Stabilize x86/x86_64 SIMD #49664

alexcrichton · 2018-04-04T19:44:12Z

This commit stabilizes the SIMD in Rust for the x86/x86_64 platforms. Notably
this commit is stabilizing:

The std::arch::{x86, x86_64} modules and the intrinsics contained inside.
The is_x86_feature_detected! macro in the standard library
The #[target_feature(enable = "...")] attribute
The #[cfg(target_feature = "...")] matcher

Stabilization of the module and intrinsics were primarily done in
rust-lang/stdarch#414 and the two attribute stabilizations are done in
this commit. The standard library is also tweaked a bit with the new way that
stdsimd is integrated.

Note that other architectures like std::arch::arm are not stabilized as part
of this commit, they will likely stabilize in the future after they've been
implemented and fleshed out. Similarly the std::simd module is also not being
stabilized in this commit, only std::arch. Finally, nothing related to __m64
is stabilized in this commit either (MMX), only SSE and up types and intrinsics
are stabilized.

Closes #29717
Closes #44839
Closes #48556

alexcrichton · 2018-04-04T19:45:48Z

Note it's not intended that this should merge before the FCP lapses in #48556, but rather this is submitted early to get some more time for review

TimNN · 2018-04-04T21:49:07Z

Your PR failed on Travis. 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.

Resolving deltas: 100% (611497/611497), completed with 4859 local objects.

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.

TimNN · 2018-04-04T23:55:33Z

Your PR failed on Travis. 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.

