Add f16 and f128 as simd types in LLVM

[tgross35]

@sayantn is working on adding SIMD for f16 and hitting the FloatingPointVector error. This should fix it and unblock adding support for simd_fma and simd_fabs in stdarch.

SIMD work: rust-lang/stdarch#1605

[rustbot]

r? @cjgillot

rustbot has assigned @cjgillot.
They will have a look at your PR within the next two weeks and either review your PR or reassign to another reviewer.

Use r? to explicitly pick a reviewer

[tgross35]

@rust-lang/wg-gcc-backend I am not sure how GCC types are named, what should the equivalent update be here?

https://github.com/tgross35/rust/blob/506512391b1a75ae450d36c9420978402a91abcc/compiler/rustc_codegen_gcc/src/intrinsic/simd.rs#L594-L605

[GuillaumeGomez]

Support was added very recently on our side: https://github.com/antoyo/rustc_codegen_gcc/pull/17/files

[tgross35]

With this I am assuming that e.g. llvm.fabs.v32f16/llvm.fabs.v4f128 are the correct LLVM names, @sayantn any chance you could double check that?

@GuillaumeGomez it doesn't look like the simd support landed there, do you know what the appropriate suffixes for the intrinsics are for GCC? I am not sure where this would be documented. (should be okay to leave this unimplemented in GCC for now if there isn't a straightforward answer)

[sayantn]

@tgross35 that is the correct name as in LangRef. The GCC suffix is l for __float128. Could not find info about __float16

[rustbot]

Some changes occurred in compiler/rustc_codegen_gcc

cc @antoyo, @GuillaumeGomez

[tgross35]

Great thanks for confirming, I did the GCC updates.

[tgross35]

Ah, I see you edited your comment about GCC 😆 I just dropped that patch, I will take a look but we can just add it later if it's not well supported yet.

[tgross35]

🤔 failure looks spurious, will force push something to retry

[Amanieu]

@bors r+

[bors]

📌 Commit 04caa55 has been approved by Amanieu

It is now in the queue for this repository.

[bors]

⌛ Testing commit 04caa55 with merge bcf1f6d...

[bors]

☀️ Test successful - checks-actions
Approved by: Amanieu
Pushing bcf1f6d to master...

[rust-timer]

Finished benchmarking commit (bcf1f6d): comparison URL.

Overall result: no relevant changes - no action needed

@rustbot label: -perf-regression

Instruction count

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

Max RSS (memory usage)

Results (primary 2.0%)

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)	2.0%	[2.0%, 2.0%]	1
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	2.0%	[2.0%, 2.0%]	1

Cycles

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

Binary size

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

Bootstrap: 704.358s -> 703.929s (-0.06%)
Artifact size: 328.86 MiB -> 328.72 MiB (-0.04%)