Optimize try_eval_bits to avoid layout queries by Mark-Simulacrum · Pull Request #64673 · rust-lang/rust

Mark-Simulacrum · 2019-09-21T22:47:42Z

This specifically targets match checking, but is possibly more widely
useful as well. In code with large, single-value match statements, we
were previously spending a lot of time running layout_of for the
primitive types (integers, chars) -- which is essentially useless. This
optimizes the code to avoid those query calls by directly obtaining the
size for these types, when possible.

It may be worth considering adding a size_of query in the future which
might be far faster, especially if specialized for "const" cases --
match arms being the most obvious example. It's possibly such a function
would benefit from not being a query as well, since it's trivially
evaluatable from the sty for many cases whereas a query needs to hash
the input and such.

Mark-Simulacrum · 2019-09-21T22:47:56Z

@bors try @rust-timer queue

rust-timer · 2019-09-21T22:47:57Z

Awaiting bors try build completion

bors · 2019-09-21T22:48:08Z

⌛ Trying commit 511352c with merge 9b2e58c...

@Centril

Optimize match checking to avoid layout queries In code with large, single-value match statements, we were previously spending a lot of time running layout_of for the primitive types (integers, chars) -- which is essentially useless. This optimizes the code to avoid those query calls by directly obtaining the size for these types, when possible. We fallback to the (slower) previous code if that fails, so this is not a behavior change. r? @Centril who I believe knows this code enough, but if not feel free to re-assign

Centril · 2019-09-21T22:49:19Z

r? @oli-obk cc @varkor @arielb1

bors · 2019-09-22T01:51:02Z

☀️ Try build successful - checks-azure
Build commit: 9b2e58c

rust-timer · 2019-09-22T01:51:04Z

Queued 9b2e58c with parent ed8b708, future comparison URL.

rust-timer · 2019-09-22T03:32:56Z

Finished benchmarking try commit 9b2e58c, comparison URL.

bjorn3 · 2019-09-22T09:26:41Z

unicode_normalization is ~30% faster! 3 other benchea got ~2% slower. The rest is stable.

Mark-Simulacrum · 2019-09-22T13:01:07Z

Okay, moved into the try_eval_bits function -- locally that shows that this unicode_normalization is a bit slower (~0.05s) after doing so, but that might be noise (although I get pretty consistent results).

Let's get some official results though @bors try @rust-timer queue

rust-timer · 2019-09-22T13:01:08Z

Awaiting bors try build completion

bors · 2019-09-22T13:01:18Z

⌛ Trying commit 9c62117 with merge 64687bb...

@Centril

Optimize match checking to avoid layout queries In code with large, single-value match statements, we were previously spending a lot of time running layout_of for the primitive types (integers, chars) -- which is essentially useless. This optimizes the code to avoid those query calls by directly obtaining the size for these types, when possible. We fallback to the (slower) previous code if that fails, so this is not a behavior change. r? @Centril who I believe knows this code enough, but if not feel free to re-assign

bors · 2019-09-22T15:45:38Z

☀️ Try build successful - checks-azure
Build commit: 64687bb

rust-timer · 2019-09-22T15:45:39Z

Queued 64687bb with parent 4ff32c0, future comparison URL.

bjorn3 · 2019-09-22T15:59:03Z

Compilation of four crates failed during benchmarking:

