Stream: git-wasmtime

Topic: wasmtime / PR #10316 asm: SSE ALU Instructions

Wasmtime GitHub notifications bot (Mar 01 2025 at 19:46):

rahulchaphalkar opened PR #10316 from rahulchaphalkar:sse-alu-instructions to bytecodealliance:main:

Implement the following SSE instructions:

• xorpd
• xorps
• andpd
• andps
• addpd
• addps
• subpd
• subps
• orps

The opcode for *ps variants of these instructions is NP 0F xx. NP has the following explanation in manual:

NP — Indicates the use of 66/F2/F3 prefixes (beyond those already part of the instructions opcode) are not
allowed with the instruction. Such use will either cause an invalid-opcode exception (#UD) or result in the
encoding for a different instruction.

There is an existing NoPrefix prefix, but I have not reused that to serve above NP as there might be some difference in the intended meaning. Also, NP doesn't classify as a prefix technically. I am thinking to address this in a later patch which would deal with prefix parsing. Any thoughts or feedback on this is welcome.
My thinking is NP doesn't functionally do anything, so it is fine to ignore it. We just need a way for validate function to check this.

Wasmtime GitHub notifications bot (Mar 01 2025 at 19:46):

rahulchaphalkar requested cfallin for a review on PR #10316.

Wasmtime GitHub notifications bot (Mar 01 2025 at 19:46):

rahulchaphalkar requested wasmtime-compiler-reviewers for a review on PR #10316.

Wasmtime GitHub notifications bot (Mar 02 2025 at 05:04):

rahulchaphalkar updated PR #10316.

Wasmtime GitHub notifications bot (Mar 03 2025 at 19:28):

abrown requested abrown for a review on PR #10316.

Wasmtime GitHub notifications bot (Mar 05 2025 at 19:39):

abrown submitted PR review:

Had a couple of code movement comments but overall this all makes sense to me. Thanks!

Wasmtime GitHub notifications bot (Mar 05 2025 at 19:39):

abrown created PR review comment:

Let's remove the old emission. In a recent PR I began removing these for scalar instructions. It ensures that our is_xmm_mem is operating as expected.

Wasmtime GitHub notifications bot (Mar 05 2025 at 19:39):

abrown created PR review comment:

Some minor nits about organizing these new instructions:

• let's put them at the bottom of the file with their own comment header, e.g., // SSE vector addition. or the like
• can we order them from smallest to largest? I.e., packed single-precision (ps = f32x4) first and then packed double-precision (pd = f64x2).

Wasmtime GitHub notifications bot (Mar 05 2025 at 19:39):

abrown created PR review comment:

And on that note, I believe that the ISLE we generate already has the signature (... (Xmm XmmMem) Xmm) which is what we need here. We can drop the is_xmm_mem extractor for these instructions.

Wasmtime GitHub notifications bot (Mar 06 2025 at 19:45):

rahulchaphalkar updated PR #10316.

Wasmtime GitHub notifications bot (Mar 06 2025 at 19:50):

rahulchaphalkar submitted PR review.

Wasmtime GitHub notifications bot (Mar 06 2025 at 19:50):

rahulchaphalkar created PR review comment:

Done

Wasmtime GitHub notifications bot (Mar 06 2025 at 19:51):

rahulchaphalkar submitted PR review.

Wasmtime GitHub notifications bot (Mar 06 2025 at 19:51):

rahulchaphalkar created PR review comment:

Done

Wasmtime GitHub notifications bot (Mar 06 2025 at 19:51):

rahulchaphalkar submitted PR review.

Wasmtime GitHub notifications bot (Mar 06 2025 at 19:51):

rahulchaphalkar created PR review comment:

Done

Wasmtime GitHub notifications bot (Mar 06 2025 at 20:48):

abrown submitted PR review.

Wasmtime GitHub notifications bot (Mar 06 2025 at 21:11):

abrown merged PR #10316.

Last updated: Apr 18 2025 at 00:13 UTC