-
Notifications
You must be signed in to change notification settings - Fork 13.3k
Remove avx512dq
and avx512vl
implication for avx512fp16
#140389
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Conversation
|
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
cc @a4lg there seems to be some doc errors in the PS: I am quite confused how they weren't caught in the stdarch CI |
On hindsight, this seems like a bug in rustdoc with cc @rust-lang/rustdoc @rust-lang/wg-gcc-backend edit: not spurious |
This comment has been minimized.
This comment has been minimized.
@sayantn It seems the Linkcheck tool on the CI does its own job and the error itself seems valid. However, multiple references to a single footnote is mandatory for simplicity of the new macro documentation with the platform guide. If you prefer the short term solution to pass CI, you may revert rust-lang/stdarch#1779. |
I created a rustdoc PR #140434 (implementing it was easier than I thought) to enable adopting rust-lang/stdarch#1779 in the future. I'm not sure whether adoption of this PR will result in immediate success (considering the default stage for building docs (stage 0), it might not be immediate) but I hope this is accepted. |
082ec44
to
dd281cb
Compare
This comment has been minimized.
This comment has been minimized.
I opened a PR rust-lang/stdarch#1792 that will revert rust-lang/stdarch#1779 for now. If this PR is applied, stdarch should be safely merged into the master branch. Even if #140434 is merged now, because the CI failure says this is caused by the stage0 compiler, we have to wait the next beta to merge rust-lang/stdarch#1779. |
Will the recent changes to the bootstrapping process help this? I am talking about #119899 |
I am trying this out with rust-lang/stdarch#1792 |
I'm glad to see that this change is working! (I miss carefully designed tables and guides though) |
Nice, I will ask someone to start a try job to further make sure nothing is breaking update: try runs are not needed, as most tests are already done. we don't need to do crater runs because this is (still) unstable
We can re-update stdarch with your PR after the Rustdoc fix is merged, and do another stdarch update. In the meantime, could you re-open that PR so that we don't forget about it? Thanks |
@sayantn I filed a new issue on stdarch instead (rust-lang/stdarch#1793). |
@bors r+ rollup=iffy |
Remove `avx512dq` and `avx512vl` implication for `avx512fp16` According to Intel, `avx512fp16` requires only `avx512bw`, but LLVM also enables `avx512vl` and `avx512dq` when `avx512fp16` is active. This is relic code, and will be fixed in LLVM soon. We should remove this from Rust too asap, especially before the stabilization of AVX512 Related: - llvm/llvm-project#136209 - rust-lang#138940 - rust-lang/stdarch#1781 - rust-lang#111137 `@rustbot` label O-x86_64 O-x86_32 A-SIMD A-target-feature T-compiler -T-libs r? `@Amanieu` **Update: the LLVM fix has been merged** cc `@rust-lang/wg-llvm` will it be possible to update the rustc llvm version to something after llvm/llvm-project#137450
Rollup of 12 pull requests Successful merges: - rust-lang#134034 (handle paren in macro expand for let-init-else expr) - rust-lang#137474 (pretty-print: Print shebang at the top of the output) - rust-lang#138872 (rustc_target: RISC-V `Zfinx` is incompatible with `{ILP32,LP64}[FD]` ABIs) - rust-lang#139046 (Improve `Lifetime::suggestion`) - rust-lang#139206 (std: use the address of `errno` to identify threads in `unique_thread_exit`) - rust-lang#139608 (Clarify `async` block behaviour) - rust-lang#139847 (Delegate to inner `vec::IntoIter` from `env::ArgsOs`) - rust-lang#140159 (Avoid redundant WTF-8 checks in `PathBuf`) - rust-lang#140197 (Document breaking out of a named code block) - rust-lang#140389 (Remove `avx512dq` and `avx512vl` implication for `avx512fp16`) - rust-lang#140430 (Improve test coverage of HIR pretty printing.) - rust-lang#140507 (rustc_target: RISC-V: feature addition batch 3) r? `@ghost` `@rustbot` modify labels: rollup
Rollup merge of rust-lang#140389 - sayantn:avx512fp16, r=Amanieu Remove `avx512dq` and `avx512vl` implication for `avx512fp16` According to Intel, `avx512fp16` requires only `avx512bw`, but LLVM also enables `avx512vl` and `avx512dq` when `avx512fp16` is active. This is relic code, and will be fixed in LLVM soon. We should remove this from Rust too asap, especially before the stabilization of AVX512 Related: - llvm/llvm-project#136209 - rust-lang#138940 - rust-lang/stdarch#1781 - rust-lang#111137 ``@rustbot`` label O-x86_64 O-x86_32 A-SIMD A-target-feature T-compiler -T-libs r? ``@Amanieu`` **Update: the LLVM fix has been merged** cc ``@rust-lang/wg-llvm`` will it be possible to update the rustc llvm version to something after llvm/llvm-project#137450
According to Intel,
avx512fp16
requires onlyavx512bw
, but LLVM also enablesavx512vl
andavx512dq
whenavx512fp16
is active. This is relic code, and will be fixed in LLVM soon. We should remove this from Rust too asap, especially before the stabilization of AVX512Related:
fma
andf16c
foravx512f
instd_detect
stdarch#1781@rustbot label O-x86_64 O-x86_32 A-SIMD A-target-feature T-compiler -T-libs
r? @Amanieu
Update: the LLVM fix has been merged
cc @rust-lang/wg-llvm will it be possible to update the rustc llvm version to something after llvm/llvm-project#137450