thread \'rustc\' panicked at \'assertion failed: pos.checked_add(num_bytes).unwrap() <= self.mapped_file.len()\', /cargo/registry/src/github.com-1ecc6299db9ec823/measureme-0.3.0/src/mmap_serialization_sink.rs:38:9
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace.
error: internal compiler error: unexpected panic
note: the compiler unexpectedly panicked. this is a bug.
note: we would appreciate a bug report: https://github.com/rust-lang/rust/blob/master/CONTRIBUTING.md#bug-reports
note: rustc 1.39.0-nightly (ed8b708c1 2019-09-21) running on x86_64-unknown-linux-gnu
note: compiler flags: -Z self-profile=/tmp/.tmp1D64N5/self-profile-output -Z self-profile-events=all -C debuginfo=2 --crate-type lib
note: some of the compiler flags provided by cargo are hidden
thread \'rustc\' panicked at \'index 1073741840 out of range for slice of length 1073741824\', src/libcore/slice/mod.rs:2583:5
stack backtrace:
0: 0x7fd1648a2b04 - backtrace::backtrace::libunwind::trace::h61b15987b9420dc8
at /cargo/registry/src/github.com-1ecc6299db9ec823/backtrace-0.3.37/src/backtrace/libunwind.rs:88
1: 0x7fd1648a2b04 - backtrace::backtrace::trace_unsynchronized::h944547918bca7d09
at /cargo/registry/src/github.com-1ecc6299db9ec823/backtrace-0.3.37/src/backtrace/mod.rs:66
2: 0x7fd1648a2b04 - std::sys_common::backtrace::_print_fmt::hf631db7a19c7ecfe
at src/libstd/sys_common/backtrace.rs:76
3: 0x7fd1648a2b04 - <std::sys_common::backtrace::_print::DisplayBacktrace as core::fmt::Display>::fmt::h356b821d79ead967
at src/libstd/sys_common/backtrace.rs:60
4: 0x7fd1648db14c - core::fmt::write::haa7725ecee710b81
at src/libcore/fmt/mod.rs:1030
5: 0x7fd164896d27 - std::io::Write::write_fmt::h677b5c4b9e48abad
at src/libstd/io/mod.rs:1412
6: 0x7fd1648a7335 - std::sys_common::backtrace::_print::hdc27c79deedd181f
at src/libstd/sys_common/backtrace.rs:64
7: 0x7fd1648a7335 - std::sys_common::backtrace::print::h0bb3a218c68a1b38
at src/libstd/sys_common/backtrace.rs:49
8: 0x7fd1648a7335 - std::panicking::default_hook::{{closure}}::h9160d687734b4c2f
at src/libstd/panicking.rs:196
9: 0x7fd1648a7026 - std::panicking::default_hook::h298b832ea14df44f
at src/libstd/panicking.rs:210
10: 0x7fd164ddcf63 - rustc_driver::report_ice::hce2a6b74528a3743
11: 0x7fd1648a7b1c - std::panicking::rust_panic_with_hook::h7c6406c2637b219f
at src/libstd/panicking.rs:477
12: 0x7fd1648a75d2 - std::panicking::continue_panic_fmt::h8e5e175fd262b206
at src/libstd/panicking.rs:380
13: 0x7fd1648a74c6 - rust_begin_unwind
at src/libstd/panicking.rs:307
14: 0x7fd1648d4ada - core::panicking::panic_fmt::h864c751c34920017
at src/libcore/panicking.rs:85
15: 0x7fd1648d5216 - core::slice::slice_index_len_fail::h5579666bd5db7c44
at src/libcore/slice/mod.rs:2583
16: 0x7fd166856244 - <measureme::mmap_serialization_sink::MmapSerializationSink as core::ops::drop::Drop>::drop::hfdb5ca3a69a780c5
17: 0x7fd164de1948 - alloc::sync::Arc<T>::drop_slow::h1e9b9b3bcf38be1c
18: 0x7fd164de1c3e - alloc::sync::Arc<T>::drop_slow::h2d5c08f8d08065e0
19: 0x7fd164df4f15 - <alloc::rc::Rc<T> as core::ops::drop::Drop>::drop::h226d778706bf29d2
20: 0x7fd164db01bc - core::ptr::real_drop_in_place::h78ba5d2f1d1836e8
21: 0x7fd164daa59c - rustc_interface::interface::run_compiler_in_existing_thread_pool::hf9ebc0f8adadba07
[...]

rust-timer · 2019-09-22T17:28:07Z

Finished benchmarking try commit 64687bb, comparison URL.

Mark-Simulacrum · 2019-09-22T19:08:16Z

Failures are unrelated to this PR; they're due to newly introduced self-profile functionality for rustc.

Mark-Simulacrum · 2019-09-23T11:53:02Z

Looks like ~no difference which is good -- more general code is probably better. It's probably more likely this would've had a difference with const generics being used in the set of crate benchmarks.

