Subtree update cg gcc 2024 03 05 #122042

GuillaumeGomez · 2024-03-05T19:00:39Z

Reopening of #121390.

Implements lane-local byte swapping through vector shuffles. While this is more setup than non-vector shuffles, this implementation can shuffle multiple integers concurrently. Signed-off-by: Andy Sadler <[email protected]>

Signed-off-by: Andy Sadler <[email protected]>

The simd intrinsic handler was delegating implementation of `simd_frem` to `Builder::frem`, which wasn't able to handle vector-typed inputs. To fix this, teach this method how to handle vector inputs. Signed-off-by: Andy Sadler <[email protected]>

If we're running against a patched libgccjit, use an algorithm similar to what LLVM uses for this intrinsic. Otherwise, fallback to a per-element bitreverse. Signed-off-by: Andy Sadler <[email protected]>

Signed-off-by: Andy Sadler <[email protected]>

gcc_not would panic upon encountering a vector type, which is not what we want here. Signed-off-by: Andy Sadler <[email protected]>

This test now passes when tested with a patched libgccjit. However, due to [some compiler bugs][1], we can't enable this for non-patched libgccjit yet. [1]: https://github.com/sadlerap/rustc_codegen_gcc/actions/runs/6820180639/job/18548672444#step:15:4375 Signed-off-by: Andy Sadler <[email protected]>

…ctions Fix build instructions

`sess` is a terribly misleading name for a `Handler`! This confused me for a bit.

…g to it

…back` function

bors · 2024-03-08T12:56:19Z

☔ The latest upstream changes (presumably #122182) made this pull request unmergeable. Please resolve the merge conflicts.

Mark-Simulacrum · 2024-03-09T15:05:11Z

Is there a chance we could land the 250+ commits in this PR without the infra changes somehow? I'd much prefer to review a smaller PR. If that means we temporarily disable testing or something on CI, that would be preferable in my eyes to trying to review and figure out the much larger PR (also less bitrotty).

Mark-Simulacrum

OK, having looked at those commits they seem minimal enough. I'm OK with this moving forward. Caching should be automatic as you're adding to the Docker builds which are always cached.

r=me on the infra changes.

src/ci/docker/host-x86_64/dist-x86_64-linux/libgccjit.version

GuillaumeGomez · 2024-03-10T01:30:46Z

Fixed the merge conflict and applied Mark's suggestion. Considering it's very conflict-prone, adding a higher priority.

@bors r=MarkSimulacrum p=1 rollup=never

bors · 2024-03-10T01:30:49Z

📌 Commit c5c6729 has been approved by MarkSimulacrum

It is now in the queue for this repository.

bors · 2024-03-10T03:40:36Z

⌛ Testing commit c5c6729 with merge 5bc7b9a...

bors · 2024-03-10T06:06:30Z

☀️ Test successful - checks-actions
Approved by: MarkSimulacrum
Pushing 5bc7b9a to master...

rust-timer · 2024-03-10T07:51:48Z

Finished benchmarking commit (5bc7b9a): comparison URL.

Overall result: ❌ regressions - no action needed

@rustbot label: -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
Regressions ❌ (secondary)	1.6%	[1.6%, 1.6%]	1
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	-	-	0

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)	-	-	0
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-2.7%	[-8.6%, -1.2%]	48
Improvements ✅ (secondary)	-3.1%	[-6.0%, -0.7%]	119
All ❌✅ (primary)	-2.7%	[-8.6%, -1.2%]	48

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)	2.1%	[2.1%, 2.1%]	1
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	-	-	0

Binary size

This benchmark run did not return any relevant results for this metric.

Bootstrap: 645.087s -> 647.582s (0.39%)
Artifact size: 309.96 MiB -> 310.00 MiB (0.02%)

sadlerap and others added 30 commits November 28, 2023 21:25

implement simd_bswap intrinsic

cc7c9be

Implements lane-local byte swapping through vector shuffles. While this is more setup than non-vector shuffles, this implementation can shuffle multiple integers concurrently. Signed-off-by: Andy Sadler <[email protected]>

remove generic-bswap-byte from failing test list

6d13f94

Signed-off-by: Andy Sadler <[email protected]>

fix simd_frem intrinsic implementation

70586a2

The simd intrinsic handler was delegating implementation of `simd_frem` to `Builder::frem`, which wasn't able to handle vector-typed inputs. To fix this, teach this method how to handle vector inputs. Signed-off-by: Andy Sadler <[email protected]>

impl simd_bitreverse intrinsic

8d42a82

If we're running against a patched libgccjit, use an algorithm similar to what LLVM uses for this intrinsic. Otherwise, fallback to a per-element bitreverse. Signed-off-by: Andy Sadler <[email protected]>

impl simd_ctlz/simd_cttz intrinsic

03e11a2

Signed-off-by: Andy Sadler <[email protected]>

fix simd_neg implementation for ints

3a22132

gcc_not would panic upon encountering a vector type, which is not what we want here. Signed-off-by: Andy Sadler <[email protected]>

Fix build instructions

bb4d0be

Merge pull request rust-lang#391 from GuillaumeGomez/fix-build-instru…

0a67e9c

…ctions Fix build instructions

Rename some arguments.

f9a228d

`sess` is a terribly misleading name for a `Handler`! This confused me for a bit.

