`intrinsic-test`: Adding x86 behavioural testing. #1894

madhav-madhusoodanan · 2025-08-03T13:31:06Z

Context

This is a redo of PR #1814, since a lot of details have changed with PRs #1863, #1862, #1861, #1852.

crates/intrinsic-test/src/x86/intrinsic.rs

crates/intrinsic-test/src/x86/mod.rs

folkertdev · 2025-08-05T10:36:48Z

crates/intrinsic-test/src/x86/types.rs

+                match str::parse::<u32>(etype_processed.as_str()) {
+                    Ok(value) => data.bit_len = Some(value),
+                    Err(_) => {
+                        data.bit_len = match data.kind() {
+                            TypeKind::Char(_) => Some(8),
+                            TypeKind::BFloat => Some(16),
+                            TypeKind::Int(_) => Some(32),
+                            TypeKind::Float => Some(32),
+                            _ => None,
+                        };
+                    }
+                }


why are only some type kinds covered here? Maybe this could be a method on TypeKind?

crates/intrinsic-test/src/x86/compile.rs

folkertdev

you should rebase on top of the upstream master branch instead of merging it in. That keeps the git history clean.

folkertdev · 2025-08-05T22:48:30Z

ci/run.sh

+    x86_64-unknown-linux-gnu*)
+        CPPFLAGS="${TEST_CPPFLAGS}" RUSTFLAGS="${HOST_RUSTFLAGS}" RUST_LOG=warn \
+            cargo run "${INTRINSIC_TEST}" "${PROFILE}"  \
+            --bin intrinsic-test -- intrinsics_data/x86-intel.xml \
+            --runner "${TEST_RUNNER}" \
+            --cppcompiler "${TEST_CXX_COMPILER}" \
+            --target "${TARGET}"
+        ;;


we'll have to see how to do this exactly, but we want to split these out of the main CI job to speed it up

madhav-madhusoodanan · 2025-09-05T08:19:24Z

Seems like the CI run at this point failed due to this error. I'll retry shortly:

#6 39.45   Could not connect to archive.ubuntu.com:80 (185.125.190.82), connection timed out Could not connect to archive.ubuntu.com:80 (185.125.190.83), connection timed out Could not connect to archive.ubuntu.com:80 (91.189.91.83), connection timed out Could not connect to archive.ubuntu.com:80 (185.125.190.81), connection timed out Could not connect to archive.ubuntu.com:80 (91.189.91.81), connection timed out Could not connect to archive.ubuntu.com:80 (185.125.190.36), connection timed out Could not connect to archive.ubuntu.com:80 (185.125.190.39), connection timed out Could not connect to archive.ubuntu.com:80 (91.189.91.82), connection timed out
#6 39.45   Unable to connect to archive.ubuntu.com:80:
#6 39.45 Fetched 126 kB in 39s (3208 B/s)
#6 39.45 Reading package lists...
#6 39.46 W: Failed to fetch http://archive.ubuntu.com/ubuntu/dists/questing/InRelease  Unable to connect to archive.ubuntu.com:80:
#6 39.46 W: Failed to fetch http://archive.ubuntu.com/ubuntu/dists/questing-updates/InRelease  Unable to connect to archive.ubuntu.com:80:
#6 39.46 W: Failed to fetch http://archive.ubuntu.com/ubuntu/dists/questing-backports/InRelease  Unable to connect to archive.ubuntu.com:80:
#6 39.46 W: Some index files failed to download. They have been ignored, or old ones used instead.

Notes: 1. chunk_info has been moved to `common/mod.rs` since it will be needed for all architectures

struct

being killed. Also separated declarations and definitions for C++ testfiles.

madhav-madhusoodanan · 2025-09-29T11:30:08Z

@folkertdev Seems like the clang++-19 was being passed as the value to the --cppcompiler flag, instead of clang++. That was the reason behind the No such file or directory error. Fixed it.

I ran into an issue with the process being killed (likely due to resource constraints since the clang++ processes are running in parallel). Would it be okay if I increase the CPU and memory limits within the Github workflow file?

For now I've made it sequential so that I can step over that error, but the amount of time it's taking has got me reconsidering it.

folkertdev · 2025-09-29T18:57:19Z

I ran into an issue with the process being killed (likely due to resource constraints since the clang++ processes are running in parallel). Would it be okay if I increase the CPU and memory limits within the Github workflow file?

I don't think that is how it works? You just get a CI machine (typically with 4 cores) for some duration. Do you know how large these files get (how many intrinsics do you generate a test for)?