W: GPG error: https://dl.hhvm.com/ubuntu trusty InRelease: The following signatures couldn't be verified because the public key is not available: NO_PUBKEY B4112585D386EB94
W: The repository 'https://dl.hhvm.com/ubuntu trusty InRelease' is not signed.
W: There is no public key available for the following key IDs:
---
Resolving deltas: 100% (611526/611526), completed with 4863 local objects.
---
[00:00:48] configure: rust.quiet-tests     := True
---
[00:37:53] ..........................................................................i.........................
[00:37:59] .................i..................................................................................
---
[00:38:33] ..........................................................................................i.........
[00:38:40] ..............................................................i.....................................
---
[00:39:30] .............................................i......................................................
---
[00:43:06] .............................i......................................................................
[00:43:20] ..............................................................i.....................................
[00:43:35] ...............................................i....................................................
[00:43:53] ....................................................................................................
[00:44:14] ....................................................................................................
[00:44:34] ....................................................................................................
[00:44:57] .i...............................................................................................i..
[00:45:26] ..........................................................................................test [run-pass] run-pass/mir_heavy_promoted.rs has been running for over 60 seconds
[00:45:31] ..........
[00:45:59] ....................................................................................................
[00:46:32] .............................................................ii.....................................
[00:47:18] ........................i....................................................i.ii...............test [run-pass] run-pass/saturating-float-casts.rs has been running for over 60 seconds
[00:47:20] ....
[00:47:57] .....................................................................................iiiiiii........
---
[00:49:53] ....................................i...............................................................
[00:50:01] ....................................................................................................
[00:50:08] ..................i............................................................ii.iii...............
[00:50:15] ....................................................................................................
[00:50:22] ........i..............................i............................................................
[00:50:29] ....................................................................................................
[00:50:36] ....................i...............................................................................
[00:50:43] ....................................................................................................
[00:50:53] ....................................................................................................
[00:51:03] ....................................................................................................
[00:51:13] ....................................................................................................
[00:51:26] ....................................................................................................
[00:51:34] .............i......................................................................................
[00:51:43] ................i..ii...............................................................................
[00:51:53] ....................................................................................................
[00:52:02] ....................................................................................................
[00:52:11] ..................................................................................i.................
[00:52:22] ............................i.......................................................................
---
[00:52:57] ............................i.......................................................................
[00:52:58] ....................................................................i...............................
[00:52:59] ................i.......................................................
---
[00:53:13] ...........i........................
---
[00:53:41] i...i..ii....i.............ii........iii......i..i...i...ii..i..i..ii.....
---
[00:53:43] i.......i......................i......
---
[00:54:19] iiii.......i..i........i..i.i.............i..........iiii...........i...i..........ii.i.i.......ii..
[00:54:20] ....ii...
---
[01:02:45] ...i................................................................................................
---
[01:04:32] .....................................i..............................................................
[01:04:50] ....................................................................................................
[01:05:08] ..........................................i.........................................................
---
[01:06:28] .........................................................ii.........................................
---
[01:06:54] FFFFFFF.............................................................................................
[01:07:10] ....................................................................................................
[01:07:28] ..............................................................i.....................................
---
[01:10:18] ---- ../stdsimd/coresimd/macros.rs - coresimd::x86::__m128 (line 27) stdout ----
[01:10:18]  error: this feature has been stable since 1.27.0. Attribute no longer needed
[01:10:18]  --> ../stdsimd/coresimd/macros.rs:28:12
[01:10:18]   |
[01:10:18] 3 | #![feature(cfg_target_feature, target_feature, stdsimd)]
---
[01:10:18]   = note: #[deny(stable_features)] implied by #[deny(warnings)]
