbp
mov rbp, rsp
- ;; size=4 bbWeight=1 PerfScore 1.25+ ;; size=7 bbWeight=1 PerfScore 1.25
G_M39926_IG02: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref
xor eax, eax
- ;; size=2 bbWeight=1 PerfScore 0.25+ ;; size=4 bbWeight=1 PerfScore 0.25
G_M39926_IG03: ; bbWeight=1, epilog, nogc, extend
pop rbp
ret - ;; size=2 bbWeight=1 PerfScore 1.50+ ;; size=4 bbWeight=1 PerfScore 1.50-; Total bytes of code 8, prolog size 4, PerfScore 3.00, instruction count 5, allocated bytes for code 8 (MethodHash=d45d6409) for method ManagedTests.DynamicCSharp.Conformance.dynamic.declarations.backwardscompatible.dynamictypedeclared017.dynamictypedeclared017.A:MainMethod():int (Tier0)+; Total bytes of code 15, prolog size 7, PerfScore 3.00, instruction count 5, allocated bytes for code 15 (MethodHash=d45d6409) for method ManagedTests.DynamicCSharp.Conformance.dynamic.declarations.backwardscompatible.dynamictypedeclared017.dynamictypedeclared017.A:MainMethod():int (Tier0)
; ============================================================
Unwind Info:
@@ -29,9 +29,9 @@ Unwind Info:
>> End offset : 0xd1ffab1e (not in unwind data)
Version : 1
Flags : 0x00
- SizeOfProlog : 0x01+ SizeOfProlog : 0x03
CountOfUnwindCodes: 1
FrameRegister : none (0)
FrameOffset : N/A (no FrameRegister) (Value=0)
UnwindCodes :
- CodeOffset: 0x01 UnwindOp: UWOP_PUSH_NONVOL (0) OpInfo: rbp (5)+ CodeOffset: 0x03 UnwindOp: UWOP_PUSH_NONVOL (0) OpInfo: rbp (5)
+7 (+87.50%) : 561668.dasm - System.Runtime.Intrinsics.Vector128`1[System.Int128]:get_IsSupported():ubyte (Tier0)
@@ -12,16 +12,16 @@
G_M6228_IG01: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref, nogc <-- Prolog IG
push rbp
mov rbp, rsp
- ;; size=4 bbWeight=1 PerfScore 1.25+ ;; size=7 bbWeight=1 PerfScore 1.25
G_M6228_IG02: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref
xor eax, eax
- ;; size=2 bbWeight=1 PerfScore 0.25+ ;; size=4 bbWeight=1 PerfScore 0.25
G_M6228_IG03: ; bbWeight=1, epilog, nogc, extend
pop rbp
ret - ;; size=2 bbWeight=1 PerfScore 1.50+ ;; size=4 bbWeight=1 PerfScore 1.50-; Total bytes of code 8, prolog size 4, PerfScore 3.00, instruction count 5, allocated bytes for code 8 (MethodHash=9c00e7ab) for method System.Runtime.Intrinsics.Vector128`1[System.Int128]:get_IsSupported():ubyte (Tier0)+; Total bytes of code 15, prolog size 7, PerfScore 3.00, instruction count 5, allocated bytes for code 15 (MethodHash=9c00e7ab) for method System.Runtime.Intrinsics.Vector128`1[System.Int128]:get_IsSupported():ubyte (Tier0)
; ============================================================
Unwind Info:
@@ -29,9 +29,9 @@ Unwind Info:
>> End offset : 0xd1ffab1e (not in unwind data)
Version : 1
Flags : 0x00
- SizeOfProlog : 0x01+ SizeOfProlog : 0x03
CountOfUnwindCodes: 1
FrameRegister : none (0)
FrameOffset : N/A (no FrameRegister) (Value=0)
UnwindCodes :
- CodeOffset: 0x01 UnwindOp: UWOP_PUSH_NONVOL (0) OpInfo: rbp (5)+ CodeOffset: 0x03 UnwindOp: UWOP_PUSH_NONVOL (0) OpInfo: rbp (5)
+7 (+87.50%) : 259812.dasm - System.Runtime.CompilerServices.RuntimeHelpers:IsBitwiseEquatable[System.Collections.Frozen.Tests.SimpleNonComparableStruct]():ubyte (Tier0)
@@ -12,16 +12,16 @@
G_M54291_IG01: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref, nogc <-- Prolog IG
push rbp
mov rbp, rsp
- ;; size=4 bbWeight=1 PerfScore 1.25+ ;; size=7 bbWeight=1 PerfScore 1.25
G_M54291_IG02: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref
xor eax, eax
- ;; size=2 bbWeight=1 PerfScore 0.25+ ;; size=4 bbWeight=1 PerfScore 0.25
G_M54291_IG03: ; bbWeight=1, epilog, nogc, extend
pop rbp
ret - ;; size=2 bbWeight=1 PerfScore 1.50+ ;; size=4 bbWeight=1 PerfScore 1.50-; Total bytes of code 8, prolog size 4, PerfScore 3.00, instruction count 5, allocated bytes for code 8 (MethodHash=c4122bec) for method System.Runtime.CompilerServices.RuntimeHelpers:IsBitwiseEquatable[System.Collections.Frozen.Tests.SimpleNonComparableStruct]():ubyte (Tier0)+; Total bytes of code 15, prolog size 7, PerfScore 3.00, instruction count 5, allocated bytes for code 15 (MethodHash=c4122bec) for method System.Runtime.CompilerServices.RuntimeHelpers:IsBitwiseEquatable[System.Collections.Frozen.Tests.SimpleNonComparableStruct]():ubyte (Tier0)
; ============================================================
Unwind Info:
@@ -29,9 +29,9 @@ Unwind Info:
>> End offset : 0xd1ffab1e (not in unwind data)
Version : 1
Flags : 0x00
- SizeOfProlog : 0x01+ SizeOfProlog : 0x03
CountOfUnwindCodes: 1
FrameRegister : none (0)
FrameOffset : N/A (no FrameRegister) (Value=0)
UnwindCodes :
- CodeOffset: 0x01 UnwindOp: UWOP_PUSH_NONVOL (0) OpInfo: rbp (5)+ CodeOffset: 0x03 UnwindOp: UWOP_PUSH_NONVOL (0) OpInfo: rbp (5)
libraries_tests_no_tiered_compilation.run.windows.x64.Release.mch
+1 (+0.14%) : 179042.dasm - System.Numerics.Matrix4x4+Impl:Transform(byref,byref):System.Numerics.Matrix4x4+Impl (FullOpts)
@@ -199,14 +199,14 @@ G_M8955_IG03: ; bbWeight=1, extend
vmovups xmmword ptr [rcx+0x30], xmm0
mov rax, rcx
; byrRegs +[rax]
- ;; size=354 bbWeight=1 PerfScore 161.25+ ;; size=355 bbWeight=1 PerfScore 161.25
G_M8955_IG04: ; bbWeight=1, epilog, nogc, extend
ret ;; size=1 bbWeight=1 PerfScore 1.00
RWD00 dd	3F800000h ; 1
-; Total bytes of code 693, prolog size 0, PerfScore 345.25, instruction count 118, allocated bytes for code 693 (MethodHash=5906dd04) for method System.Numerics.Matrix4x4+Impl:Transform(byref,byref):System.Numerics.Matrix4x4+Impl (FullOpts)+; Total bytes of code 694, prolog size 0, PerfScore 345.25, instruction count 118, allocated bytes for code 694 (MethodHash=5906dd04) for method System.Numerics.Matrix4x4+Impl:Transform(byref,byref):System.Numerics.Matrix4x4+Impl (FullOpts)
; ============================================================
Unwind Info:
+1 (+0.15%) : 212365.dasm - System.Runtime.Intrinsics.VectorMath:LogDouble[System.Runtime.Intrinsics.Vector512`1[double],System.Runtime.Intrinsics.Vector512`1[long],System.Runtime.Intrinsics.Vector512`1[ulong]](System.Runtime.Intrinsics.Vector512`1[double]):System.Runtime.Intrinsics.Vector512`1[double] (FullOpts)
@@ -161,7 +161,7 @@ G_M22781_IG04: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0006 {rcx rdx},
vmovups zmmword ptr [rcx], zmm2
mov rax, rcx
; byrRegs +[rax]
- ;; size=437 bbWeight=1 PerfScore 201.42+ ;; size=438 bbWeight=1 PerfScore 201.42
G_M22781_IG05: ; bbWeight=1, epilog, nogc, extend
vzeroupper ret @@ -203,7 +203,7 @@ RWD488 dq	BF2BD0105C610CA8h	; -0.00021219444
RWD496 dq	3FE6300000000000h	; 0.693359375