@oli-obk I believe this should be ready to merge.

oli-obk · 2019-09-23T13:42:16Z

@bors r+

bors · 2019-09-23T13:42:17Z

📌 Commit 9c62117 has been approved by oli-obk

Centril · 2019-09-23T16:58:47Z

@bors rollup=never

Centril · 2019-09-28T20:11:36Z

@bors p=3

bors · 2019-09-29T13:41:35Z

⌛ Testing commit 9c62117 with merge eea282a8233db92be72b4de43c575023a00f07f8...

rust-highfive · 2019-09-29T13:52:48Z

The job x86_64-gnu-debug of your PR failed (pretty log, raw log). Through arcane magic we have determined that the following fragments from the build log may contain information about the problem.

Click to expand the log.

2019-09-29T13:50:20.2613831Z == clock drift check ==
2019-09-29T13:50:20.2613979Z local time: Sun Sep 29 13:50:19 UTC 2019
2019-09-29T13:50:20.2614140Z network time: Sun, 29 Sep 2019 13:50:19 GMT
2019-09-29T13:50:20.2614297Z == end clock drift check ==
2019-09-29T13:50:20.7406115Z ##[error]Bash exited with code '1'.
2019-09-29T13:50:20.7458169Z ##[section]Starting: Upload CPU usage statistics
2019-09-29T13:50:20.7461403Z ==============================================================================
2019-09-29T13:50:20.7461479Z Task : Bash
2019-09-29T13:50:20.7461552Z Description : Run a Bash script on macOS, Linux, or Windows

I'm a bot! I can only do what humans tell me to, so if this was not helpful or you have suggestions for improvements, please ping or otherwise contact @TimNN. (Feature Requests)

bors · 2019-09-29T13:53:03Z

💔 Test failed - checks-azure

This specifically targets match checking, but is possibly more widely useful as well. In code with large, single-value match statements, we were previously spending a lot of time running layout_of for the primitive types (integers, chars) -- which is essentially useless. This optimizes the code to avoid those query calls by directly obtaining the size for these types, when possible. It may be worth considering adding a `size_of` query in the future which might be far faster, especially if specialized for "const" cases -- match arms being the most obvious example. It's possibly such a function would benefit from *not* being a query as well, since it's trivially evaluatable from the sty for many cases whereas a query needs to hash the input and such.

Mark-Simulacrum · 2019-09-29T22:14:01Z

@bors r=oli-obk

bors · 2019-09-29T22:14:02Z

📌 Commit 06c6e75 has been approved by oli-obk

bors · 2019-09-29T22:21:49Z

⌛ Testing commit 06c6e75 with merge d16ee89...

Optimize try_eval_bits to avoid layout queries This specifically targets match checking, but is possibly more widely useful as well. In code with large, single-value match statements, we were previously spending a lot of time running layout_of for the primitive types (integers, chars) -- which is essentially useless. This optimizes the code to avoid those query calls by directly obtaining the size for these types, when possible. It may be worth considering adding a `size_of` query in the future which might be far faster, especially if specialized for "const" cases -- match arms being the most obvious example. It's possibly such a function would benefit from *not* being a query as well, since it's trivially evaluatable from the sty for many cases whereas a query needs to hash the input and such.

bors · 2019-09-30T02:09:36Z

☀️ Test successful - checks-azure
Approved by: oli-obk
Pushing d16ee89 to master...

The `if let Some(val) = value.try_eval_bits(...)` branch in `from_const()` is very hot for the `unicode_normalization` benchmark. This commit introduces a special-case alternative for scalars that avoids `try_eval_bits()` and all the functions it calls (`Const::eval()`, `ConstValue::try_to_bits()`, `ConstValue::try_to_scalar()`, and `Scalar::to_bits()`), instead extracting the result immediately. The type and value checking done by `Scalar::to_bits()` is replicated by moving it into a new function `Scalar::check_raw()` and using that new function in the special case. PR rust-lang#64673 introduced some special-case handling of scalar types in `Const::try_eval_bits()`. This handling is now moved out of that function into the new `IntRange::integral_size_and_signed_bias` function. This commit reduces the instruction count for `unicode_normalization-check-clean` by about 10%.