[01:10:18]
[01:10:18] error: this feature has been stable since 1.27.0. Attribute no longer needed
[01:10:18]  --> ../stdsimd/coresimd/macros.rs:28:32
[01:10:18]   |
[01:10:18] 3 | #![feature(cfg_target_feature, target_feature, stdsimd)]
[01:10:18]   |                                ^^^^^^^^^^^^^^
[01:10:18]
[01:10:18] thread '../stdsimd/coresimd/macros.rs - coresimd::x86::__m128 (line 27)' panicked at 'couldn't compile the test', librustdoc/test.rs:306:13
[01:10:18] note: Run with `RUST_BACKTRACE=1` for a backtrace.
[01:10:18]
[01:10:18] ---- ../stdsimd/coresimd/macros.rs - coresimd::x86::__m128d (line 27) stdout ----
[01:10:18]  error: this feature has been stable since 1.27.0. Attribute no longer needed
[01:10:18]  --> ../stdsimd/coresimd/macros.rs:28:12
[01:10:18]   |
[01:10:18] 3 | #![feature(cfg_target_feature, target_feature, stdsimd)]
---
[01:10:18]   = note: #[deny(stable_features)] implied by #[deny(warnings)]
[01:10:18]
[01:10:18] error: this feature has been stable since 1.27.0. Attribute no longer needed
[01:10:18]  --> ../stdsimd/coresimd/macros.rs:28:32
[01:10:18]   |
[01:10:18] 3 | #![feature(cfg_target_feature, target_feature, stdsimd)]
[01:10:18]   |                                ^^^^^^^^^^^^^^
[01:10:18]
[01:10:18] thread '../stdsimd/coresimd/macros.rs - coresimd::x86::__m128d (line 27)' panicked at 'couldn't compile the test', librustdoc/test.rs:306:13
[01:10:18]
[01:10:18] ---- ../stdsimd/coresimd/macros.rs - coresimd::x86::__m128i (line 34) stdout ----
[01:10:18]  error: this feature has been stable since 1.27.0. Attribute no longer needed
[01:10:18]  --> ../stdsimd/coresimd/macros.rs:35:12
[01:10:18]   |
[01:10:18] 3 | #![feature(cfg_target_feature, target_feature, stdsimd)]
---
[01:10:18]   = note: #[deny(stable_features)] implied by #[deny(warnings)]
[01:10:18]
[01:10:18] error: this feature has been stable since 1.27.0. Attribute no longer needed
[01:10:18]  --> ../stdsimd/coresimd/macros.rs:35:32
[01:10:18]   |
[01:10:18] 3 | #![feature(cfg_target_feature, target_feature, stdsimd)]
[01:10:18]   |                                ^^^^^^^^^^^^^^
[01:10:18]
[01:10:18] thread '../stdsimd/coresimd/macros.rs - coresimd::x86::__m128i (line 34)' panicked at 'couldn't compile the test', librustdoc/test.rs:306:13
[01:10:18]
[01:10:18] ---- ../stdsimd/coresimd/macros.rs - coresimd::x86::__m256 (line 27) stdout ----
[01:10:18]  error: this feature has been stable since 1.27.0. Attribute no longer needed
[01:10:18]  --> ../stdsimd/coresimd/macros.rs:28:12
[01:10:18]   |
[01:10:18] 3 | #![feature(cfg_target_feature, target_feature, stdsimd)]
---
[01:10:18]   = note: #[deny(stable_features)] implied by #[deny(warnings)]
[01:10:18]
[01:10:18] error: this feature has been stable since 1.27.0. Attribute no longer needed
[01:10:18]  --> ../stdsimd/coresimd/macros.rs:28:32
[01:10:18]   |
[01:10:18] 3 | #![feature(cfg_target_feature, target_feature, stdsimd)]
[01:10:18]   |                                ^^^^^^^^^^^^^^
[01:10:18]
[01:10:18] thread '../stdsimd/coresimd/macros.rs - coresimd::x86::__m256 (line 27)' panicked at 'couldn't compile the test', librustdoc/test.rs:306:13
[01:10:18]
[01:10:18] ---- ../stdsimd/coresimd/macros.rs - coresimd::x86::__m64 (line 34) stdout ----
[01:10:18]  error: this feature has been stable since 1.27.0. Attribute no longer needed
[01:10:18]  --> ../stdsimd/coresimd/macros.rs:35:12
[01:10:18]   |
[01:10:18] 3 | #![feature(cfg_target_feature, target_feature, stdsimd)]
---
[01:10:18]   = note: #[deny(stable_features)] implied by #[deny(warnings)]
[01:10:18]
[01:10:18] error: this feature has been stable since 1.27.0. Attribute no longer needed
[01:10:18]  --> ../stdsimd/coresimd/macros.rs:35:32
[01:10:18]   |
[01:10:18] 3 | #![feature(cfg_target_feature, target_feature, stdsimd)]
[01:10:18]   |                                ^^^^^^^^^^^^^^
[01:10:18]
[01:10:18] thread '../stdsimd/coresimd/macros.rs - coresimd::x86::__m64 (line 34)' panicked at 'couldn't compile the test', librustdoc/test.rs:306:13
[01:10:18]
[01:10:18] ---- ../stdsimd/coresimd/macros.rs - coresimd::x86::__m256d (line 27) stdout ----
[01:10:18]  error: this feature has been stable since 1.27.0. Attribute no longer needed
[01:10:18]  --> ../stdsimd/coresimd/macros.rs:28:12
[01:10:18]   |
[01:10:18] 3 | #![feature(cfg_target_feature, target_feature, stdsimd)]
---
[01:10:18]   = note: #[deny(stable_features)] implied by #[deny(warnings)]
[01:10:18]
[01:10:18] error: this feature has been stable since 1.27.0. Attribute no longer needed
[01:10:18]  --> ../stdsimd/coresimd/macros.rs:28:32
[01:10:18]   |
[01:10:18] 3 | #![feature(cfg_target_feature, target_feature, stdsimd)]
[01:10:18]   |                                ^^^^^^^^^^^^^^
[01:10:18]
[01:10:18] thread '../stdsimd/coresimd/macros.rs - coresimd::x86::__m256d (line 27)' panicked at 'couldn't compile the test', librustdoc/test.rs:306:13
[01:10:18]
[01:10:18] ---- ../stdsimd/coresimd/macros.rs - coresimd::x86::__m256i (line 31) stdout ----
[01:10:18]  error: this feature has been stable since 1.27.0. Attribute no longer needed
[01:10:18]  --> ../stdsimd/coresimd/macros.rs:32:12
[01:10:18]   |
[01:10:18] 3 | #![feature(cfg_target_feature, target_feature, stdsimd)]
---
[01:10:18]   = note: #[deny(stable_features)] implied by #[deny(warnings)]
[01:10:18]
[01:10:18] error: this feature has been stable since 1.27.0. Attribute no longer needed
[01:10:18]  --> ../stdsimd/coresimd/macros.rs:32:32
[01:10:18]   |
[01:10:18] 3 | #![feature(cfg_target_feature, target_feature, stdsimd)]
[01:10:18]   |                                ^^^^^^^^^^^^^^
[01:10:18]
[01:10:18] thread '../stdsimd/coresimd/macros.rs - coresimd::x86::__m256i (line 31)' panicked at 'couldn't compile the test', librustdoc/test.rs:306:13
---
[01:10:18] command did not execute successfully: "/checkout/obj/build/x86_64-unknown-linux-gnu/stage0/bin/cargo" "test" "--target" "x86_64-unknown-linux-gnu" "--release" "--locked" "--color" "always" "--features" "panic-unwind jemalloc backtrace" "--manifest-path" "/checkout/src/libstd/Cargo.toml" "-p" "core" "--" "--quiet"
[01:10:18] expected success, got: exit code: 101
[01:10:18]
[01:10:18]
[01:10:18] failed to run: /checkout/obj/build/bootstrap/debug/bootstrap test
[01:10:18] Build completed unsuccessfully in 0:33:32
[01:10:18] Makefile:58: recipe for target 'check' failed
[01:10:18] make: *** [check] Error 1
---
$ dmesg | grep -i kill
[   10.853990] init: failsafe main process (1093) killed by TERM signal