-; Total bytes of code 665, prolog size 0, PerfScore 237.75, instruction count 91, allocated bytes for code 667 (MethodHash=3c50a702) for method System.Runtime.Intrinsics.VectorMath:LogDouble[System.Runtime.Intrinsics.Vector512`1[double],System.Runtime.Intrinsics.Vector512`1[long],System.Runtime.Intrinsics.Vector512`1[ulong]](System.Runtime.Intrinsics.Vector512`1[double]):System.Runtime.Intrinsics.Vector512`1[double] (FullOpts)+; Total bytes of code 666, prolog size 0, PerfScore 237.75, instruction count 91, allocated bytes for code 668 (MethodHash=3c50a702) for method System.Runtime.Intrinsics.VectorMath:LogDouble[System.Runtime.Intrinsics.Vector512`1[double],System.Runtime.Intrinsics.Vector512`1[long],System.Runtime.Intrinsics.Vector512`1[ulong]](System.Runtime.Intrinsics.Vector512`1[double]):System.Runtime.Intrinsics.Vector512`1[double] (FullOpts)
; ============================================================
Unwind Info:
+1 (+0.15%) : 211948.dasm - System.Runtime.Intrinsics.VectorMath:Log2Double[System.Runtime.Intrinsics.Vector512`1[double],System.Runtime.Intrinsics.Vector512`1[long],System.Runtime.Intrinsics.Vector512`1[ulong]](System.Runtime.Intrinsics.Vector512`1[double]):System.Runtime.Intrinsics.Vector512`1[double] (FullOpts)
@@ -161,7 +161,7 @@ G_M63855_IG04: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0006 {rcx rdx},
vmovups zmmword ptr [rcx], zmm2
mov rax, rcx
; byrRegs +[rax]
- ;; size=437 bbWeight=1 PerfScore 201.42+ ;; size=438 bbWeight=1 PerfScore 201.42
G_M63855_IG05: ; bbWeight=1, epilog, nogc, extend
vzeroupper ret @@ -203,7 +203,7 @@ RWD488 dq	3ECB295C17F0BBBEh	; 3.23791045e-06
RWD496 dq	3FF7154400000000h	; 1.4426918
-; Total bytes of code 665, prolog size 0, PerfScore 237.75, instruction count 91, allocated bytes for code 667 (MethodHash=4d7b0690) for method System.Runtime.Intrinsics.VectorMath:Log2Double[System.Runtime.Intrinsics.Vector512`1[double],System.Runtime.Intrinsics.Vector512`1[long],System.Runtime.Intrinsics.Vector512`1[ulong]](System.Runtime.Intrinsics.Vector512`1[double]):System.Runtime.Intrinsics.Vector512`1[double] (FullOpts)+; Total bytes of code 666, prolog size 0, PerfScore 237.75, instruction count 91, allocated bytes for code 668 (MethodHash=4d7b0690) for method System.Runtime.Intrinsics.VectorMath:Log2Double[System.Runtime.Intrinsics.Vector512`1[double],System.Runtime.Intrinsics.Vector512`1[long],System.Runtime.Intrinsics.Vector512`1[ulong]](System.Runtime.Intrinsics.Vector512`1[double]):System.Runtime.Intrinsics.Vector512`1[double] (FullOpts)
; ============================================================
Unwind Info:
+4 (+80.00%) : 18935.dasm - LibraryImportGenerator.IntegrationTests.FunctionPointerTests:g__Callback|2_0(int,int):int (FullOpts)
@@ -18,12 +18,12 @@ G_M41111_IG01: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref,
G_M41111_IG02: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref
mov eax, ecx
sub eax, edx
- ;; size=4 bbWeight=1 PerfScore 0.50+ ;; size=8 bbWeight=1 PerfScore 0.50
G_M41111_IG03: ; bbWeight=1, epilog, nogc, extend
ret ;; size=1 bbWeight=1 PerfScore 1.00
-; Total bytes of code 5, prolog size 0, PerfScore 1.50, instruction count 3, allocated bytes for code 5 (MethodHash=f1415f68) for method LibraryImportGenerator.IntegrationTests.FunctionPointerTests:<CalledWithArgumentsInOrder>g__Callback|2_0(int,int):int (FullOpts)+; Total bytes of code 9, prolog size 0, PerfScore 1.50, instruction count 3, allocated bytes for code 9 (MethodHash=f1415f68) for method LibraryImportGenerator.IntegrationTests.FunctionPointerTests:<CalledWithArgumentsInOrder>g__Callback|2_0(int,int):int (FullOpts)
; ============================================================
Unwind Info:
+14 (+82.35%) : 146650.dasm - System.Linq.Parallel.Tests.JoinTests+LeftOrderingCollisionTest+<>c:b__0_0(int):int:this (FullOpts)
@@ -23,12 +23,12 @@ G_M5520_IG02: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref
mov ecx, edx
sub ecx, eax
mov eax, ecx
- ;; size=16 bbWeight=1 PerfScore 2.00+ ;; size=30 bbWeight=1 PerfScore 2.00
G_M5520_IG03: ; bbWeight=1, epilog, nogc, extend
ret ;; size=1 bbWeight=1 PerfScore 1.00
-; Total bytes of code 17, prolog size 0, PerfScore 3.00, instruction count 8, allocated bytes for code 17 (MethodHash=3061ea6f) for method System.Linq.Parallel.Tests.JoinTests+LeftOrderingCollisionTest+<>c:<ReorderLeft>b__0_0(int):int:this (FullOpts)+; Total bytes of code 31, prolog size 0, PerfScore 3.00, instruction count 8, allocated bytes for code 31 (MethodHash=3061ea6f) for method System.Linq.Parallel.Tests.JoinTests+LeftOrderingCollisionTest+<>c:<ReorderLeft>b__0_0(int):int:this (FullOpts)
; ============================================================
Unwind Info:
+6 (+85.71%) : 176318.dasm - System.Numerics.Tensors.TensorPrimitives+InvertedBinaryOperator`2[System.Numerics.Tensors.TensorPrimitives+DivideOperator`1[uint],uint]:Invoke(uint,uint):uint (FullOpts)
@@ -20,12 +20,12 @@ G_M23137_IG02: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref
mov eax, edx
xor edx, edx
div edx:eax, ecx
- ;; size=6 bbWeight=1 PerfScore 25.50+ ;; size=12 bbWeight=1 PerfScore 25.50
G_M23137_IG03: ; bbWeight=1, epilog, nogc, extend
ret ;; size=1 bbWeight=1 PerfScore 1.00
-; Total bytes of code 7, prolog size 0, PerfScore 26.50, instruction count 4, allocated bytes for code 7 (MethodHash=86f0a59e) for method System.Numerics.Tensors.TensorPrimitives+InvertedBinaryOperator`2[System.Numerics.Tensors.TensorPrimitives+DivideOperator`1[uint],uint]:Invoke(uint,uint):uint (FullOpts)+; Total bytes of code 13, prolog size 0, PerfScore 26.50, instruction count 4, allocated bytes for code 13 (MethodHash=86f0a59e) for method System.Numerics.Tensors.TensorPrimitives+InvertedBinaryOperator`2[System.Numerics.Tensors.TensorPrimitives+DivideOperator`1[uint],uint]:Invoke(uint,uint):uint (FullOpts)
; ============================================================
Unwind Info:
realworld.run.windows.x64.checked.mch
+1 (+0.11%) : 1372.dasm - BepuPhysics.Collidables.MeshInertiaHelper:ComputeTetrahedronContribution(byref,byref,byref,float,byref) (FullOpts)
@@ -56,7 +56,7 @@
G_M56806_IG01: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref, nogc <-- Prolog IG
mov rax, bword ptr [rsp+0x28]
; byrRegs +[rax]
- ;; size=5 bbWeight=1 PerfScore 1.00+ ;; size=6 bbWeight=1 PerfScore 1.00
G_M56806_IG02: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0107 {rax rcx rdx r8}, byref
; byrRegs +[rcx rdx r8]
vmulss xmm0, xmm3, dword ptr [reloc @RWD00]
@@ -236,7 +236,7 @@ RWD08 dd	C0000000h ; -2
RWD12 dd	40000000h ; 2
