Stream: git-wasmtime

Topic: wasmtime / PR #10954 x64: Add AVX encodings for `cvt*` in...

Wasmtime GitHub notifications bot (Jun 06 2025 at 16:27):

alexcrichton opened PR #10954 from alexcrichton:x64-vcvt to bytecodealliance:main:

Helps to light the CvtIntToFloatVex instruction shape on :fire:.

<!--
Please make sure you include the following information:

• If this work has been discussed elsewhere, please include a link to that
  conversation. If it was discussed in an issue, just mention "issue #...".

• Explain why this change is needed. If the details are in an issue already,
  this can be brief.

Our development process is documented in the Wasmtime book:
https://docs.wasmtime.dev/contributing-development-process.html

Please ensure all communication follows the code of conduct:
https://github.com/bytecodealliance/wasmtime/blob/main/CODE_OF_CONDUCT.md
-->

Wasmtime GitHub notifications bot (Jun 06 2025 at 16:27):

alexcrichton requested abrown for a review on PR #10954.

Wasmtime GitHub notifications bot (Jun 06 2025 at 16:27):

alexcrichton requested wasmtime-compiler-reviewers for a review on PR #10954.

Wasmtime GitHub notifications bot (Jun 06 2025 at 16:32):

alexcrichton updated PR #10954.

Wasmtime GitHub notifications bot (Jun 06 2025 at 16:53):

abrown created PR review comment:

Huh, weird... :shrug:

Wasmtime GitHub notifications bot (Jun 06 2025 at 16:53):

abrown submitted PR review:

Nice!

Wasmtime GitHub notifications bot (Jun 06 2025 at 16:53):

abrown created PR review comment:

        inst("vcvtsi2ssl", fmt("B", [w(xmm1), r(xmm2), r(rm32)]), vex(LIG)._f3()._0f().w0().op(0x2A).r(), _64b | compat | avx),
        inst("vcvtsi2ssq", fmt("B", [w(xmm1), r(xmm2), r(rm64)]), vex(LIG)._f3()._0f().w1().op(0x2A).r(), _64b | avx),

It's unfortunate I had to turn off rustfmt...

Wasmtime GitHub notifications bot (Jun 06 2025 at 17:22):

alexcrichton updated PR #10954.

Wasmtime GitHub notifications bot (Jun 06 2025 at 17:24):

alexcrichton submitted PR review.

Wasmtime GitHub notifications bot (Jun 06 2025 at 17:24):

alexcrichton created PR review comment:

I think this is to disambiguate between

VEX.128.66.0F.WIG E6 /r
VCVTTPD2DQ xmm1, xmm2/m128

and

VEX.256.66.0F.WIG E6 /r
VCVTTPD2DQ xmm1, ymm2/m256

where in the memory-to-register case the att-style assembler, without the x suffix, would look like vcvttpd2dq (%rax), %xmm1 and it's ambiguous as to which instruction you want. I think the x and y suffixes are used to disambiguate at the assembler level. Now if we used intel syntax...

Wasmtime GitHub notifications bot (Jun 06 2025 at 17:24):

alexcrichton has enabled auto merge for PR #10954.

Wasmtime GitHub notifications bot (Jun 06 2025 at 17:58):

alexcrichton merged PR #10954.

Last updated: Dec 06 2025 at 06:05 UTC