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.

parched · 2018-04-05T07:30:50Z

Doesn't this stabilise all the (whitelisted) feature names for all the architectures, or am I mistaken? The RFC mentioned this specific set https://github.com/rust-lang/rfcs/blob/master/text/2325-stable-simd.md#the-target_feature-attribute.

bors · 2018-04-05T09:59:08Z

☔ The latest upstream changes (presumably #48851) made this pull request unmergeable. Please resolve the merge conflicts.

Zoxc · 2018-04-05T10:59:20Z

.gitmodules

@@ -49,7 +49,7 @@
 	url = https://github.com/rust-lang/llvm
 [submodule "src/stdsimd"]
 	path = src/stdsimd
-	url = https://github.com/rust-lang-nursery/stdsimd
+	url = https://github.com/alexcrichton/stdsimd


Why doesn't this still use rust-lang-nursery?

Because this isn't ready to land yet, it's waiting on FCP to finish.

Zoxc · 2018-04-05T11:21:39Z

The std::arch::{x86, x86_64} modules and the intrinsics contained inside.

What is the difference between the std::arch::x86 and std::arch::x86_64? Typically x86 includes x86_64 so it seems odd that it is separate. Does x86 refer to 32-bit x86 only?

is_x86_feature_detected!

Why do we have a macro specific to x86? Does that also apply to the std::arch::x86_64 or just std::arch::x86? Do we no longer have a global namespace of CPU features?

steveklabnik · 2018-04-05T12:42:25Z

Why do we have a macro specific to x86? Does that also apply to the std::arch::x86_64 or just std::arch::x86? Do we no longer have a global namespace of CPU features?

From IRC last night:

20:51 < Gankro> acrichto: why is the is_x86_feature_detected! so precisely named? Why can't it be is_feature_detected?
20:55 <~acrichto> Gankro: it doesn't work on other paltforms
20:55 <~acrichto> in that the arguments are x86-specific
20:55 <~acrichto> (so there's a macro for each arch)
20:56 < Gankro> acrichto: why couldn't it just return false on the plats?
20:57 < Gankro> *other plats
20:57 <~acrichto> Gankro: catches typos and such
20:57 <~acrichto> in that it's currently a deliberate decision
20:57 <~acrichto> (we've discussed this a bunch historically)
20:57 < Gankro> ok
20:57 <~acrichto> it turns out that if the macros is cross paltform it doesn't really help at all
20:57 <~acrichto> it's only invoked in platform-specific contexts anyway

alexcrichton · 2018-04-05T14:03:21Z

@Zoxc

What is the difference between the std::arch::x86 and std::arch::x86_64? Typically x86 includes x86_64 so it seems odd that it is separate. Does x86 refer to 32-bit x86 only?

There is a module per target_arch which is used conditionally, and the module name, like std::os, indicates the non-portability. The ~~std::arch::x86~~ std::arch::x86_64 module is not defined on i686-unknown-linux-gnu, for example.

Why do we have a macro specific to x86? Does that also apply to the std::arch::x86_64 or just std::arch::x86? Do we no longer have a global namespace of CPU features?

As explained by @steveklabnik the macro is x86-specific in its feature detection and list of accepted features. For convenience there is no is_x86_64_feature_detected! macro as it's much more difficult to conditionally invoke macros than work with use. Each platform has its own macro with its own accepted set of arguments, again indicated by the name of the macro.

Zoxc · 2018-04-05T14:19:33Z

There is a module per target_arch which is used conditionally, and the module name, like std::os, indicates the non-portability. The std::arch::x86 module is not defined on i686-unknown-linux-gnu, for example.

So are there modules for i286, i386, i486, i586, i786 too? It doesn't seem terribly useful to have this many variants of x86. What is the justification for this?

As explained by @steveklabnik the macro is x86-specific in its feature detection and list of accepted features.

That doesn't answer the question. Why do we have platform specific macros instead of a single macro? and why does the justification for that not apply to #target_feature(..) or #[cfg(target_feature = ..))?

alexcrichton · 2018-04-05T15:05:51Z

@parched ah yes, thanks! I've pushed a commit which should allow us to tweak the stability of each target feature.

@Zoxc

So are there modules for i286, i386, i486, i586, i786 too? It doesn't seem terribly useful to have this many variants of x86. What is the justification for this?

No, there are not all those modules. I'd recommend reading the RFC and the issue tracker for stdsimd for rationale here. There are only modules for each target_arch, similar to how std::os has only the modules that match #[cfg] like unix/windows/linux/...

That doesn't answer the question. Why do we have platform specific macros instead of a single macro? and why does the justification for that not apply to #target_feature(..) or #[cfg(target_feature = ..))?

The answer to your question was indicated in the IRC logs, a platform-independent macro in no way makes SIMD easier to use. Previous attempts at using the same name and same namespace for attributes didn't provide any ergonomics so a platform-specific macro is the conservative option.

TimNN · 2018-04-05T15:10:39Z

Your PR failed on Travis. 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.