-; Total bytes of code 870, prolog size 0, PerfScore 531.00, instruction count 165, allocated bytes for code 870 (MethodHash=1b132219) for method BepuPhysics.Collidables.MeshInertiaHelper:ComputeTetrahedronContribution(byref,byref,byref,float,byref) (FullOpts)+; Total bytes of code 871, prolog size 0, PerfScore 531.00, instruction count 165, allocated bytes for code 871 (MethodHash=1b132219) for method BepuPhysics.Collidables.MeshInertiaHelper:ComputeTetrahedronContribution(byref,byref,byref,float,byref) (FullOpts)
; ============================================================
Unwind Info:
+1 (+0.21%) : 1203.dasm - BepuPhysics.Collidables.Compound:GetRotatedChildPose(byref,byref,byref,byref) (FullOpts)
@@ -100,7 +100,7 @@ G_M10677_IG02: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0306 {rcx rdx r
vmulps ymm17, ymm3, ymm0
vmulps ymm0, ymm5, ymm0
vmulps ymm18, ymm3, ymm2
- ;; size=275 bbWeight=1 PerfScore 256.50+ ;; size=276 bbWeight=1 PerfScore 256.50
G_M10677_IG03: ; bbWeight=1, extend
vmulps ymm2, ymm5, ymm2
vmulps ymm4, ymm5, ymm4
@@ -146,7 +146,7 @@ G_M10677_IG04: ; bbWeight=1, epilog, nogc, extend
RWD00 dq	3F8000003F800000h, 3F8000003F800000h, 3F8000003F800000h, 3F8000003F800000h
-; Total bytes of code 476, prolog size 0, PerfScore 393.50, instruction count 97, allocated bytes for code 476 (MethodHash=d5c8d64a) for method BepuPhysics.Collidables.Compound:GetRotatedChildPose(byref,byref,byref,byref) (FullOpts)+; Total bytes of code 477, prolog size 0, PerfScore 393.50, instruction count 97, allocated bytes for code 477 (MethodHash=d5c8d64a) for method BepuPhysics.Collidables.Compound:GetRotatedChildPose(byref,byref,byref,byref) (FullOpts)
; ============================================================
Unwind Info:
+5 (+0.36%) : 1326.dasm - BepuUtilities.Symmetric6x6Wide:LDLTSolve(byref,byref,byref,byref,byref,byref,byref) (FullOpts)
@@ -59,7 +59,7 @@ G_M52182_IG01: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref,
; byrRegs +[rax]
mov r10, bword ptr [rsp+0x90]
; byrRegs +[r10]
- ;; size=57 bbWeight=1 PerfScore 13.25+ ;; size=61 bbWeight=1 PerfScore 13.25
G_M52182_IG02: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0F07 {rax rcx rdx r8 r9 r10 r11}, byref
; byrRegs +[rcx rdx r8-r9]
vmovups ymm0, ymmword ptr [r8]
@@ -305,11 +305,11 @@ G_M52182_IG06: ; bbWeight=1, epilog, nogc, extend
vmovaps xmm10, xmmword ptr [rsp]
add rsp, 88
ret - ;; size=37 bbWeight=1 PerfScore 22.25+ ;; size=38 bbWeight=1 PerfScore 22.25
RWD00 dq	3F8000003F800000h, 3F8000003F800000h, 3F8000003F800000h, 3F8000003F800000h
-; Total bytes of code 1406, prolog size 33, PerfScore 914.50, instruction count 244, allocated bytes for code 1406 (MethodHash=9c303429) for method BepuUtilities.Symmetric6x6Wide:LDLTSolve(byref,byref,byref,byref,byref,byref,byref) (FullOpts)+; Total bytes of code 1411, prolog size 34, PerfScore 914.50, instruction count 244, allocated bytes for code 1411 (MethodHash=9c303429) for method BepuUtilities.Symmetric6x6Wide:LDLTSolve(byref,byref,byref,byref,byref,byref,byref) (FullOpts)
; ============================================================
Unwind Info:
@@ -317,19 +317,19 @@ Unwind Info:
>> End offset : 0xd1ffab1e (not in unwind data)
Version : 1
Flags : 0x00
- SizeOfProlog : 0x21+ SizeOfProlog : 0x22
CountOfUnwindCodes: 11
FrameRegister : none (0)
FrameOffset : N/A (no FrameRegister) (Value=0)
UnwindCodes :
- CodeOffset: 0x21 UnwindOp: UWOP_SAVE_XMM128 (8) OpInfo: XMM10 (10)+ CodeOffset: 0x22 UnwindOp: UWOP_SAVE_XMM128 (8) OpInfo: XMM10 (10)
Scaled Small Offset: 0 * 16 = 0 = 0x00000
- CodeOffset: 0x1C UnwindOp: UWOP_SAVE_XMM128 (8) OpInfo: XMM9 (9)+ CodeOffset: 0x1D UnwindOp: UWOP_SAVE_XMM128 (8) OpInfo: XMM9 (9)
Scaled Small Offset: 1 * 16 = 16 = 0x00010
- CodeOffset: 0x16 UnwindOp: UWOP_SAVE_XMM128 (8) OpInfo: XMM8 (8)+ CodeOffset: 0x17 UnwindOp: UWOP_SAVE_XMM128 (8) OpInfo: XMM8 (8)
Scaled Small Offset: 2 * 16 = 32 = 0x00020
- CodeOffset: 0x10 UnwindOp: UWOP_SAVE_XMM128 (8) OpInfo: XMM7 (7)+ CodeOffset: 0x11 UnwindOp: UWOP_SAVE_XMM128 (8) OpInfo: XMM7 (7)
Scaled Small Offset: 3 * 16 = 48 = 0x00030
- CodeOffset: 0x0A UnwindOp: UWOP_SAVE_XMM128 (8) OpInfo: XMM6 (6)+ CodeOffset: 0x0B UnwindOp: UWOP_SAVE_XMM128 (8) OpInfo: XMM6 (6)
Scaled Small Offset: 4 * 16 = 64 = 0x00040
- CodeOffset: 0x04 UnwindOp: UWOP_ALLOC_SMALL (2) OpInfo: 10 * 8 + 8 = 88 = 0x58+ CodeOffset: 0x05 UnwindOp: UWOP_ALLOC_SMALL (2) OpInfo: 10 * 8 + 8 = 88 = 0x58
+6 (+75.00%) : 8466.dasm - Microsoft.ML.Data.VectorDataViewType+<>c:<.ctor>b__4_0(int):ubyte:this (FullOpts)
@@ -19,12 +19,12 @@ G_M20425_IG02: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref
mov eax, edx
not eax
shr eax, 31
- ;; size=7 bbWeight=1 PerfScore 1.00+ ;; size=13 bbWeight=1 PerfScore 1.00
G_M20425_IG03: ; bbWeight=1, epilog, nogc, extend
ret ;; size=1 bbWeight=1 PerfScore 1.00
-; Total bytes of code 8, prolog size 0, PerfScore 2.00, instruction count 4, allocated bytes for code 8 (MethodHash=067ab036) for method Microsoft.ML.Data.VectorDataViewType+<>c:<.ctor>b__4_0(int):ubyte:this (FullOpts)+; Total bytes of code 14, prolog size 0, PerfScore 2.00, instruction count 4, allocated bytes for code 14 (MethodHash=067ab036) for method Microsoft.ML.Data.VectorDataViewType+<>c:<.ctor>b__4_0(int):ubyte:this (FullOpts)
; ============================================================
Unwind Info:
+10 (+76.92%) : 20363.dasm - Microsoft.CodeAnalysis.CSharp.NullableWalker+LocalState:get_Capacity():int:this (FullOpts)
@@ -25,12 +25,12 @@ G_M6705_IG02: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0002 {rcx}, byre
shr ecx, 31
add eax, ecx
sar eax, 1
- ;; size=12 bbWeight=1 PerfScore 3.50+ ;; size=22 bbWeight=1 PerfScore 3.50
G_M6705_IG03: ; bbWeight=1, epilog, nogc, extend
ret ;; size=1 bbWeight=1 PerfScore 1.00
-; Total bytes of code 13, prolog size 0, PerfScore 4.50, instruction count 6, allocated bytes for code 13 (MethodHash=ab86e5ce) for method Microsoft.CodeAnalysis.CSharp.NullableWalker+LocalState:get_Capacity():int:this (FullOpts)+; Total bytes of code 23, prolog size 0, PerfScore 4.50, instruction count 6, allocated bytes for code 23 (MethodHash=ab86e5ce) for method Microsoft.CodeAnalysis.CSharp.NullableWalker+LocalState:get_Capacity():int:this (FullOpts)
; ============================================================
Unwind Info:
+4 (+80.00%) : 17010.dasm - Microsoft.CodeAnalysis.CSharp.Symbols.ErrorTypeSymbol:HasInlineArrayAttribute(byref):ubyte:this (FullOpts)
@@ -19,13 +19,13 @@ G_M52199_IG02: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0004 {rdx}, byr
