Implement constant propagation on top of MIR SSA analysis #116012

cjgillot · 2023-09-20T22:04:55Z

This implements the idea I proposed in #110719 (comment)

The value numbering "GVN" pass formulates each rvalue that appears in MIR with an abstract form (the Value enum), and assigns an integer VnIndex to each. This abstract form can be used to deduplicate values, reusing an earlier local that holds the same value instead of recomputing. This part is proposed in #109597.

From this abstract representation, we can perform more involved simplifications, for example in #111344.

With the abstract representation Value, we can also attempt to evaluate each to a constant using the interpreter. This builds a VnIndex -> OpTy map. From this map, we can opportunistically replace an operand or a rvalue with a constant if their value has an associated OpTy.

The most relevant commit is Evaluated computed values to constants."

r? @oli-obk

rustbot · 2023-09-20T22:05:06Z

This PR changes MIR

cc @oli-obk, @RalfJung, @JakobDegen, @davidtwco, @celinval, @vakaras

Some changes occurred to the CTFE / Miri engine

cc @rust-lang/miri

The Miri subtree was changed

cc @rust-lang/miri

Some changes occurred to MIR optimizations

cc @rust-lang/wg-mir-opt

bors · 2023-09-21T15:04:05Z

☔ The latest upstream changes (presumably #116027) made this pull request unmergeable. Please resolve the merge conflicts.

rustbot · 2023-09-21T17:11:45Z

This PR changes MIR

cc @oli-obk, @RalfJung, @JakobDegen, @davidtwco, @celinval, @vakaras

The Miri subtree was changed

cc @rust-lang/miri

Some changes occurred to the CTFE / Miri engine

cc @rust-lang/miri

Some changes occurred to MIR optimizations

cc @rust-lang/wg-mir-opt

bors · 2023-09-22T18:08:01Z

☔ The latest upstream changes (presumably #115696) made this pull request unmergeable. Please resolve the merge conflicts.

cjgillot · 2023-09-23T11:22:06Z

@bors try @rust-timer queue

bors · 2023-09-23T11:22:17Z

⌛ Trying commit 81d5ac4 with merge 7b24347...

Implement constant propagation on top of MIR SSA analysis This implements the idea I proposed in rust-lang#110719 (comment) Based on rust-lang#109597 The value numbering "GVN" pass formulates each rvalue that appears in MIR with an abstract form (the `Value` enum), and assigns an integer `VnIndex` to each. This abstract form can be used to deduplicate values, reusing an earlier local that holds the same value instead of recomputing. This part is proposed in rust-lang#109597. From this abstract representation, we can perform more involved simplifications, for example in rust-lang#111344. With the abstract representation `Value`, we can also attempt to evaluate each to a constant using the interpreter. This builds a `VnIndex -> OpTy` map. From this map, we can opportunistically replace an operand or a rvalue with a constant if their value has an associated `OpTy`. The most relevant commit is [Evaluated computed values to constants.](rust-lang@2767c49)" r? `@oli-obk`

bors · 2023-09-23T12:33:12Z

☀️ Try build successful - checks-actions
Build commit: 7b24347 (7b2434719a7e59a9df73c635255e0ca739d29a4c)

bors · 2023-09-23T12:33:12Z

☀️ Try build successful - checks-actions
Build commit: 7b24347 (7b2434719a7e59a9df73c635255e0ca739d29a4c)

rust-timer · 2023-09-23T14:33:22Z

Finished benchmarking commit (7b24347): comparison URL.

Overall result: ❌✅ regressions and improvements - ACTION NEEDED

Benchmarking this pull request likely means that it is perf-sensitive, so we're automatically marking it as not fit for rolling up. While you can manually mark this PR as fit for rollup, we strongly recommend not doing so since this PR may lead to changes in compiler perf.

