Do not embed result type in VN for scalar hw intrinsics by kunalspathak · Pull Request #72895 · dotnet/runtime

kunalspathak · 2022-07-26T23:44:58Z

For scalar hw intrinsics, we were creating VNForSimdType and it was manifesting in vnDumpSimdType().

Fixes: #69551

kunalspathak · 2022-07-26T23:45:11Z

@SingleAccretion , @dotnet/jit-contrib , @tannergooding

ghost · 2022-07-26T23:45:14Z

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

Issue Details

For scalar hw intrinsics, we were creating VNForSimdType and it was manifesting in vnDumpSimdType().

Fixes: #69551

Author:	kunalspathak
Assignees:	-
Labels:	`area-CodeGen-coreclr`
Milestone:	-

tannergooding · 2022-07-28T15:20:37Z

+ if (HWIntrinsicInfo::lookupCategory(hwIntrinsicID) == HW_Category_Scalar)
+ {
+ return false;
+ }


We may also want a path for HW_Category_Special or maybe better check for:
Suggested change
if (HWIntrinsicInfo::lookupCategory(hwIntrinsicID) == HW_Category_Scalar)
{
returnfalse;
}
unsigned simdSize = 0;
if (HWIntrinsicInfo::tryLookupSimdSize(hwIntrinsicID, &simdSize) && (simdSize == 0))
{
returnfalse;
}
All scalar intrinsics, regardless of category, should have a simdSize == 0. Likewise, all SIMD intrinsics should have a size of 8, 16, 32, or -1 (where -1 implies the intrinsic supports multiple SIMD sizes and is not a scalar).

I think the current check is "accurate enough" today since the few "special" scalars tend to have all INS_invalid entries and are specially handled in codegen instead, but its probably better to be more explicit here.
There's also a consideration for the type encoding that certain special or helper intrinsics might specify INS_invalid and get specialized handling in lowering/codegen and so it might be important to have the type anyways. I'm unsure if there is any actual issue here or not, but its probably worth auditing at some point in the future.

I agree, except that it doesn't set the simdSize if it was -1. So probably I should initialize simdSize to something like 100 or something and then see if became 0. Wil fix it.

That's what the tryLookupSimdSize check was for. It should return false if simdSize == -1 and so the second check (simdSize == 0) won't get hit and we'll treat it as SIMD rather than Scalar

ah. was an oversight on my part.

kunalspathak · 2022-07-26T23:44:58Z

For scalar hw intrinsics, we were creating VNForSimdType and it was manifesting in vnDumpSimdType().

Fixes: #69551

kunalspathak · 2022-07-26T23:45:11Z

@SingleAccretion , @dotnet/jit-contrib , @tannergooding

ghost · 2022-07-26T23:45:14Z

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

Issue Details

For scalar hw intrinsics, we were creating VNForSimdType and it was manifesting in vnDumpSimdType().

Fixes: #69551

Author:	kunalspathak
Assignees:	-
Labels:	`area-CodeGen-coreclr`
Milestone:	-

tannergooding · 2022-07-28T15:20:37Z

+ if (HWIntrinsicInfo::lookupCategory(hwIntrinsicID) == HW_Category_Scalar)
+ {
+ return false;
+ }


We may also want a path for HW_Category_Special or maybe better check for:
Suggested change
if (HWIntrinsicInfo::lookupCategory(hwIntrinsicID) == HW_Category_Scalar)
{
returnfalse;
}
unsigned simdSize = 0;
if (HWIntrinsicInfo::tryLookupSimdSize(hwIntrinsicID, &simdSize) && (simdSize == 0))
{
returnfalse;
}
All scalar intrinsics, regardless of category, should have a simdSize == 0. Likewise, all SIMD intrinsics should have a size of 8, 16, 32, or -1 (where -1 implies the intrinsic supports multiple SIMD sizes and is not a scalar).