; byrRegs +[rdx]
xor eax, eax
mov dword ptr [rdx], eax
- ;; size=4 bbWeight=1 PerfScore 1.25+ ;; size=8 bbWeight=1 PerfScore 1.25
G_M52199_IG03: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref, epilog, nogc
; byrRegs -[rdx]
ret ;; size=1 bbWeight=1 PerfScore 1.00
-; Total bytes of code 5, prolog size 0, PerfScore 2.25, instruction count 3, allocated bytes for code 5 (MethodHash=79e23418) for method Microsoft.CodeAnalysis.CSharp.Symbols.ErrorTypeSymbol:HasInlineArrayAttribute(byref):ubyte:this (FullOpts)+; Total bytes of code 9, prolog size 0, PerfScore 2.25, instruction count 3, allocated bytes for code 9 (MethodHash=79e23418) for method Microsoft.CodeAnalysis.CSharp.Symbols.ErrorTypeSymbol:HasInlineArrayAttribute(byref):ubyte:this (FullOpts)
; ============================================================
Unwind Info:
smoke_tests.nativeaot.windows.x64.checked.mch
-1 (-16.67%) : 14296.dasm - Program:X86SerializeX64IsSupported():ubyte (FullOpts)
@@ -14,13 +14,13 @@
G_M13406_IG01: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref, nogc <-- Prolog IG
;; size=0 bbWeight=1 PerfScore 0.00
G_M13406_IG02: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref
- mov eax, 1- ;; size=5 bbWeight=1 PerfScore 0.25+ xor eax, eax+ ;; size=4 bbWeight=1 PerfScore 0.25
G_M13406_IG03: ; bbWeight=1, epilog, nogc, extend
ret ;; size=1 bbWeight=1 PerfScore 1.00
-; Total bytes of code 6, prolog size 0, PerfScore 1.25, instruction count 2, allocated bytes for code 6 (MethodHash=00a5cba1) for method Program:X86SerializeX64IsSupported():ubyte (FullOpts)+; Total bytes of code 5, prolog size 0, PerfScore 1.25, instruction count 2, allocated bytes for code 5 (MethodHash=00a5cba1) for method Program:X86SerializeX64IsSupported():ubyte (FullOpts)
; ============================================================
Unwind Info:
-1 (-16.67%) : 21532.dasm - Program:AesX64IsSupported():ubyte (FullOpts)
@@ -14,13 +14,13 @@
G_M55817_IG01: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref, nogc <-- Prolog IG
;; size=0 bbWeight=1 PerfScore 0.00
G_M55817_IG02: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref
- mov eax, 1- ;; size=5 bbWeight=1 PerfScore 0.25+ xor eax, eax+ ;; size=4 bbWeight=1 PerfScore 0.25
G_M55817_IG03: ; bbWeight=1, epilog, nogc, extend
ret ;; size=1 bbWeight=1 PerfScore 1.00
-; Total bytes of code 6, prolog size 0, PerfScore 1.25, instruction count 2, allocated bytes for code 6 (MethodHash=c5da25f6) for method Program:AesX64IsSupported():ubyte (FullOpts)+; Total bytes of code 5, prolog size 0, PerfScore 1.25, instruction count 2, allocated bytes for code 5 (MethodHash=c5da25f6) for method Program:AesX64IsSupported():ubyte (FullOpts)
; ============================================================
Unwind Info:
-1 (-16.67%) : 19229.dasm - Program:X86SerializeX64IsSupported():ubyte (FullOpts)
@@ -14,13 +14,13 @@
G_M13406_IG01: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref, nogc <-- Prolog IG
;; size=0 bbWeight=1 PerfScore 0.00
G_M13406_IG02: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref
- mov eax, 1- ;; size=5 bbWeight=1 PerfScore 0.25+ xor eax, eax+ ;; size=4 bbWeight=1 PerfScore 0.25
G_M13406_IG03: ; bbWeight=1, epilog, nogc, extend
ret ;; size=1 bbWeight=1 PerfScore 1.00
-; Total bytes of code 6, prolog size 0, PerfScore 1.25, instruction count 2, allocated bytes for code 6 (MethodHash=00a5cba1) for method Program:X86SerializeX64IsSupported():ubyte (FullOpts)+; Total bytes of code 5, prolog size 0, PerfScore 1.25, instruction count 2, allocated bytes for code 5 (MethodHash=00a5cba1) for method Program:X86SerializeX64IsSupported():ubyte (FullOpts)
; ============================================================
Unwind Info:
+5 (+166.67%) : 19199.dasm - Program:AvxVnniX64IsSupported():ubyte (FullOpts)
@@ -14,13 +14,13 @@
G_M60430_IG01: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref, nogc <-- Prolog IG
;; size=0 bbWeight=1 PerfScore 0.00
G_M60430_IG02: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref
- xor eax, eax- ;; size=2 bbWeight=1 PerfScore 0.25+ mov eax, 1+ ;; size=7 bbWeight=1 PerfScore 0.25
G_M60430_IG03: ; bbWeight=1, epilog, nogc, extend
ret ;; size=1 bbWeight=1 PerfScore 1.00
-; Total bytes of code 3, prolog size 0, PerfScore 1.25, instruction count 2, allocated bytes for code 3 (MethodHash=e20b13f1) for method Program:AvxVnniX64IsSupported():ubyte (FullOpts)+; Total bytes of code 8, prolog size 0, PerfScore 1.25, instruction count 2, allocated bytes for code 8 (MethodHash=e20b13f1) for method Program:AvxVnniX64IsSupported():ubyte (FullOpts)
; ============================================================
Unwind Info:
+5 (+166.67%) : 21515.dasm - Program:FmaX64IsSupported():ubyte (FullOpts)
@@ -14,13 +14,13 @@
G_M2260_IG01: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref, nogc <-- Prolog IG
;; size=0 bbWeight=1 PerfScore 0.00
G_M2260_IG02: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref
- xor eax, eax- ;; size=2 bbWeight=1 PerfScore 0.25+ mov eax, 1+ ;; size=7 bbWeight=1 PerfScore 0.25
G_M2260_IG03: ; bbWeight=1, epilog, nogc, extend
ret ;; size=1 bbWeight=1 PerfScore 1.00
-; Total bytes of code 3, prolog size 0, PerfScore 1.25, instruction count 2, allocated bytes for code 3 (MethodHash=36a7f72b) for method Program:FmaX64IsSupported():ubyte (FullOpts)+; Total bytes of code 8, prolog size 0, PerfScore 1.25, instruction count 2, allocated bytes for code 8 (MethodHash=36a7f72b) for method Program:FmaX64IsSupported():ubyte (FullOpts)
; ============================================================
Unwind Info:
+5 (+166.67%) : 21526.dasm - Program:Avx2X64IsSupported():ubyte (FullOpts)
@@ -14,13 +14,13 @@
G_M13187_IG01: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref, nogc <-- Prolog IG
;; size=0 bbWeight=1 PerfScore 0.00
G_M13187_IG02: ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref
- xor eax, eax- ;; size=2 bbWeight=1 PerfScore 0.25+ mov eax, 1+ ;; size=7 bbWeight=1 PerfScore 0.25
G_M13187_IG03: ; bbWeight=1, epilog, nogc, extend
ret ;; size=1 bbWeight=1 PerfScore 1.00
-; Total bytes of code 3, prolog size 0, PerfScore 1.25, instruction count 2, allocated bytes for code 3 (MethodHash=f683cc7c) for method Program:Avx2X64IsSupported():ubyte (FullOpts)+; Total bytes of code 8, prolog size 0, PerfScore 1.25, instruction count 2, allocated bytes for code 8 (MethodHash=f683cc7c) for method Program:Avx2X64IsSupported():ubyte (FullOpts)
; ============================================================
Unwind Info:
Details