[00:01:29] configure: rust.quiet-tests     := True
---
]       in the 'ui' test suite, with its failures due to
[00:05:07]       missing usage of #![feature(aarch64_target_feature)].
[00:05:07] Hint: If you already have such a test and don't want to rename it,
[00:05:07]       you can also add a // gate-test-aarch64_target_feature line to the test file.
[00:05:07] Expected a gate test for the feature 'mmx_target_feature'.
[00:05:07] Hint: create a failing test file named 'feature-gate-mmx_target_feature.rs'
[00:05:07]       in the 'ui' test suite, with its failures due to
[00:05:07]       missing usage of #![feature(mmx_target_feature)].
[00:05:07] Hint: If you already have such a test and don't want to rename it,
[00:05:07]       you can also add a // gate-test-mmx_target_feature line to the test file.
[00:05:07] Expected a gate test for the feature 'powerpc_target_feature'.
[00:05:07] Hint: create a failing test file named 'feature-gate-powerpc_target_feature.rs'
[00:05:07]       in the 'ui' test suite, with its failures due to
[00:05:07]       missing usage of #![feature(powerpc_target_feature)].
[00:05:07] Hint: If you already have such a test and don't want to rename it,
[00:05:07]       you can also add a // gate-test-powerpc_target_feature line to the test file.
[00:05:07] Expected a gate test for the feature 'sse4a_target_feature'.
[00:05:07] Hint: create a failing test file named 'feature-gate-sse4a_target_feature.rs'
[00:05:07]       in the 'ui' test suite, with its failures due to
[00:05:07]       missing usage of #![feature(sse4a_target_feature)].
[00:05:07] Hint: If you already have such a test and don't want to rename it,
[00:05:07]       you can also add a // gate-test-sse4a_target_feature line to the test file.
[00:05:07] Expected a gate test for the feature 'avx512_target_feature'.
[00:05:07] Hint: create a failing test file named 'feature-gate-avx512_target_feature.rs'
[00:05:07]       in the 'ui' test suite, with its failures due to
[00:05:07]       missing usage of #![feature(avx512_target_feature)].
[00:05:07] Hint: If you already have such a test and don't want to rename it,
[00:05:07]       you can also add a // gate-test-avx512_target_feature line to the test file.
[00:05:07] Expected a gate test for the feature 'tbm_target_feature'.
[00:05:07] Hint: create a failing test file named 'feature-gate-tbm_target_feature.rs'
[00:05:07]       in the 'ui' test suite, with its failures due to
[00:05:07]       missing usage of #![feature(tbm_target_feature)].
[00:05:07] Hint: If you already have such a test and don't want to rename it,
[00:05:07]       you can also add a // gate-test-tbm_target_feature line to the test file.
---
$ ls -lat $HOME/Library/Logs/DiagnosticReports/
ls: cannot access /home/travis/Library/Logs/DiagnosticReports/: No such file or directory
travis_time:end:3995dc75:start=1522941035763776228,finish=1522941035769506889,duration=5730661
travis_fold:end:after_failure.2
travis_fold:start:after_failure.3
travis_time:start:1833ec00
$ find $HOME/Library/Logs/DiagnosticReports -type f -name '*.crash' -not -name '*.stage2-*.crash' -not -name 'com.apple.CoreSimulator.CoreSimulatorService-*.crash' -exec printf travis_fold":start:crashlog\n\033[31;1m%s\033[0m\n" {} \; -exec head -750 {} \; -exec echo travis_fold":"end:crashlog \; || true
find: `/home/travis/Library/Logs/DiagnosticReports': No such file or directory
travis_time:end:1833ec00:start=1522941035774367335,finish=1522941035779994962,duration=5627627
travis_fold:end:after_failure.3
travis_fold:start:after_failure.4
travis_time:start:0fb35062
$ dmesg | grep -i kill
[   10.571310] init: failsafe main process (1093) killed by TERM signal

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.

hanna-kruppe · 2018-04-05T15:37:09Z

@alexcrichton Did you have a typo in this sentence?

The std::arch::x86 module is not defined on i686-unknown-linux-gnu, for example.

I would have expected either of:

The std::arch::x86 module is not defined on x86_64-unknonw-linux-gnu
The std::arch::x86_64 module is not defined on i686-unknown-linux-gnu

TimNN · 2018-04-05T15:38:31Z

Your PR failed on Travis. 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.