I think the current check is "accurate enough" today since the few "special" scalars tend to have all INS_invalid entries and are specially handled in codegen instead, but its probably better to be more explicit here.
There's also a consideration for the type encoding that certain special or helper intrinsics might specify INS_invalid and get specialized handling in lowering/codegen and so it might be important to have the type anyways. I'm unsure if there is any actual issue here or not, but its probably worth auditing at some point in the future.

I agree, except that it doesn't set the simdSize if it was -1. So probably I should initialize simdSize to something like 100 or something and then see if became 0. Wil fix it.

That's what the tryLookupSimdSize check was for. It should return false if simdSize == -1 and so the second check (simdSize == 0) won't get hit and we'll treat it as SIMD rather than Scalar

ah. was an oversight on my part.

kunalspathak · 2022-07-26T23:44:58Z

For scalar hw intrinsics, we were creating VNForSimdType and it was manifesting in vnDumpSimdType().

Fixes: #69551

kunalspathak · 2022-07-26T23:45:11Z

@SingleAccretion , @dotnet/jit-contrib , @tannergooding

ghost · 2022-07-26T23:45:14Z

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

Issue Details

For scalar hw intrinsics, we were creating VNForSimdType and it was manifesting in vnDumpSimdType().

Fixes: #69551

Author:	kunalspathak
Assignees:	-
Labels:	`area-CodeGen-coreclr`
Milestone:	-

tannergooding · 2022-07-28T15:20:37Z

+ if (HWIntrinsicInfo::lookupCategory(hwIntrinsicID) == HW_Category_Scalar)
+ {
+ return false;
+ }


We may also want a path for HW_Category_Special or maybe better check for:
Suggested change
if (HWIntrinsicInfo::lookupCategory(hwIntrinsicID) == HW_Category_Scalar)
{
returnfalse;
}
unsigned simdSize = 0;
if (HWIntrinsicInfo::tryLookupSimdSize(hwIntrinsicID, &simdSize) && (simdSize == 0))
{
returnfalse;
}
All scalar intrinsics, regardless of category, should have a simdSize == 0. Likewise, all SIMD intrinsics should have a size of 8, 16, 32, or -1 (where -1 implies the intrinsic supports multiple SIMD sizes and is not a scalar).

I think the current check is "accurate enough" today since the few "special" scalars tend to have all INS_invalid entries and are specially handled in codegen instead, but its probably better to be more explicit here.
There's also a consideration for the type encoding that certain special or helper intrinsics might specify INS_invalid and get specialized handling in lowering/codegen and so it might be important to have the type anyways. I'm unsure if there is any actual issue here or not, but its probably worth auditing at some point in the future.

I agree, except that it doesn't set the simdSize if it was -1. So probably I should initialize simdSize to something like 100 or something and then see if became 0. Wil fix it.

That's what the tryLookupSimdSize check was for. It should return false if simdSize == -1 and so the second check (simdSize == 0) won't get hit and we'll treat it as SIMD rather than Scalar

ah. was an oversight on my part.

kunalspathak · 2022-07-26T23:44:58Z

For scalar hw intrinsics, we were creating VNForSimdType and it was manifesting in vnDumpSimdType().

Fixes: #69551

kunalspathak · 2022-07-26T23:45:11Z

@SingleAccretion , @dotnet/jit-contrib , @tannergooding

ghost · 2022-07-26T23:45:14Z

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

Issue Details

For scalar hw intrinsics, we were creating VNForSimdType and it was manifesting in vnDumpSimdType().