Size improvements/regressions per collection

CollectionContexts with diffsImprovementsRegressionsSame sizeImprovements (bytes)Regressions (bytes)
aspnet.run.windows.x64.checked.mch140,5270140,5270-0+10,392,179
benchmarks.run.windows.x64.checked.mch37,922037,9220-0+3,013,399
benchmarks.run_pgo.windows.x64.checked.mch120,0200120,0200-0+8,962,474
benchmarks.run_tiered.windows.x64.checked.mch76,575076,5750-0+4,199,746
coreclr_tests.run.windows.x64.checked.mch639,8903639,8870-14+84,314,025
libraries.crossgen2.windows.x64.checked.mch274,8480274,8480-0+11,739,139
libraries.pmi.windows.x64.checked.mch309,1491309,1480-1+15,125,726
libraries_tests.run.windows.x64.Release.mch811,9140811,9140-0+69,356,394
libraries_tests_no_tiered_compilation.run.windows.x64.Release.mch339,2840339,2840-0+32,521,941
realworld.run.windows.x64.checked.mch28,087028,0870-0+2,545,126
smoke_tests.nativeaot.windows.x64.checked.mch30,547830,5390-8+1,394,449
2,808,763122,808,7510-23+243,564,598

PerfScore improvements/regressions per collection

CollectionContexts with diffsImprovementsRegressionsSame PerfScoreImprovements (PerfScore)Regressions (PerfScore)PerfScore Overall in FullOpts
aspnet.run.windows.x64.checked.mch140,52700140,5270.00%0.00%0.0000%
benchmarks.run.windows.x64.checked.mch37,9220037,9220.00%0.00%0.0000%
benchmarks.run_pgo.windows.x64.checked.mch120,02000120,0200.00%0.00%0.0000%
benchmarks.run_tiered.windows.x64.checked.mch76,5750076,5750.00%0.00%0.0000%
coreclr_tests.run.windows.x64.checked.mch639,89021639,887-7.87%+0.03%0.0000%
libraries.crossgen2.windows.x64.checked.mch274,84800274,8480.00%0.00%0.0000%
libraries.pmi.windows.x64.checked.mch309,14900309,1490.00%0.00%0.0000%
libraries_tests.run.windows.x64.Release.mch811,91400811,9140.00%0.00%0.0000%
libraries_tests_no_tiered_compilation.run.windows.x64.Release.mch339,28400339,2840.00%0.00%0.0000%
realworld.run.windows.x64.checked.mch28,0870028,0870.00%0.00%0.0000%
smoke_tests.nativeaot.windows.x64.checked.mch30,5470030,5470.00%0.00%0.0000%

Context information

CollectionDiffed contextsMinOptsFullOptsMissed, baseMissed, diff
aspnet.run.windows.x64.checked.mch141,22477,32463,9000 (0.00%)0 (0.00%)
benchmarks.run.windows.x64.checked.mch38,352638,3460 (0.00%)0 (0.00%)
benchmarks.run_pgo.windows.x64.checked.mch120,28068,10352,1770 (0.00%)0 (0.00%)
benchmarks.run_tiered.windows.x64.checked.mch76,87656,35820,5180 (0.00%)0 (0.00%)
coreclr_tests.run.windows.x64.checked.mch642,813393,776249,0370 (0.00%)5 (0.00%)
libraries.crossgen2.windows.x64.checked.mch276,88915276,8740 (0.00%)2 (0.00%)
libraries.pmi.windows.x64.checked.mch316,0106316,0040 (0.00%)1 (0.00%)
libraries_tests.run.windows.x64.Release.mch814,679567,674247,0050 (0.00%)0 (0.00%)
libraries_tests_no_tiered_compilation.run.windows.x64.Release.mch343,89521,994321,9010 (0.00%)0 (0.00%)
realworld.run.windows.x64.checked.mch28,368328,3650 (0.00%)0 (0.00%)
smoke_tests.nativeaot.windows.x64.checked.mch31,2021031,1920 (0.00%)3 (0.01%)
2,830,5881,185,2691,645,3190 (0.00%)11 (0.00%)

jit-analyze output

Comments: No Decode Failure or assertion failure is reported in the logs, only except some assert fails about unsupported ISAs, this should is also attributed to the APX CPUID changes. The huge code size is expected as we are forcing all the compatible legacy instructions to be encoded in REX2 regradless if it is needed.