[00:00:46] configure: rust.quiet-tests     := True
---
[00:37:24] ..........................................................................i.........................
[00:37:29] .................i..................................................................................
---
[00:38:03] ...........................................................................................i........
[00:38:09] .................................................................i..................................
---
[00:38:59] .............................................i......................................................
---
[00:42:31] .............................i......................................................................
[00:42:44] ..............................................................i.....................................
[00:42:59] ...............................................i....................................................
[00:43:17] ....................................................................................................
[00:43:37] ....................................................................................................
[00:43:57] ....................................................................................................
[00:44:20] .i...............................................................................................i..
[00:44:49] ............................................................................................test [run-pass] run-pass/mir_heavy_promoted.rs has been running for over 60 seconds
[00:44:53] ........
[00:45:21] ....................................................................................................
[00:45:54] .............................................................ii.....................................
[00:46:40] ........................i....................................................i.ii................test [run-pass] run-pass/saturating-float-casts.rs has been running for over 60 seconds
[00:46:41] ...
[00:47:18] .....................................................................................iiiiiii........
---
[00:49:27] ..................i............................................................ii.iii...............
[00:49:34] ....................................................................................................
[00:49:42] ........i..............................i............................................................
[00:49:49] ....................................................................................................
[00:49:56] ....................i...............................................................................
[00:50:03] ....................................................................................................
[00:50:13] ....................................................................................................
[00:50:23] ....................................................................................................
[00:50:33] ....................................................................................................
[00:50:46] ....................................................................................................
[00:50:54] .............i......................................................................................
[00:51:03] ................i..ii...............................................................................
[00:51:12] ....................................................................................................
[00:51:22] ....................................................................................................
[00:51:31] ..................................................................................i.................
[00:51:41] ............................i.......................................................................
---
[00:52:16] ...........................i........................................................................
[00:52:18] ....................................................................i...............................
[00:52:19] ................i.......................................................
---
[00:52:32] ...........i........................
---
[00:53:00] i....i.ii....i.............ii........iii......i..i...i...ii..i..i..ii.....
---
[00:53:03] i.......i......................i......
---
[00:53:37] iiii.......i..i........i..i.i.............i..........iiii...........i...i..........ii.i.i.......ii..
[00:53:38] ....ii...
---
[01:01:52] .....i..............................................................................................
---
[01:03:35] .....................................i..............................................................
[01:03:53] ....................................................................................................
[01:04:10] ..........................................i.........................................................
---
[01:05:28] .........................................................ii.........................................
---
[01:06:26] ..............................................................i.....................................
---
[01:10:52] iiFF................................................................................................
[01:11:10] ....................................................................................................
[01:11:25] ...................iii......i......i...i......i.....................................................
[01:11:35] ....................................................................................................
[01:11:49] .........................................iiii........ii.............................................
[01:12:00] ....................................................................................................
[01:12:16] ........................................................................................i...........
[01:12:38] ....................................................................................................
[01:12:48] ..................................................................................................ii
[01:12:57] ii...............................................
[01:12:57] failures:
[01:12:57]
[01:12:57] ---- ../stdsimd/stdsimd/mod.rs - stdsimd::arch (line 207) stdout ----
[01:12:57]  error: this feature has been stable since 1.27.0. Attribute no longer needed
[01:12:57]  --> ../stdsimd/stdsimd/mod.rs:208:12
[01:12:57]   |
[01:12:57] 3 | #![feature(cfg_target_feature, target_feature, stdsimd)]
---
[01:12:57]   = note: #[deny(stable_features)] implied by #[deny(warnings)]
[01:12:57]
[01:12:57] error: this feature has been stable since 1.27.0. Attribute no longer needed
[01:12:57]  --> ../stdsimd/stdsimd/mod.rs:208:32
[01:12:57]   |
[01:12:57] 3 | #![feature(cfg_target_feature, target_feature, stdsimd)]
[01:12:57]   |                                ^^^^^^^^^^^^^^
[01:12:57]
[01:12:57] thread '../stdsimd/stdsimd/mod.rs - stdsimd::arch (line 207)' panicked at 'couldn't compile the test', librustdoc/test.rs:306:13
[01:12:57] note: Run with `RUST_BACKTRACE=1` for a backtrace.
[01:12:57]
[01:12:57] ---- ../stdsimd/stdsimd/mod.rs - stdsimd::arch (line 249) stdout ----
[01:12:57]  error: this feature has been stable since 1.27.0. Attribute no longer needed
[01:12:57]  --> ../stdsimd/stdsimd/mod.rs:250:12
[01:12:57]   |
[01:12:57] 3 | #![feature(cfg_target_feature, target_feature, stdsimd)]
---
[01:12:57]   = note: #[deny(stable_features)] implied by #[deny(warnings)]
[01:12:57]
[01:12:57] error: this feature has been stable since 1.27.0. Attribute no longer needed
[01:12:57]  --> ../stdsimd/stdsimd/mod.rs:250:32
[01:12:57]   |
[01:12:57] 3 | #![feature(cfg_target_feature, target_feature, stdsimd)]
[01:12:57]   |                                ^^^^^^^^^^^^^^
[01:12:57]
[01:12:57] thread '../stdsimd/stdsimd/mod.rs - stdsimd::arch (line 249)' panicked at 'couldn't compile the test', librustdoc/test.rs:306:13
---
[01:12:57] command did not execute successfully: "/checkout/obj/build/x86_64-unknown-linux-gnu/stage0/bin/cargo" "test" "--target" "x86_64-unknown-linux-gnu" "--release" "--locked" "--color" "always" "--features" "panic-unwind jemalloc backtrace" "--manifest-path" "/checkout/src/libstd/Cargo.toml" "-p" "std" "--" "--quiet"
[01:12:57] expected success, got: exit code: 101
[01:12:57]
[01:12:57]
[01:12:57] failed to run: /checkout/obj/build/bootstrap/debug/bootstrap test
[01:12:57] Build completed unsuccessfully in 0:36:39
[01:12:57] make: *** [check] Error 1
[01:12:57] Makefile:58: recipe for target 'check' failed
---
$ ls -lat $HOME/Library/Logs/DiagnosticReports/
ls: cannot access /home/travis/Library/Logs/DiagnosticReports/: No such file or directory
travis_time:end:00701732:start=1522942702499426526,finish=1522942702506606725,duration=7180199
travis_fold:end:after_failure.2
travis_fold:start:after_failure.3
travis_time:start:13db3254
$ find $HOME/Library/Logs/DiagnosticReports -type f -name '*.crash' -not -name '*.stage2-*.crash' -not -name 'com.apple.CoreSimulator.CoreSimulatorService-*.crash' -exec printf travis_fold":start:crashlog\n\033[31;1m%s\033[0m\n" {} \; -exec head -750 {} \; -exec echo travis_fold":"end:crashlog \; || true
find: `/home/travis/Library/Logs/DiagnosticReports': No such file or directory
travis_time:end:13db3254:start=1522942702512819864,finish=1522942702520055466,duration=7235602
travis_fold:end:after_failure.3
travis_fold:start:after_failure.4
travis_time:start:015d41a2
$ dmesg | grep -i kill
[   10.213694] init: failsafe main process (1096) killed by TERM signal