Fixes: #69551

Author:	kunalspathak
Assignees:	-
Labels:	`area-CodeGen-coreclr`
Milestone:	-

tannergooding · 2022-07-28T15:20:37Z

+ if (HWIntrinsicInfo::lookupCategory(hwIntrinsicID) == HW_Category_Scalar)
+ {
+ return false;
+ }


We may also want a path for HW_Category_Special or maybe better check for:
Suggested change
if (HWIntrinsicInfo::lookupCategory(hwIntrinsicID) == HW_Category_Scalar)
{
returnfalse;
}
unsigned simdSize = 0;
if (HWIntrinsicInfo::tryLookupSimdSize(hwIntrinsicID, &simdSize) && (simdSize == 0))
{
returnfalse;
}
All scalar intrinsics, regardless of category, should have a simdSize == 0. Likewise, all SIMD intrinsics should have a size of 8, 16, 32, or -1 (where -1 implies the intrinsic supports multiple SIMD sizes and is not a scalar).

I think the current check is "accurate enough" today since the few "special" scalars tend to have all INS_invalid entries and are specially handled in codegen instead, but its probably better to be more explicit here.
There's also a consideration for the type encoding that certain special or helper intrinsics might specify INS_invalid and get specialized handling in lowering/codegen and so it might be important to have the type anyways. I'm unsure if there is any actual issue here or not, but its probably worth auditing at some point in the future.

I agree, except that it doesn't set the simdSize if it was -1. So probably I should initialize simdSize to something like 100 or something and then see if became 0. Wil fix it.

That's what the tryLookupSimdSize check was for. It should return false if simdSize == -1 and so the second check (simdSize == 0) won't get hit and we'll treat it as SIMD rather than Scalar

ah. was an oversight on my part.

kunalspathak · 2022-07-26T23:44:58Z

For scalar hw intrinsics, we were creating VNForSimdType and it was manifesting in vnDumpSimdType().

Fixes: #69551

kunalspathak · 2022-07-26T23:45:11Z

@SingleAccretion , @dotnet/jit-contrib , @tannergooding

ghost · 2022-07-26T23:45:14Z

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

Issue Details

For scalar hw intrinsics, we were creating VNForSimdType and it was manifesting in vnDumpSimdType().

Fixes: #69551

Author:	kunalspathak
Assignees:	-
Labels:	`area-CodeGen-coreclr`
Milestone:	-

tannergooding · 2022-07-28T15:20:37Z

+ if (HWIntrinsicInfo::lookupCategory(hwIntrinsicID) == HW_Category_Scalar)
+ {
+ return false;
+ }


We may also want a path for HW_Category_Special or maybe better check for:
Suggested change
if (HWIntrinsicInfo::lookupCategory(hwIntrinsicID) == HW_Category_Scalar)
{
returnfalse;
}
unsigned simdSize = 0;
if (HWIntrinsicInfo::tryLookupSimdSize(hwIntrinsicID, &simdSize) && (simdSize == 0))
{
returnfalse;
}
All scalar intrinsics, regardless of category, should have a simdSize == 0. Likewise, all SIMD intrinsics should have a size of 8, 16, 32, or -1 (where -1 implies the intrinsic supports multiple SIMD sizes and is not a scalar).

I think the current check is "accurate enough" today since the few "special" scalars tend to have all INS_invalid entries and are specially handled in codegen instead, but its probably better to be more explicit here.
There's also a consideration for the type encoding that certain special or helper intrinsics might specify INS_invalid and get specialized handling in lowering/codegen and so it might be important to have the type anyways. I'm unsure if there is any actual issue here or not, but its probably worth auditing at some point in the future.

I agree, except that it doesn't set the simdSize if it was -1. So probably I should initialize simdSize to something like 100 or something and then see if became 0. Wil fix it.

That's what the tryLookupSimdSize check was for. It should return false if simdSize == -1 and so the second check (simdSize == 0) won't get hit and we'll treat it as SIMD rather than Scalar

ah. was an oversight on my part.

kunalspathak · 2022-07-26T23:44:58Z

For scalar hw intrinsics, we were creating VNForSimdType and it was manifesting in vnDumpSimdType().

Fixes: #69551

kunalspathak · 2022-07-26T23:45:11Z

@SingleAccretion , @dotnet/jit-contrib , @tannergooding

ghost · 2022-07-26T23:45:14Z

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

Issue Details

For scalar hw intrinsics, we were creating VNForSimdType and it was manifesting in vnDumpSimdType().