2.3 TpDiff - REX2 off (no or little tp impact expected)

TP impact with REX2 off compared with base main:

Overall (+0.08% to +0.19%)
CollectionPDIFF
aspnet.run.windows.x64.checked.mch+0.13%
benchmarks.run.windows.x64.checked.mch+0.08%
benchmarks.run_pgo.windows.x64.checked.mch+0.12%
benchmarks.run_tiered.windows.x64.checked.mch+0.19%
coreclr_tests.run.windows.x64.checked.mch+0.18%
libraries.crossgen2.windows.x64.checked.mch+0.11%
libraries.pmi.windows.x64.checked.mch+0.09%
libraries_tests.run.windows.x64.Release.mch+0.15%
libraries_tests_no_tiered_compilation.run.windows.x64.Release.mch+0.10%
realworld.run.windows.x64.checked.mch+0.09%
smoke_tests.nativeaot.windows.x64.checked.mch+0.08%
MinOpts (+0.24% to +0.43%)
CollectionPDIFF
aspnet.run.windows.x64.checked.mch+0.37%
benchmarks.run.windows.x64.checked.mch+0.36%
benchmarks.run_pgo.windows.x64.checked.mch+0.35%
benchmarks.run_tiered.windows.x64.checked.mch+0.34%
coreclr_tests.run.windows.x64.checked.mch+0.27%
libraries.crossgen2.windows.x64.checked.mch+0.36%
libraries.pmi.windows.x64.checked.mch+0.24%
libraries_tests.run.windows.x64.Release.mch+0.37%
libraries_tests_no_tiered_compilation.run.windows.x64.Release.mch+0.30%
realworld.run.windows.x64.checked.mch+0.43%
smoke_tests.nativeaot.windows.x64.checked.mch+0.29%
FullOpts (+0.07% to +0.11%)
CollectionPDIFF
aspnet.run.windows.x64.checked.mch+0.08%
benchmarks.run.windows.x64.checked.mch+0.08%
benchmarks.run_pgo.windows.x64.checked.mch+0.07%
benchmarks.run_tiered.windows.x64.checked.mch+0.08%
coreclr_tests.run.windows.x64.checked.mch+0.10%
libraries.crossgen2.windows.x64.checked.mch+0.11%
libraries.pmi.windows.x64.checked.mch+0.09%
libraries_tests.run.windows.x64.Release.mch+0.08%
libraries_tests_no_tiered_compilation.run.windows.x64.Release.mch+0.10%
realworld.run.windows.x64.checked.mch+0.08%
smoke_tests.nativeaot.windows.x64.checked.mch+0.08%
Details

All contexts:

CollectionBase # instructionsDiff # instructionsPDIFF
aspnet.run.windows.x64.checked.mch142,394,809,547142,582,075,252+0.13%
benchmarks.run.windows.x64.checked.mch55,370,986,62455,417,510,475+0.08%
benchmarks.run_pgo.windows.x64.checked.mch121,883,543,862122,027,057,184+0.12%
benchmarks.run_tiered.windows.x64.checked.mch34,231,112,72434,297,405,288+0.19%
coreclr_tests.run.windows.x64.checked.mch809,468,778,745810,902,734,493+0.18%
libraries.crossgen2.windows.x64.checked.mch154,853,677,569155,028,932,749+0.11%
libraries.pmi.windows.x64.checked.mch269,020,941,364269,270,900,769+0.09%
libraries_tests.run.windows.x64.Release.mch815,708,776,365816,960,864,737+0.15%
libraries_tests_no_tiered_compilation.run.windows.x64.Release.mch577,085,986,658577,668,396,449+0.10%
realworld.run.windows.x64.checked.mch49,400,011,36349,442,097,903+0.09%
smoke_tests.nativeaot.windows.x64.checked.mch22,690,369,63122,708,625,728+0.08%

MinOpts contexts:

CollectionBase # instructionsDiff # instructionsPDIFF
aspnet.run.windows.x64.checked.mch24,056,213,49124,145,743,102+0.37%
benchmarks.run.windows.x64.checked.mch705,633708,145+0.36%
benchmarks.run_pgo.windows.x64.checked.mch19,880,799,80619,950,333,595+0.35%
benchmarks.run_tiered.windows.x64.checked.mch15,022,302,54115,073,432,822+0.34%
coreclr_tests.run.windows.x64.checked.mch347,233,426,241348,186,424,612+0.27%
libraries.crossgen2.windows.x64.checked.mch2,084,9092,092,477+0.36%
libraries.pmi.windows.x64.checked.mch132,525,396132,849,510+0.24%
libraries_tests.run.windows.x64.Release.mch206,423,819,906207,191,499,654+0.37%
libraries_tests_no_tiered_compilation.run.windows.x64.Release.mch12,115,105,17212,152,028,039+0.30%
realworld.run.windows.x64.checked.mch348,063,722349,547,131+0.43%
smoke_tests.nativeaot.windows.x64.checked.mch1,254,1671,257,840+0.29%

FullOpts contexts:

CollectionBase # instructionsDiff # instructionsPDIFF
aspnet.run.windows.x64.checked.mch118,338,596,056118,436,332,150+0.08%
benchmarks.run.windows.x64.checked.mch55,370,280,99155,416,802,330+0.08%
benchmarks.run_pgo.windows.x64.checked.mch102,002,744,056102,076,723,589+0.07%
benchmarks.run_tiered.windows.x64.checked.mch19,208,810,18319,223,972,466+0.08%
coreclr_tests.run.windows.x64.checked.mch462,235,352,504462,716,309,881+0.10%
libraries.crossgen2.windows.x64.checked.mch154,851,592,660155,026,840,272+0.11%
libraries.pmi.windows.x64.checked.mch268,888,415,968269,138,051,259+0.09%
libraries_tests.run.windows.x64.Release.mch609,284,956,459609,769,365,083+0.08%
libraries_tests_no_tiered_compilation.run.windows.x64.Release.mch564,970,881,486565,516,368,410+0.10%
realworld.run.windows.x64.checked.mch49,051,947,64149,092,550,772+0.08%
smoke_tests.nativeaot.windows.x64.checked.mch22,689,115,46422,707,367,888+0.08%

3. JIT unit tests

3-1

Comments: We are not using the full JIT test suite because the emulator has its own limitation and when the test sets is too big, emulator itself will have some non-deterministic behaviors, to avoid it, we did some effort to figure out the best coverage that will generate stable testing results.

Comments: Within this subset shown in the screen shot, all the tests are passing without REX2 (DOTNET_JitStressRex2Encoding=0) and with REX2 (DOTNET_JitStressRex2Encoding = 1) with some know exceptions caused by the emulator itself, i.e. CodegenBringUpTests and IL_Comformance will break due to the fact that there are some existing try-catch structures, and some exceptions are supposed to be caught by the runtime, but first caught by the emulator.

@dotnet-policy-servicedotnet-policy-serviceBot added the community-contribution Indicates that the PR has been added by a community member label Aug 16, 2024
@dotnet-policy-service

dotnet-policy-serviceBot commented Aug 16, 2024

Copy link
Copy Markdown
Contributor

Tagging subscribers to this area: @JulieLeeMSFT, @jakobbotsch
See info in area-owners.md if you want to be subscribed.

@Ruihan-Yin

Ruihan-Yin commented Aug 16, 2024

Copy link
Copy Markdown
MemberAuthor

The base of 2 APX related PRs( CPUID: #104637 and REX2: #106557) is outdated, I will work offline to resolve the conflicts and rebase the branch.