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.

alexcrichton · 2018-04-05T15:39:50Z

Er oops sorry yeah, @rkruppe you are correct! (both of those alternatives are true)

Zoxc · 2018-04-05T16:21:26Z

I'd expect to have std::arch::x86_16, std::arch::x86_32 and std::arch::x86_64 for x86 or just a single std::arch::x86 which covers all modes. Similarly for ARM std::arch::aarch32, std::arch::aarch64 and possibly std::arch::t32 or something for Thumb.

The answer to your question was indicated in the IRC logs, a platform-independent macro in no way makes SIMD easier to use.

How does having a platform-independent macro make SIMD harder to use?

Previous attempts at using the same name and same namespace for attributes didn't provide any ergonomics so a platform-specific macro is the conservative option.

By that logic we should also have platform-specific versions of #[target_feature(..)] and #[cfg(target_feature = ..)). Using a platform-specific macro doesn't necessary in other to be conservative, just having a namespace per architecture should suffice.

TimNN · 2018-04-05T17:20:58Z

Your PR failed on Travis. 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.

[00:00:46] configure: rust.quiet-tests     := True

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.

alexcrichton · 2018-04-05T17:58:04Z

I'd expect to have std::arch::x86_16, std::arch::x86_32 and std::arch::x86_64 for x86 or just a single std::arch::x86 which covers all modes. Similarly for ARM std::arch::aarch32, std::arch::aarch64 and possibly std::arch::t32 or something for Thumb.

This does not match #[cfg(target_arch)], the modules match the well-established convention in Rust with #[cfg(target_arch)]

How does having a platform-independent macro make SIMD harder to use?

You've misread that, I said that it wasn't easier to use, not harder to use.

bors · 2018-04-16T14:27:55Z

📌 Commit b7a4532 has been approved by BurntSushi

bors · 2018-04-16T19:23:58Z

⌛ Testing commit b7a453287867cfc10d6d03bfccad00aad09e711f with merge 2d76bee9ca65c77b4f42a13ced55d2f43b5e03ef...

bors · 2018-04-16T20:35:53Z

💔 Test failed - status-appveyor

Use an explicit whitelist for what features are actually stable and can be enabled.

alexcrichton · 2018-04-16T20:59:08Z

@bors: r=BurntSushi

bors · 2018-04-16T20:59:09Z

📌 Commit 1217d70 has been approved by BurntSushi

bors · 2018-04-17T03:57:29Z

⌛ Testing commit 1217d70 with merge 186db76...

Stabilize x86/x86_64 SIMD This commit stabilizes the SIMD in Rust for the x86/x86_64 platforms. Notably this commit is stabilizing: * The `std::arch::{x86, x86_64}` modules and the intrinsics contained inside. * The `is_x86_feature_detected!` macro in the standard library * The `#[target_feature(enable = "...")]` attribute * The `#[cfg(target_feature = "...")]` matcher Stabilization of the module and intrinsics were primarily done in rust-lang/stdarch#414 and the two attribute stabilizations are done in this commit. The standard library is also tweaked a bit with the new way that stdsimd is integrated. Note that other architectures like `std::arch::arm` are not stabilized as part of this commit, they will likely stabilize in the future after they've been implemented and fleshed out. Similarly the `std::simd` module is also not being stabilized in this commit, only `std::arch`. Finally, nothing related to `__m64` is stabilized in this commit either (MMX), only SSE and up types and intrinsics are stabilized. Closes #29717 Closes #44839 Closes #48556

bors · 2018-04-17T06:46:26Z

☀️ Test successful - status-appveyor, status-travis
Approved by: BurntSushi
Pushing 186db76 to master...

rust-highfive assigned TimNN Apr 4, 2018

alexcrichton mentioned this pull request Apr 4, 2018

Tracking issue for stable SIMD in Rust #48556

Closed

3 tasks

alexcrichton added the T-libs-api Relevant to the library API team, which will review and decide on the PR/issue. label Apr 4, 2018

alexcrichton mentioned this pull request Apr 4, 2018

Log analysis error for PR 49664 rust-lang/rust-log-analyzer#4

Open

alexcrichton force-pushed the stable-simd branch from 01ec2d0 to 6de696a Compare April 4, 2018 22:37

pietroalbini added relnotes Marks issues that should be documented in the release notes of the next release. S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. labels Apr 5, 2018

Zoxc reviewed Apr 5, 2018

View reviewed changes

alexcrichton force-pushed the stable-simd branch 2 times, most recently from ffcd619 to 8a02174 Compare April 5, 2018 14:06

alexcrichton force-pushed the stable-simd branch from edf2362 to ca8a7ca Compare April 5, 2018 16:32

alexcrichton force-pushed the stable-simd branch from ca8a7ca to a97e381 Compare April 5, 2018 17:56

alexcrichton force-pushed the stable-simd branch from 9d51780 to a97e381 Compare April 5, 2018 18:37