We definitely want to be running these processes in parallel.

madhav-madhusoodanan · 2025-09-29T19:20:53Z

Ohh I see, I thought I could edit the resource allocation.
However, while running them in parallel I consistently run into the process being killed.

Let me experiment with keeping more number of mod_i.cpp and a worker pooling mechanism, seeing as we can't change the resource allocation.

folkertdev · 2025-09-29T19:25:59Z

you could also try to just cut the total number of generated tests (e.g. to 10%) to see if that helps?

By the looks of it the whole process is just way too slow, and the runners eventually time out. So, you'll have to figure out how to make it faster. I think the crux here is what I plan to do: link all the c++ and rust code into one binary that generates, in effect, a large number of rust #[test] tests, and just execute that using the given runner.

But that is probably not feasible for you in the remaining time, so by just cutting the total number of tests (maybe using some random sampling, e.g. keep every 10th test) we'll be able to merge your code without completely tanking our CI times.

processing

madhav-madhusoodanan · 2025-09-29T20:11:08Z

Good idea @folkertdev. Let me first check whether processing smaller-sized chunks would be of any help.

to C++ version of the same.

base type

madhav-madhusoodanan · 2025-10-06T05:46:02Z

@sayantn was this change related to the change we made for some x86 intrinsics? Found this in the CI tests

Illegal instruction at address = 56435cedbf49: f2 0f 2b 00 0f ae f8 f2 0f 10 04 24 66 0f 2e 
Image name: /checkout/target/x86_64-unknown-linux-gnu/release/deps/core_arch-92d6a56b530b6fbf
Offset in image: 0x6cef49
If you believe your application should attempt to execute
this illegal instruction (and others that may be present),
Then use this knob: -emit-illegal-insts 0
and this error message will be avoided.
Use -print-exception-details to get more details.

SDE ERROR: Illegal instruction at address = 56435cedbf49: f2 0f 2b 00 0f ae f8 f2 0f 10 04 24 66 0f 2e 
Image name: /checkout/target/x86_64-unknown-linux-gnu/release/deps/core_arch-92d6a56b530b6fbf
Offset in image: 0x6cef49
If you believe your application should attempt to execute
this illegal instruction (and others that may be present),
Then use this knob: -emit-illegal-insts 0
and this error message will be avoided.
Use -print-exception-details to get more details.

sayantn · 2025-10-07T03:28:19Z

I don't think I have seen this before, seems to be in _mm_stream_sd. I did recently update the implementation in #1929, but the CI did pass there (and locally). It shouldn't happen if we didn't change anything, might be spurious. Worth a retry in CI

instead of letting C++ do it (and potentially change the bits)

Amanieu · 2025-10-07T16:10:58Z

_mm_stream_sd and _mm_stream_ss seem to be in sse4a, not sse2.

sayantn · 2025-10-07T16:17:48Z

Yea, and the CI failures looked like it got to sse4a and then failed (although I don't know how much we can trust output order, considering that the tests are parallelized)

Amanieu · 2025-10-07T16:24:40Z

SSE4a is an AMD extension, so intel-sde may not support it. Does the intrinsic tester check for feature availability before using those features?

sayantn · 2025-10-07T17:04:42Z

Iirc sde supports (most) amd extensions. It even has a cpu preset named amd-future. Currently we just enable the sse4a, vp2intersect, xop and tbm in the future cpu preset of sde

rustbot assigned folkertdev Aug 3, 2025

madhav-madhusoodanan commented Aug 3, 2025

View reviewed changes

crates/intrinsic-test/src/x86/intrinsic.rs Outdated Show resolved Hide resolved

madhav-madhusoodanan force-pushed the intrinsic-test-x86-addition branch 2 times, most recently from 41db5a8 to 5fc0f3b Compare August 5, 2025 10:16

folkertdev reviewed Aug 5, 2025

View reviewed changes

madhav-madhusoodanan force-pushed the intrinsic-test-x86-addition branch from 9e28106 to 111cd5d Compare August 5, 2025 16:22

folkertdev reviewed Aug 5, 2025

View reviewed changes

madhav-madhusoodanan force-pushed the intrinsic-test-x86-addition branch from f3f87f2 to 2ec747c Compare August 9, 2025 12:20

madhav-madhusoodanan force-pushed the intrinsic-test-x86-addition branch from 2ec747c to a8313d0 Compare September 5, 2025 08:16

madhav-madhusoodanan force-pushed the intrinsic-test-x86-addition branch 9 times, most recently from a119be1 to 0f3900b Compare September 8, 2025 19:25

madhav-madhusoodanan added 11 commits September 10, 2025 00:10

feat: added the skeleton structure of the x86 module

1eee9ed

feat: added the XML intrinsic parser for x86

07f20f4

feat: updated intrinsics creation

191614a

feat: update building C code for x86 architecture.