Next Steps: If you can justify the regressions found in this try perf run, please indicate this with @rustbot label: +perf-regression-triaged along with sufficient written justification. If you cannot justify the regressions please fix the regressions and do another perf run. If the next run shows neutral or positive results, the label will be automatically removed.

@bors rollup=never
@rustbot label: -S-waiting-on-perf +perf-regression

Instruction count

This is a highly reliable metric that was used to determine the overall result at the top of this comment.

	mean	range	count
Regressions ❌ (primary)	0.4%	[0.4%, 0.4%]	1
Regressions ❌ (secondary)	2.2%	[1.0%, 2.9%]	4
Improvements ✅ (primary)	-0.6%	[-1.6%, -0.3%]	13
Improvements ✅ (secondary)	-0.4%	[-0.9%, -0.2%]	13
All ❌✅ (primary)	-0.6%	[-1.6%, 0.4%]	14

Max RSS (memory usage)

Results

This is a less reliable metric that may be of interest but was not used to determine the overall result at the top of this comment.

	mean	range	count
Regressions ❌ (primary)	1.9%	[0.2%, 3.0%]	3
Regressions ❌ (secondary)	5.0%	[2.5%, 6.4%]	8
Improvements ✅ (primary)	-3.1%	[-3.5%, -2.7%]	3
Improvements ✅ (secondary)	-3.1%	[-5.1%, -0.8%]	5
All ❌✅ (primary)	-0.6%	[-3.5%, 3.0%]	6

Cycles

Results

This is a less reliable metric that may be of interest but was not used to determine the overall result at the top of this comment.

	mean	range	count
Regressions ❌ (primary)	-	-	0
Regressions ❌ (secondary)	3.3%	[2.3%, 4.5%]	4
Improvements ✅ (primary)	-1.3%	[-1.6%, -1.0%]	2
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	-1.3%	[-1.6%, -1.0%]	2

Binary size

Results

This is a less reliable metric that may be of interest but was not used to determine the overall result at the top of this comment.

	mean	range	count
Regressions ❌ (primary)	0.3%	[0.0%, 1.0%]	58
Regressions ❌ (secondary)	0.1%	[0.0%, 0.4%]	17
Improvements ✅ (primary)	-0.6%	[-4.8%, -0.0%]	32
Improvements ✅ (secondary)	-0.6%	[-2.4%, -0.0%]	19
All ❌✅ (primary)	-0.0%	[-4.8%, 1.0%]	90

Bootstrap: 634.28s -> 634.677s (0.06%)
Artifact size: 317.90 MiB -> 317.85 MiB (-0.02%)

rust-log-analyzer · 2023-12-30T01:45:09Z

A job failed! Check out the build log: (web) (plain)

Click to see the possible cause of the failure (guessed by this bot)

saethlin · 2023-12-30T01:51:58Z

Looks to me like the timeout was in uploading to s3

@bors retry

bors · 2023-12-30T03:46:02Z

⌛ Testing commit 6d7f063 with merge 8d76d07...

bors · 2023-12-30T05:42:32Z

☀️ Test successful - checks-actions
Approved by: oli-obk
Pushing 8d76d07 to master...

rust-timer · 2023-12-30T07:03:11Z

Finished benchmarking commit (8d76d07): comparison URL.

Overall result: ❌✅ regressions and improvements - ACTION NEEDED

Next Steps: If you can justify the regressions found in this perf run, please indicate this with @rustbot label: +perf-regression-triaged along with sufficient written justification. If you cannot justify the regressions please open an issue or create a new PR that fixes the regressions, add a comment linking to the newly created issue or PR, and then add the perf-regression-triaged label to this PR.