Fixes: #69551

Author:	kunalspathak
Assignees:	-
Labels:	`area-CodeGen-coreclr`
Milestone:	-

tannergooding · 2022-07-28T15:20:37Z

+ if (HWIntrinsicInfo::lookupCategory(hwIntrinsicID) == HW_Category_Scalar)
+ {
+ return false;
+ }


We may also want a path for HW_Category_Special or maybe better check for:
Suggested change
if (HWIntrinsicInfo::lookupCategory(hwIntrinsicID) == HW_Category_Scalar)
{
returnfalse;
}
unsigned simdSize = 0;
if (HWIntrinsicInfo::tryLookupSimdSize(hwIntrinsicID, &simdSize) && (simdSize == 0))
{
returnfalse;
}
All scalar intrinsics, regardless of category, should have a simdSize == 0. Likewise, all SIMD intrinsics should have a size of 8, 16, 32, or -1 (where -1 implies the intrinsic supports multiple SIMD sizes and is not a scalar).

I think the current check is "accurate enough" today since the few "special" scalars tend to have all INS_invalid entries and are specially handled in codegen instead, but its probably better to be more explicit here.
There's also a consideration for the type encoding that certain special or helper intrinsics might specify INS_invalid and get specialized handling in lowering/codegen and so it might be important to have the type anyways. I'm unsure if there is any actual issue here or not, but its probably worth auditing at some point in the future.

I agree, except that it doesn't set the simdSize if it was -1. So probably I should initialize simdSize to something like 100 or something and then see if became 0. Wil fix it.

That's what the tryLookupSimdSize check was for. It should return false if simdSize == -1 and so the second check (simdSize == 0) won't get hit and we'll treat it as SIMD rather than Scalar

ah. was an oversight on my part.

kunalspathak · 2022-07-26T23:44:58Z

For scalar hw intrinsics, we were creating VNForSimdType and it was manifesting in vnDumpSimdType().

Fixes: #69551

kunalspathak · 2022-07-26T23:45:11Z

@SingleAccretion , @dotnet/jit-contrib , @tannergooding

ghost · 2022-07-26T23:45:14Z

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

Issue Details

For scalar hw intrinsics, we were creating VNForSimdType and it was manifesting in vnDumpSimdType().

Fixes: #69551

Author:	kunalspathak
Assignees:	-
Labels:	`area-CodeGen-coreclr`
Milestone:	-

tannergooding · 2022-07-28T15:20:37Z

+ if (HWIntrinsicInfo::lookupCategory(hwIntrinsicID) == HW_Category_Scalar)
+ {
+ return false;
+ }


We may also want a path for HW_Category_Special or maybe better check for:
Suggested change
if (HWIntrinsicInfo::lookupCategory(hwIntrinsicID) == HW_Category_Scalar)
{
returnfalse;
}
unsigned simdSize = 0;
if (HWIntrinsicInfo::tryLookupSimdSize(hwIntrinsicID, &simdSize) && (simdSize == 0))
{
returnfalse;
}
All scalar intrinsics, regardless of category, should have a simdSize == 0. Likewise, all SIMD intrinsics should have a size of 8, 16, 32, or -1 (where -1 implies the intrinsic supports multiple SIMD sizes and is not a scalar).

I think the current check is "accurate enough" today since the few "special" scalars tend to have all INS_invalid entries and are specially handled in codegen instead, but its probably better to be more explicit here.
There's also a consideration for the type encoding that certain special or helper intrinsics might specify INS_invalid and get specialized handling in lowering/codegen and so it might be important to have the type anyways. I'm unsure if there is any actual issue here or not, but its probably worth auditing at some point in the future.

I agree, except that it doesn't set the simdSize if it was -1. So probably I should initialize simdSize to something like 100 or something and then see if became 0. Wil fix it.