Allow rustfmt to run on build_system

04f32f2

Rustify test.sh

7b76ac4

Remove test.sh, config.sh and all calls and documentation pointin…

84ca4f5

…g to it

Display stdout and stderr if a command failed to run

d3e14a4

Fix invalid path in build_sysroot_inner

8cc024c

Add missing --build-sysroot option

694a80d

Correctly handle channel in config

c27fe3e

Clone rust repository before modifying it

9d104a0

Only read rust test files

87c284c

Replace xargs command with pure Rust

23c97b5

Remove newline for llvm FileCheck binary path

673661d

Correctly pass cfg option

d793f80

Correctly pass toolchain to cargo command

4bed89f

Use the correct folder when deleting rust UI tests

3c6bae7

Add missing code comment

7013ecc

Set RUSTDOCFLAGS environment variable in `run_cargo_command_with_call…

ad1d541

…back` function

Correctly set --cap-lints when running regex tests

7d71b87

Fix build_sysroot by adding missing RUSTFLAGS environment variable

970b2c7

Remove --target option

ff04316

Show command which failed

53b2759

Mark-Simulacrum reviewed Mar 9, 2024

View reviewed changes

src/ci/docker/host-x86_64/dist-x86_64-linux/libgccjit.version Outdated Show resolved Hide resolved

Merge remote-tracking branch 'upstream/master' into HEAD

2b5b43e

This comment has been minimized.

Sign in to view

GuillaumeGomez force-pushed the subtree-update_cg_gcc_2024-03-05 branch from 62b838a to b85b11a Compare March 9, 2024 17:21

This comment has been minimized.

Sign in to view

GuillaumeGomez force-pushed the subtree-update_cg_gcc_2024-03-05 branch from b85b11a to 2d4c379 Compare March 9, 2024 17:40

This comment has been minimized.

Sign in to view

GuillaumeGomez added 2 commits March 10, 2024 01:01

Fix cg_gcc merge

1d058a0

Use full commit hash and remove libgccjit.version file

c5c6729

GuillaumeGomez force-pushed the subtree-update_cg_gcc_2024-03-05 branch from 2d4c379 to c5c6729 Compare March 10, 2024 00:01

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 Mar 10, 2024

bors added the merged-by-bors This PR was explicitly merged by bors. label Mar 10, 2024

bors merged commit 5bc7b9a into rust-lang:master Mar 10, 2024
12 checks passed

rustbot added this to the 1.78.0 milestone Mar 10, 2024

GuillaumeGomez deleted the subtree-update_cg_gcc_2024-03-05 branch March 10, 2024 10:51

This was referenced Mar 14, 2024

Distribute it with rustup rust-lang/rustc_codegen_gcc#465

Open

Greatly reduce GCC build logs #122496

Merged

RalfJung mentioned this pull request Mar 15, 2024

GCC builds are taking 30 min in PR CI #122546

Open

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Subtree update cg gcc 2024 03 05 #122042

Subtree update cg gcc 2024 03 05 #122042

GuillaumeGomez commented Mar 5, 2024

bors commented Mar 8, 2024

Mark-Simulacrum commented Mar 9, 2024

Mark-Simulacrum left a comment

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

GuillaumeGomez commented Mar 10, 2024

bors commented Mar 10, 2024

bors commented Mar 10, 2024

bors commented Mar 10, 2024

rust-timer commented Mar 10, 2024

Subtree update cg gcc 2024 03 05 #122042

Subtree update cg gcc 2024 03 05 #122042

Conversation

GuillaumeGomez commented Mar 5, 2024

bors commented Mar 8, 2024

Mark-Simulacrum commented Mar 9, 2024

Mark-Simulacrum left a comment

Choose a reason for hiding this comment

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

GuillaumeGomez commented Mar 10, 2024

bors commented Mar 10, 2024

bors commented Mar 10, 2024

bors commented Mar 10, 2024

rust-timer commented Mar 10, 2024

Overall result: ❌ regressions - no action needed

Instruction count

Max RSS (memory usage)

Cycles

Binary size