We are willing to discuss the design and tests here, please feel free to leave a comment if any question or suggestion

@Ruihan-YinRuihan-Yin mentioned this pull request Aug 16, 2024
Merged
@BruceForstallBruceForstall added the apx Related to the Intel Advanced Performance Extensions (APX) label Sep 5, 2024
@dotnet-policy-service

dotnet-policy-serviceBot commented Oct 5, 2024

Copy link
Copy Markdown
Contributor

Draft Pull Request was automatically closed for 30 days of inactivity. Please let us know if you'd like to reopen it.

@Ruihan-YinRuihan-Yin mentioned this pull request Oct 11, 2024
Merged
@BruceForstallBruceForstall mentioned this pull request Oct 15, 2024
Closed
48 tasks
@anthonycanino

anthonycanino commented Oct 21, 2024

Copy link
Copy Markdown
Contributor

@dotnet/avx512-contrib can we reopen this as a PR ready to review?

@BruceForstall

BruceForstall commented Oct 21, 2024

Copy link
Copy Markdown
Contributor

@anthonycanino I re-opened it (it wasn't clear to me if your question implied you did not have permission to do so). Either you or @Ruihan-Yin need to update to latest main and resolve the conflicts, then mark it ready-for-review.

This was referenced Oct 23, 2024
Closed
Closed
Closed
Closed
Open
Closed
Closed
Closed
@Ruihan-Yin
Ruihan-Yin marked this pull request as ready for review November 13, 2024 01:27
This was referenced Nov 13, 2024
Open
Closed
Open
@Ruihan-Yin

Ruihan-Yin commented Nov 18, 2024

Copy link
Copy Markdown
MemberAuthor

Hi @tannergooding, thanks for the reviews in #104637, it seems like the CPUID changes are just pending merge and there should be no major changes expected, so while waiting, I wonder if we can start the conversion on this PR?

@tannergooding

tannergooding commented Nov 18, 2024

Copy link
Copy Markdown
Member

@Ruihan-Yin, just got #104637 merged. If we could get this PR updated so it contains just the new changes, that should make it a lot simpler to review and get in.

@Ruihan-Yin

Ruihan-Yin commented Nov 18, 2024

Copy link
Copy Markdown
MemberAuthor

Thanks! I will work on it soon.

@BruceForstall

BruceForstall commented Nov 20, 2024

Copy link
Copy Markdown
Contributor

Do you have plans to bump the LLVM version for .NET 10?

Yes. Also, I expect to re-tool the cordistools build process to make it easier than it is currently to update the dependent LLVM version.

@MichalPetryka thanks for that link. That might make it easier to build cordistools with our AzureLinux containers (currently, the libc in Ubuntu 16.04 I think is limiting our ability to build new LLVM). However, note that coredistools version of LLVM and .NET 10 version of LLVM are not currently related, and hopefully can still be chosen independently, in the future.

@Ruihan-Yin

Ruihan-Yin commented Nov 22, 2024

Copy link
Copy Markdown
MemberAuthor

Hi @tannergooding, I tried to refactor the stress mode for REX2, can you please check if it is as expected?

Plus, there are a few questions regarding to the comments, would appreciate it if you can take a look.

Thanks!

@BruceForstall

BruceForstall commented Dec 2, 2024

Copy link
Copy Markdown
Contributor

@tannergooding ping on @Ruihan-Yin 's latest questions

@build-analysisbuild-analysisBot mentioned this pull request Dec 3, 2024
Closed
tannergooding
tannergooding reviewed Dec 4, 2024
View reviewed changes
Comment threadsrc/coreclr/jit/compiler.cpp Outdated
@@ -2297,6 +2297,10 @@ void Compiler::compSetProcessor()
codeGen->GetEmitter()->SetUseEvexEncoding(true);
// TODO-XArch-AVX512 : Revisit other flags to be set once avx512 instructions are added.

@tannergoodingtannergoodingDec 4, 2024

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Not related to this PR, but this comment is likely irrelevant now.

tannergooding
tannergooding reviewed Dec 4, 2024
View reviewed changes
Comment threadsrc/coreclr/jit/emitxarch.cpp Outdated
Comment on lines +1386 to +1388
// TODO-apx:
// there are duplicated stress logics here and in HasExtendedGPReg()
// need to clean up later.

@tannergoodingtannergoodingDec 4, 2024

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This comment looks outdated now

tannergooding
tannergooding reviewed Dec 4, 2024
View reviewed changes
Comment threadsrc/coreclr/jit/emitxarch.cpp Outdated
Comment on lines +1885 to +1892
// TODO-apx: It would be better to have stress mode on LSRA to forcely allocate EGPRs,
// instead of stressing here.
#if defined(DEBUG)
if (emitComp->DoJitStressRex2Encoding())
{
return true;
}
#endif // DEBUG

@tannergoodingtannergoodingDec 4, 2024

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

We should make sure a tracking issue exists for this.

@tannergoodingtannergoodingDec 4, 2024

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Actually, I'm not quite sure why this check is needed. We don't have/need such a path for EVEX, we simply emit the EVEX encoding always under stress mode.

We then use the JitStressReg knobs to preference different register sets.

@Ruihan-YinRuihan-YinDec 4, 2024

Copy link
Copy Markdown
MemberAuthor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Yes, I agree that this part is now not needed. I will remove it.

tannergooding
tannergooding reviewed Dec 4, 2024
View reviewed changes
Comment threadsrc/coreclr/jit/emitxarch.cpp Outdated
}
#endif // DEBUG

if (UseRex2Encoding())

@tannergoodingtannergoodingDec 4, 2024

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Is this check actually necessary?

IsExtendedReg freely returns true for XMM16-XMM31 without a corresponding EVEX check. So I would have expected this can simply be (reg >= REG_R16) && (reg <= REG_R31) -- with a temporary path doing || (DoJitStressRex2Encoding() && (reg >= REG_RAX) && (reg <= REG_R15)) until LSRA is properly updated, but as per the above I'm not sure that's really needed

@Ruihan-YinRuihan-YinDec 4, 2024

Copy link
Copy Markdown
MemberAuthor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

IsExtendedGPReg is only used in HasExtendedGPReg, which is then used in TakesRex2Prefix, I think it should be fine to leave a simple check like (reg >= REG_R16) && (reg <= REG_R31) there. If we stress REX2 encoding, TakesRex2Prefix will return true by DoJitStressRex2Encoding check and never reach the register check.

For now, since the EGPR definition is still missing, I will simply return false there, and come back when EGPRs are defined. I will document this with an issue. (I'd expect this could be addressed within the PR for register allocator

tannergooding
tannergooding approved these changes Dec 4, 2024
View reviewed changes

@tannergoodingtannergooding left a comment

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The changes generally look good/correct to me now for a foundational PR.

I think we need at least one tracking issue covering the TODOs in here and ensuring they're cleaned up as the other work gets completed.

This needs an additional review from @dotnet/jit-contrib

@Ruihan-YinRuihan-Yin mentioned this pull request Dec 4, 2024
Closed
3 tasks
@Ruihan-Yin

Ruihan-Yin commented Dec 4, 2024

Copy link
Copy Markdown
MemberAuthor

Thanks for the review!

TODOs are now tracked in #110414.

This was referenced Dec 4, 2024
Closed
Closed
BruceForstall
BruceForstall reviewed Dec 14, 2024
View reviewed changes

@BruceForstallBruceForstall left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

A few somewhat minor suggestions and comments.