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 Apr 16, 2018

Separately gate each target_feature feature

1217d70

Use an explicit whitelist for what features are actually stable and can be enabled.

alexcrichton force-pushed the stable-simd branch from b7a4532 to 1217d70 Compare April 16, 2018 20:58

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 Apr 16, 2018

bors merged commit 1217d70 into rust-lang:master Apr 17, 2018

bors mentioned this pull request Apr 17, 2018

Implementation of RFC 2086 - Allow Irrefutable Let patterns #49469

Merged

alexcrichton deleted the stable-simd branch April 17, 2018 14:18

KodrAus mentioned this pull request Apr 18, 2018

Use std::arch for crc checking if available BurntSushi/rust-snappy#19

Closed

This was referenced May 7, 2018

Migrate to stabilized x86/x86_64 SIMD sunchao/parquet-rs#113

Closed

Compile with stable Rust sunchao/parquet-rs#119

Open

alexcrichton mentioned this pull request May 26, 2018

Tracking issue for RFC 2045: improving #[target_feature] #44839

Open

22 tasks

codeworm96 mentioned this pull request Jun 25, 2018

circleci: Specify tarpaulin version tikv/tikv#3241

Merged

mssun mentioned this pull request Aug 14, 2018

Consider using SIMD primitives to accelerate computation yogcrypt/yogcrypt#16

Open

zzau13 mentioned this pull request Nov 19, 2018

Add SIMD escape djc/askama#160

Closed

pnkfelix mentioned this pull request Dec 3, 2018

NLL error on closure, but not on equivalent function #55526

Closed

ehuss mentioned this pull request Mar 18, 2019

Document target_feature and cfg_target_feature. rust-lang/reference#545

Merged

m-ou-se mentioned this pull request Sep 17, 2020

Feature gate for unstable target_feature values not being checked on nightly #76842

Open

workingjubilee mentioned this pull request Oct 20, 2020

Consider limiting #[rustc_args_required_const(...)] to intrinsics. #70271

Closed

RalfJung mentioned this pull request Aug 9, 2023

Non-temporal stores (and _mm_stream operations in stdarch) break our memory model #114582

Open

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Stabilize x86/x86_64 SIMD #49664

Stabilize x86/x86_64 SIMD #49664

alexcrichton commented Apr 4, 2018

alexcrichton commented Apr 4, 2018

TimNN commented Apr 4, 2018

TimNN commented Apr 4, 2018

parched commented Apr 5, 2018

bors commented Apr 5, 2018

Zoxc Apr 5, 2018

alexcrichton Apr 5, 2018

Zoxc commented Apr 5, 2018

steveklabnik commented Apr 5, 2018

alexcrichton commented Apr 5, 2018 •

edited

Zoxc commented Apr 5, 2018

alexcrichton commented Apr 5, 2018

TimNN commented Apr 5, 2018

hanna-kruppe commented Apr 5, 2018

TimNN commented Apr 5, 2018

alexcrichton commented Apr 5, 2018

Zoxc commented Apr 5, 2018

TimNN commented Apr 5, 2018

alexcrichton commented Apr 5, 2018

bors commented Apr 16, 2018

bors commented Apr 16, 2018

bors commented Apr 16, 2018

alexcrichton commented Apr 16, 2018

bors commented Apr 16, 2018

bors commented Apr 17, 2018

bors commented Apr 17, 2018

Stabilize x86/x86_64 SIMD #49664

Stabilize x86/x86_64 SIMD #49664

Conversation

alexcrichton commented Apr 4, 2018

alexcrichton commented Apr 4, 2018

TimNN commented Apr 4, 2018

TimNN commented Apr 4, 2018

parched commented Apr 5, 2018

bors commented Apr 5, 2018

Zoxc Apr 5, 2018

Choose a reason for hiding this comment

alexcrichton Apr 5, 2018

Choose a reason for hiding this comment

Zoxc commented Apr 5, 2018

steveklabnik commented Apr 5, 2018

alexcrichton commented Apr 5, 2018 • edited

Zoxc commented Apr 5, 2018

alexcrichton commented Apr 5, 2018

TimNN commented Apr 5, 2018

hanna-kruppe commented Apr 5, 2018

TimNN commented Apr 5, 2018

alexcrichton commented Apr 5, 2018

Zoxc commented Apr 5, 2018

TimNN commented Apr 5, 2018

alexcrichton commented Apr 5, 2018

bors commented Apr 16, 2018

bors commented Apr 16, 2018

bors commented Apr 16, 2018

alexcrichton commented Apr 16, 2018

bors commented Apr 16, 2018

bors commented Apr 17, 2018

bors commented Apr 17, 2018

alexcrichton commented Apr 5, 2018 •

edited