That's what the tryLookupSimdSize check was for. It should return false if simdSize == -1 and so the second check (simdSize == 0) won't get hit and we'll treat it as SIMD rather than Scalar

ah. was an oversight on my part.

kunalspathak · 2022-07-26T23:44:58Z

For scalar hw intrinsics, we were creating VNForSimdType and it was manifesting in vnDumpSimdType().

Fixes: #69551

kunalspathak · 2022-07-26T23:45:11Z

@SingleAccretion , @dotnet/jit-contrib , @tannergooding

ghost · 2022-07-26T23:45:14Z

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

Issue Details

For scalar hw intrinsics, we were creating VNForSimdType and it was manifesting in vnDumpSimdType().

Fixes: #69551

Author:	kunalspathak
Assignees:	-
Labels:	`area-CodeGen-coreclr`
Milestone:	-

tannergooding · 2022-07-28T15:20:37Z

+ if (HWIntrinsicInfo::lookupCategory(hwIntrinsicID) == HW_Category_Scalar)
+ {
+ return false;
+ }


We may also want a path for HW_Category_Special or maybe better check for:
Suggested change
if (HWIntrinsicInfo::lookupCategory(hwIntrinsicID) == HW_Category_Scalar)
{
returnfalse;
}
unsigned simdSize = 0;
if (HWIntrinsicInfo::tryLookupSimdSize(hwIntrinsicID, &simdSize) && (simdSize == 0))
{
returnfalse;
}
All scalar intrinsics, regardless of category, should have a simdSize == 0. Likewise, all SIMD intrinsics should have a size of 8, 16, 32, or -1 (where -1 implies the intrinsic supports multiple SIMD sizes and is not a scalar).

I think the current check is "accurate enough" today since the few "special" scalars tend to have all INS_invalid entries and are specially handled in codegen instead, but its probably better to be more explicit here.
There's also a consideration for the type encoding that certain special or helper intrinsics might specify INS_invalid and get specialized handling in lowering/codegen and so it might be important to have the type anyways. I'm unsure if there is any actual issue here or not, but its probably worth auditing at some point in the future.

I agree, except that it doesn't set the simdSize if it was -1. So probably I should initialize simdSize to something like 100 or something and then see if became 0. Wil fix it.

That's what the tryLookupSimdSize check was for. It should return false if simdSize == -1 and so the second check (simdSize == 0) won't get hit and we'll treat it as SIMD rather than Scalar

ah. was an oversight on my part.

do not embed result type for scalar hw intrinsics
17feeff

ghost added the area-CodeGen-coreclr CLR JIT compiler in src/coreclr/src/jit and related components such as SuperPMI label Jul 26, 2022

ghost assigned kunalspathak Jul 26, 2022

kunalspathak changed the title ~~Do not embed result type for scalar hw intrinsics~~Do not embed result type in VN for scalar hw intrinsicsJul 26, 2022

tannergooding approved these changes Jul 28, 2022
View reviewed changes

tannergooding reviewed Jul 28, 2022
View reviewed changes

kunalspathak added 2 commits July 28, 2022 09:37

Check the simdSize
2e0d1a6

oversight fix
f765190

tannergooding approved these changes Jul 28, 2022
View reviewed changes

kunalspathak merged commit ef7c980 into dotnet:mainJul 28, 2022

kunalspathak deleted the dumpsimdType branch July 28, 2022 20:21

ghost locked as resolved and limited conversation to collaborators Aug 28, 2022

do not embed result type for scalar hw intrinsics
17feeff

ghost added the area-CodeGen-coreclr CLR JIT compiler in src/coreclr/src/jit and related components such as SuperPMI label Jul 26, 2022

ghost assigned kunalspathak Jul 26, 2022

kunalspathak changed the title ~~Do not embed result type for scalar hw intrinsics~~Do not embed result type in VN for scalar hw intrinsicsJul 26, 2022