Comment threadsrc/coreclr/jit/codegenxarch.cpp Outdated
genDefineTempLabel(genCreateTempLabel());

// This test suite needs REX2 enabled.
assert(theEmitter->UseRex2Encoding() || theEmitter->emitComp->DoJitStressRex2Encoding());

@BruceForstallBruceForstallDec 14, 2024

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Is it the case that these are all "normal" non-APX / non-REX2 instructions, that only test REX2 if REX2 stress is enabled? So to use this, you need to set DOTNET_JitStressRex2Encoding=1 and DOTNET_JitEmitUnitTestsSections=apx?

Note that these asserts will cause DOTNET_JitEmitUnitTestsSections=all (without DOTNET_JitStressRex2Encoding=1 set) to assert, which is undesirable. Instead of assert, maybe just return if REX2 is not being stressed? Or maybe it's possible that if you get here, to temporarily "turn on" REX2 stress/encoding for the duration of this function?

Presumably there will be additional cases, with EGPR high registers, where the REX2 encodings will be required.

@Ruihan-YinRuihan-YinDec 16, 2024

Copy link
Copy Markdown
MemberAuthor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

If we temporarily turn on REX2 stress only in this function, then only the code generator will do its work using REX2 encoding, after we exit the function and turn off REX2 stress, the corresponding emitter will emit code under REX2 off.

I can make it an early return if REX2 is not stressed.

And yes, currently the emitter unit test can only perform encoding on GPR0~15, while we are still waiting for the EGPR definition, that will come with the PR for LSRA. I can add a TODO item for more unit tests with EGPRs to #110414 if needed.

Comment threadsrc/coreclr/jit/codegenxarch.cpp Outdated
// it might fail due to stack value unavailable/mismatch, since these tests are mainly for
// encoding correctness check, this kind of failures may be considered as not harmful.

GenTree* stkNum = theEmitter->emitComp->stackState.esStack[0].val;

@BruceForstallBruceForstallDec 14, 2024

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I don't see the relationship to the stackState variable.

Couldn't you:

GenTreePhysReg physReg(REG_EDX);
GenTreeIndir load(TYP_INT, &physReg);

to create [EDX] addressing mode?

Comment threadsrc/coreclr/jit/compiler.h Outdated
{
// we should make sure EVEX is also stressed when REX2 is stressed, as we will need to guarantee EGPR
// functionality is properly turned on for every instructions when REX2 is stress.
assert(JitConfig.JitStressEvexEncoding());

@BruceForstallBruceForstallDec 14, 2024

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This means that if you set DOTNET_JitStressRex2Encoding=1 but don't set DOTNET_JitStressEvexEncoding=1 you will get an assert. That's annoying. Is it really necessary? That is, can you stress REX2 without also stressing EVEX?

If it's required to also stress EVEX, I suggest creating a small helper function:

bool JitStressEvexEncoding() const
{
return JitConfig.JitStressEvexEncoding() || JitConfig.JitStressRex2Encoding();
}

and replace all occurrences of calls to JitConfig.JitStressEvexEncoding() to calls to this new helper function. Then you can remove this assert.

@Ruihan-YinRuihan-YinDec 16, 2024

Copy link
Copy Markdown
MemberAuthor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

We had more discussion on this here: #106557 (comment)

I will replace the EVEX encoding check with the suggested helper, thanks for pointing out.

Comment threadsrc/coreclr/jit/emitxarch.cpp Outdated
// 2-byte
return true;
}
if ((code & 0xFF0000) == 0x0F0000)

@BruceForstallBruceForstallDec 14, 2024

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Does this need to be:

 if ((code & 0xFFFF0000) == 0x000F0000)

(that is, check that high byte 4 is zero)? If we don't check this, couldn't a 4-byte code with 0F second byte be caught?

Comment threadsrc/coreclr/jit/emitxarch.cpp Outdated

if ((code & 0xFF000000) == 0x0F000000)
{
// 4-byte, need to check if PP is prefixs

@BruceForstallBruceForstallDec 14, 2024

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

nit

Suggested change
// 4-byte, need to check if PP is prefixs
// 4-byte, need to check if PP is a prefix

Comment threadsrc/coreclr/jit/emitxarch.cpp Outdated
// TODO-xarch-apx:
// At this stage, we are only using REX2 in the case that non-simd integer instructions
// with EGPRs being used in its operands, it could be either direct register uses, or
// memory addresssig operands, i.e. index and base.

@BruceForstallBruceForstallDec 14, 2024

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

typo

Suggested change
// memory addresssig operands, i.e. index and base.
// memory addressing operands, i.e. index and base.

Comment threadsrc/coreclr/jit/emitxarch.cpp Outdated
return false;
}

#if defined(DEBUG)

@BruceForstallBruceForstallDec 14, 2024

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

It doesn't really matter, but it seems like the stress check should be last, just before the final return false;

Comment threadsrc/coreclr/jit/emitxarch.cpp Outdated
}

//------------------------------------------------------------------------
// HasExtendedGPReg: Checks if an instruction uses a extended general purpose registers - EGPRs (r16-r31)

@BruceForstallBruceForstallDec 14, 2024

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

nit

Suggested change
// HasExtendedGPReg: Checks if an instruction uses a extended generalpurpose registers - EGPRs (r16-r31)
// HasExtendedGPReg: Checks if an instruction uses an extended general-purpose register - EGPR (r16-r31)

Comment threadsrc/coreclr/jit/emitxarch.cpp Outdated
if ((code & 0xFF) == 0x0F)
{
// some map-1 instructions have opcode in forms like:
// XX0F, remove the leading 0x0F byte as it have been recoreded in REX2.

@BruceForstallBruceForstallDec 14, 2024

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Suggested change
// XX0F, remove the leading 0x0F byte as it have been recoreded in REX2.
// XX0F, remove the leading 0x0F byte as it has been recorded in REX2.

Comment threadsrc/coreclr/jit/emitxarch.cpp Outdated
}
if (false /*reg >= REG_R16 && reg <= REG_R31*/)
{
// seperate the encoding for REX2.B3/B4, REX2.B3 will be handled in `AddRexBPrefix`.

@BruceForstallBruceForstallDec 14, 2024

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Suggested change
//seperate the encoding for REX2.B3/B4, REX2.B3 will be handled in `AddRexBPrefix`.
//Separate the encoding for REX2.B3/B4, REX2.B3 will be handled in `AddRexBPrefix`.

@BruceForstall
BruceForstall merged commit 3410c76 into dotnet:mainDec 17, 2024
@Ruihan-Yin

Ruihan-Yin commented Dec 17, 2024

Copy link
Copy Markdown
MemberAuthor

Thanks all for the reviews and suggestions!

This was referenced Dec 18, 2024
Closed
Closed
@anthonycaninoanthonycanino mentioned this pull request Jan 3, 2025
Merged
@github-actionsgithub-actionsBot locked and limited conversation to collaborators Jan 16, 2025
Sign up for freeto subscribe to this conversation on GitHub. Already have an account? Sign in.

Reviewers

@tannergoodingtannergoodingtannergooding approved these changes
@MichalStrehovskyMichalStrehovskyAwaiting requested review from MichalStrehovsky
+1 more reviewer
@BruceForstallBruceForstallBruceForstall approved these changes
Reviewers whose approvals may not affect merge requirements

Assignees

No one assigned

Labels

apxRelated to the Intel Advanced Performance Extensions (APX)area-CodeGen-coreclrCLR JIT compiler in src/coreclr/src/jit and related components such as SuperPMIcommunity-contributionIndicates that the PR has been added by a community member

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

5 participants

@Ruihan-Yin@anthonycanino@BruceForstall@tannergooding@MichalPetryka