rust-highfive assigned Centril Sep 21, 2019

rust-highfive added the S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. label Sep 21, 2019

rust-highfive assigned oli-obk and unassigned Centril Sep 21, 2019

oli-obk reviewed Sep 22, 2019
View reviewed changes

Comment threadsrc/librustc_mir/hair/pattern/_match.rs Outdated

oli-obk reviewed Sep 22, 2019
View reviewed changes

Comment threadsrc/librustc_mir/hair/pattern/_match.rs Outdated

Mark-Simulacrum force-pushed the opt-match-ck branch from 511352c to 9c62117Compare September 22, 2019 13:00

Mark-Simulacrum changed the title ~~Optimize match checking to avoid layout queries~~Optimize try_eval_bits to avoid layout queries Sep 22, 2019

bors removed the S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. label Sep 23, 2019

bors added the S-waiting-on-bors Status: Waiting on bors to run and complete tests. Bors will change the label on completion. label Sep 23, 2019

bors added S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. and removed S-waiting-on-bors Status: Waiting on bors to run and complete tests. Bors will change the label on completion. labels Sep 29, 2019

Mark-Simulacrum force-pushed the opt-match-ck branch from 9c62117 to 06c6e75Compare September 29, 2019 22:01

bors added S-waiting-on-bors Status: Waiting on bors to run and complete tests. Bors will change the label on completion. and removed S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. labels Sep 29, 2019

bors added the merged-by-bors This PR was explicitly merged by bors. label Sep 30, 2019

bors merged commit 06c6e75 into rust-lang:masterSep 30, 2019

bors mentioned this pull request Sep 30, 2019
Create compilation target versions of ::alloc::Layout#64299
Closed

nnethercote mentioned this pull request Oct 4, 2019
Optimize integral pattern matching #65089
Merged

Mark-Simulacrum deleted the opt-match-ck branch October 8, 2019 21:52

nnethercote mentioned this pull request Oct 9, 2019
Long compilation time of crate "unicode-normalization" #55528
Closed

Uh oh!

Uh oh!

Conversation

Mark-Simulacrum commented Sep 21, 2019• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Mark-Simulacrum commented Sep 21, 2019

Uh oh!

rust-timer commented Sep 21, 2019

Uh oh!

bors commented Sep 21, 2019

Uh oh!

Centril commented Sep 21, 2019

Uh oh!

bors commented Sep 22, 2019

Uh oh!

rust-timer commented Sep 22, 2019

Uh oh!

rust-timer commented Sep 22, 2019

Uh oh!

bjorn3 commented Sep 22, 2019

Uh oh!

Uh oh!

Uh oh!

Mark-Simulacrum commented Sep 22, 2019

Uh oh!

rust-timer commented Sep 22, 2019

Uh oh!

bors commented Sep 22, 2019

Uh oh!

bors commented Sep 22, 2019

Uh oh!

rust-timer commented Sep 22, 2019

Uh oh!

bjorn3 commented Sep 22, 2019

Uh oh!

rust-timer commented Sep 22, 2019

Uh oh!

Mark-Simulacrum commented Sep 22, 2019

Uh oh!

Mark-Simulacrum commented Sep 23, 2019

Uh oh!

oli-obk commented Sep 23, 2019

Uh oh!

bors commented Sep 23, 2019

Uh oh!

Centril commented Sep 23, 2019

Uh oh!

Centril commented Sep 28, 2019

Uh oh!

bors commented Sep 29, 2019

Uh oh!

rust-highfive commented Sep 29, 2019

Uh oh!

bors commented Sep 29, 2019

Uh oh!

Mark-Simulacrum commented Sep 29, 2019

Uh oh!

bors commented Sep 29, 2019

Uh oh!

bors commented Sep 29, 2019

Uh oh!

bors commented Sep 30, 2019

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

7 participants

Mark-Simulacrum commented Sep 21, 2019•
edited
Loading