tannergooding approved these changes Jul 28, 2022
View reviewed changes

tannergooding reviewed Jul 28, 2022
View reviewed changes

kunalspathak added 2 commits July 28, 2022 09:37

Check the simdSize
2e0d1a6

oversight fix
f765190

tannergooding approved these changes Jul 28, 2022
View reviewed changes

kunalspathak merged commit ef7c980 into dotnet:mainJul 28, 2022

kunalspathak deleted the dumpsimdType branch July 28, 2022 20:21

ghost locked as resolved and limited conversation to collaborators Aug 28, 2022

do not embed result type for scalar hw intrinsics
17feeff

ghost added the area-CodeGen-coreclr CLR JIT compiler in src/coreclr/src/jit and related components such as SuperPMI label Jul 26, 2022

ghost assigned kunalspathak Jul 26, 2022

kunalspathak changed the title ~~Do not embed result type for scalar hw intrinsics~~Do not embed result type in VN for scalar hw intrinsicsJul 26, 2022

tannergooding approved these changes Jul 28, 2022
View reviewed changes

tannergooding reviewed Jul 28, 2022
View reviewed changes

kunalspathak added 2 commits July 28, 2022 09:37

Check the simdSize
2e0d1a6

oversight fix
f765190

tannergooding approved these changes Jul 28, 2022
View reviewed changes

kunalspathak merged commit ef7c980 into dotnet:mainJul 28, 2022

kunalspathak deleted the dumpsimdType branch July 28, 2022 20:21

ghost locked as resolved and limited conversation to collaborators Aug 28, 2022

do not embed result type for scalar hw intrinsics
17feeff

ghost added the area-CodeGen-coreclr CLR JIT compiler in src/coreclr/src/jit and related components such as SuperPMI label Jul 26, 2022

ghost assigned kunalspathak Jul 26, 2022

kunalspathak changed the title ~~Do not embed result type for scalar hw intrinsics~~Do not embed result type in VN for scalar hw intrinsicsJul 26, 2022

tannergooding approved these changes Jul 28, 2022
View reviewed changes

tannergooding reviewed Jul 28, 2022
View reviewed changes

kunalspathak added 2 commits July 28, 2022 09:37

Check the simdSize
2e0d1a6

oversight fix
f765190

tannergooding approved these changes Jul 28, 2022
View reviewed changes

kunalspathak merged commit ef7c980 into dotnet:mainJul 28, 2022

kunalspathak deleted the dumpsimdType branch July 28, 2022 20:21

ghost locked as resolved and limited conversation to collaborators Aug 28, 2022

do not embed result type for scalar hw intrinsics
17feeff

ghost added the area-CodeGen-coreclr CLR JIT compiler in src/coreclr/src/jit and related components such as SuperPMI label Jul 26, 2022

ghost assigned kunalspathak Jul 26, 2022

kunalspathak changed the title ~~Do not embed result type for scalar hw intrinsics~~Do not embed result type in VN for scalar hw intrinsicsJul 26, 2022

tannergooding approved these changes Jul 28, 2022
View reviewed changes

tannergooding reviewed Jul 28, 2022
View reviewed changes

kunalspathak added 2 commits July 28, 2022 09:37

Check the simdSize
2e0d1a6

oversight fix
f765190

tannergooding approved these changes Jul 28, 2022
View reviewed changes

kunalspathak merged commit ef7c980 into dotnet:mainJul 28, 2022

kunalspathak deleted the dumpsimdType branch July 28, 2022 20:21

ghost locked as resolved and limited conversation to collaborators Aug 28, 2022

do not embed result type for scalar hw intrinsics
17feeff

ghost added the area-CodeGen-coreclr CLR JIT compiler in src/coreclr/src/jit and related components such as SuperPMI label Jul 26, 2022

ghost assigned kunalspathak Jul 26, 2022

kunalspathak changed the title ~~Do not embed result type for scalar hw intrinsics~~Do not embed result type in VN for scalar hw intrinsicsJul 26, 2022