@rustbot label: +perf-regression
cc @rust-lang/wg-compiler-performance

Instruction count

This is a highly reliable metric that was used to determine the overall result at the top of this comment.

	mean	range	count
Regressions ❌ (primary)	0.7%	[0.2%, 1.9%]	10
Regressions ❌ (secondary)	0.6%	[0.2%, 1.1%]	15
Improvements ✅ (primary)	-0.6%	[-2.6%, -0.2%]	26
Improvements ✅ (secondary)	-31.2%	[-62.1%, -0.2%]	8
All ❌✅ (primary)	-0.2%	[-2.6%, 1.9%]	36

Max RSS (memory usage)

Results

This is a less reliable metric that may be of interest but was not used to determine the overall result at the top of this comment.

	mean	range	count
Regressions ❌ (primary)	3.1%	[0.4%, 8.5%]	8
Regressions ❌ (secondary)	4.2%	[3.3%, 5.9%]	4
Improvements ✅ (primary)	-2.6%	[-5.4%, -0.5%]	10
Improvements ✅ (secondary)	-8.9%	[-43.4%, -0.7%]	23
All ❌✅ (primary)	-0.1%	[-5.4%, 8.5%]	18

Cycles

Results

This is a less reliable metric that may be of interest but was not used to determine the overall result at the top of this comment.

	mean	range	count
Regressions ❌ (primary)	-	-	0
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-1.0%	[-1.6%, -0.5%]	5
Improvements ✅ (secondary)	-56.6%	[-70.9%, -8.8%]	5
All ❌✅ (primary)	-1.0%	[-1.6%, -0.5%]	5

Binary size

Results

This is a less reliable metric that may be of interest but was not used to determine the overall result at the top of this comment.

	mean	range	count
Regressions ❌ (primary)	0.3%	[0.0%, 1.6%]	59
Regressions ❌ (secondary)	0.2%	[0.0%, 0.4%]	12
Improvements ✅ (primary)	-0.5%	[-4.0%, -0.0%]	28
Improvements ✅ (secondary)	-2.3%	[-5.3%, -0.0%]	15
All ❌✅ (primary)	0.1%	[-4.0%, 1.6%]	87

Bootstrap: 670.291s -> 668.991s (-0.19%)
Artifact size: 312.00 MiB -> 312.11 MiB (0.03%)

rylev · 2024-01-02T12:55:44Z

It seems given the positives for this change and the fact that improvements outweigh regressions, we can call this triaged.

@rustbot label: +perf-regression-triaged

Fold arithmetic identities in GVN Extracted from rust-lang#111344 This PR implements a few arithmetic folds for unary and binary operations. This should take care of the missed optimizations introduced by rust-lang#116012.

Fold arithmetic identities in GVN Extracted from rust-lang/rust#111344 This PR implements a few arithmetic folds for unary and binary operations. This should take care of the missed optimizations introduced by rust-lang/rust#116012.

RalfJung · 2024-01-24T16:25:39Z

I'm only just seeing this... this is amazing, old const-prop is gone. :) Thanks a ton @cjgillot!

…i-obk Remove const_prop.rs Removed const_prop.rs and moved its contents to const_prop_lint.rs and dataflow_const_prop.rs where they are used. const_prop.rs does not actually do any const propagation any more. After rust-lang#116012 all it contains is code that is used by const_prop_lint.rs and one macro that is used by dataflow_const_prop.rs. So it made sense to just move it to those two places and remove this file.

Rollup merge of rust-lang#121260 - gurry:constprop-lint-cleanup, r=oli-obk Remove const_prop.rs Removed const_prop.rs and moved its contents to const_prop_lint.rs and dataflow_const_prop.rs where they are used. const_prop.rs does not actually do any const propagation any more. After rust-lang#116012 all it contains is code that is used by const_prop_lint.rs and one macro that is used by dataflow_const_prop.rs. So it made sense to just move it to those two places and remove this file.

Add FileCheck annotations to const_prop tests Unblocks rust-lang/rust#116012 Advances rust-lang/rust#116971

rustbot assigned oli-obk Sep 20, 2023

rustbot added S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. T-compiler Relevant to the compiler team, which will review and decide on the PR/issue. labels Sep 20, 2023