a952d3b

Notes: 1. chunk_info has been moved to `common/mod.rs` since it will be needed for all architectures

fix: code cleanup

417f729

chore: added Regex crate, updated the structure of X86IntrinsicType

4c19bd2

struct

feat: implemented build_rust_file of x86 module

3e3bf65

feat: implemented compare_outputs of x86 module

3e95708

feat: implement print_result_c for Intrinsic<X86IntrinsicType>

fbb8214

feat: Added x86 to CI pipeline

6c69404

fix: update arch flags being sent to the x86 compilation command

0d62fe4

madhav-madhusoodanan force-pushed the intrinsic-test-x86-addition branch from c10accb to ce16379 Compare September 28, 2025 20:15

fix: making compilation step run one by one to prevent the process from

411fbcd

being killed. Also separated declarations and definitions for C++ testfiles.

feat: attempting compilation of smaller chunks for faster parallel

ee2f1e7

processing

madhav-madhusoodanan force-pushed the intrinsic-test-x86-addition branch 2 times, most recently from c81c215 to 6733138 Compare September 30, 2025 08:06

feat: add c_programs to PATH and increase chunk size to 400

4b06d4b

madhav-madhusoodanan force-pushed the intrinsic-test-x86-addition branch 2 times, most recently from 2876624 to 6be6472 Compare October 1, 2025 13:13

feat: display __mmask8 values so that non-utf8 values are not displayed

87e39a2

madhav-madhusoodanan force-pushed the intrinsic-test-x86-addition branch from 6be6472 to 87e39a2 Compare October 2, 2025 19:47

madhav-madhusoodanan mentioned this pull request Oct 3, 2025

Fix mistake in #1928 #1931

Merged

madhav-madhusoodanan force-pushed the intrinsic-test-x86-addition branch 2 times, most recently from 20ec01d to b476cd1 Compare October 3, 2025 13:04

feat: add formatting for __m128i, __m256i, __m512i types that is similar

3b77dc5

to C++ version of the same.

madhav-madhusoodanan force-pushed the intrinsic-test-x86-addition branch from b476cd1 to 3b77dc5 Compare October 3, 2025 13:10

feat: make the debug_i16 into a generic debug_as function that adapts to

9dbc078

base type

madhav-madhusoodanan force-pushed the intrinsic-test-x86-addition branch from ceb2a37 to 9dbc078 Compare October 5, 2025 18:48

feat: casting the results of the lane function by preserving the bits

8c19290

instead of letting C++ do it (and potentially change the bits)

madhav-madhusoodanan force-pushed the intrinsic-test-x86-addition branch from 81bd07f to 8c19290 Compare October 7, 2025 05:16

intrinsic-test: Adding x86 behavioural testing. #1894

Are you sure you want to change the base?

intrinsic-test: Adding x86 behavioural testing. #1894

Uh oh!

Conversation

madhav-madhusoodanan commented Aug 3, 2025

Context

Uh oh!

Uh oh!

Uh oh!

folkertdev Aug 5, 2025

Choose a reason for hiding this comment

Uh oh!

Uh oh!

folkertdev left a comment

Choose a reason for hiding this comment

Uh oh!

folkertdev Aug 5, 2025

Choose a reason for hiding this comment

Uh oh!

madhav-madhusoodanan commented Sep 5, 2025

Uh oh!

madhav-madhusoodanan commented Sep 29, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

folkertdev commented Sep 29, 2025

Uh oh!

madhav-madhusoodanan commented Sep 29, 2025

Uh oh!

folkertdev commented Sep 29, 2025

Uh oh!

madhav-madhusoodanan commented Sep 29, 2025

Uh oh!

madhav-madhusoodanan commented Oct 6, 2025

Uh oh!

sayantn commented Oct 7, 2025

Uh oh!

Amanieu commented Oct 7, 2025

Uh oh!

sayantn commented Oct 7, 2025

Uh oh!

Amanieu commented Oct 7, 2025

Uh oh!

sayantn commented Oct 7, 2025

Uh oh!

Uh oh!

`intrinsic-test`: Adding x86 behavioural testing. #1894

`intrinsic-test`: Adding x86 behavioural testing. #1894

madhav-madhusoodanan commented Sep 29, 2025 •

edited

Loading