tannergooding approved these changes Jul 28, 2022
View reviewed changes

tannergooding reviewed Jul 28, 2022
View reviewed changes

kunalspathak added 2 commits July 28, 2022 09:37

Check the simdSize
2e0d1a6

oversight fix
f765190

tannergooding approved these changes Jul 28, 2022
View reviewed changes

kunalspathak merged commit ef7c980 into dotnet:mainJul 28, 2022

kunalspathak deleted the dumpsimdType branch July 28, 2022 20:21

ghost locked as resolved and limited conversation to collaborators Aug 28, 2022

do not embed result type for scalar hw intrinsics
17feeff

ghost added the area-CodeGen-coreclr CLR JIT compiler in src/coreclr/src/jit and related components such as SuperPMI label Jul 26, 2022

ghost assigned kunalspathak Jul 26, 2022

kunalspathak changed the title ~~Do not embed result type for scalar hw intrinsics~~Do not embed result type in VN for scalar hw intrinsicsJul 26, 2022

tannergooding approved these changes Jul 28, 2022
View reviewed changes

tannergooding reviewed Jul 28, 2022
View reviewed changes

kunalspathak added 2 commits July 28, 2022 09:37

Check the simdSize
2e0d1a6

oversight fix
f765190

tannergooding approved these changes Jul 28, 2022
View reviewed changes

kunalspathak merged commit ef7c980 into dotnet:mainJul 28, 2022

kunalspathak deleted the dumpsimdType branch July 28, 2022 20:21

ghost locked as resolved and limited conversation to collaborators Aug 28, 2022

do not embed result type for scalar hw intrinsics
17feeff

ghost added the area-CodeGen-coreclr CLR JIT compiler in src/coreclr/src/jit and related components such as SuperPMI label Jul 26, 2022

ghost assigned kunalspathak Jul 26, 2022

kunalspathak changed the title ~~Do not embed result type for scalar hw intrinsics~~Do not embed result type in VN for scalar hw intrinsicsJul 26, 2022

tannergooding approved these changes Jul 28, 2022
View reviewed changes

tannergooding reviewed Jul 28, 2022
View reviewed changes

kunalspathak added 2 commits July 28, 2022 09:37

Check the simdSize
2e0d1a6

oversight fix
f765190

tannergooding approved these changes Jul 28, 2022
View reviewed changes

kunalspathak merged commit ef7c980 into dotnet:mainJul 28, 2022

kunalspathak deleted the dumpsimdType branch July 28, 2022 20:21

ghost locked as resolved and limited conversation to collaborators Aug 28, 2022

-if (HWIntrinsicInfo::lookupCategory(hwIntrinsicID) == HW_Category_Scalar)
- {
-returnfalse;
- }
+unsigned simdSize = 0;
+if (HWIntrinsicInfo::tryLookupSimdSize(hwIntrinsicID, &simdSize) && (simdSize == 0))
+ {
+returnfalse;
+ }

Uh oh!

Conversation

kunalspathak commented Jul 26, 2022

Uh oh!

kunalspathak commented Jul 26, 2022

Uh oh!

ghost commented Jul 26, 2022

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

tannergoodingJul 28, 2022• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Uh oh!

Conversation

kunalspathak commented Jul 26, 2022

Uh oh!

kunalspathak commented Jul 26, 2022

Uh oh!

ghost commented Jul 26, 2022

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

tannergoodingJul 28, 2022• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Uh oh!

Conversation

kunalspathak commented Jul 26, 2022

Uh oh!

kunalspathak commented Jul 26, 2022

Uh oh!

ghost commented Jul 26, 2022

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

tannergoodingJul 28, 2022• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

tannergoodingJul 28, 2022•
edited
Loading

tannergoodingJul 28, 2022•
edited
Loading

tannergoodingJul 28, 2022•
edited
Loading

tannergoodingJul 28, 2022•
edited
Loading