Stream: git-wasmtime

Topic: wasmtime / PR #5948 Stop supporting vectors with `bmask`

Wasmtime GitHub notifications bot (Mar 07 2023 at 01:36):

elliottt opened PR #5948 from trevor/bmask-simd to main:

The bmask instruction does not have a lowering in any backend that will produce a vector result, but the opcode constraints indicate that it is possible to produce a vector. Any attempt to produce a vector with bmask currently yields a panic, so this PR avoids that by rejecting these cases in the verifier instead.

Additionally, it's possible to produce a vector of comparison results whose lanes will be all 1 or 0 using either icmp or fcmp, rendering this possible behavior of bmask redundant.

Co-authored-by: Jamey Sharp <jsharp@fastly.com>

<!--

Please ensure that the following steps are all taken care of before submitting
the PR.

• [ ] This has been discussed in issue #..., or if not, please tell us why
  here.

• [ ] A short description of what this does, why it is needed; if the
  description becomes long, the matter should probably be discussed in an issue
  first.

• [ ] This PR contains test cases, if meaningful.

• [ ] A reviewer from the core maintainer team has been assigned for this PR.
  If you don't know who could review this, please indicate so. The list of
  suggested reviewers on the right can help you.

Please ensure all communication adheres to the code of conduct.
-->

Wasmtime GitHub notifications bot (Mar 07 2023 at 01:36):

elliottt updated PR #5948 from trevor/bmask-simd to main.

Wasmtime GitHub notifications bot (Mar 07 2023 at 01:38):

elliottt edited PR #5948 from trevor/bmask-simd to main:

The bmask instruction does not have a lowering in any backend that will produce a vector result, but the opcode constraints indicate that it is possible to produce a vector. Any attempt to produce a vector with bmask currently yields a panic, so this PR avoids that by rejecting these cases in the verifier instead.

Additionally, it's possible to produce a vector of comparison results whose lanes will be all 1 or 0 using either icmp or fcmp, rendering this possible behavior of bmask redundant. @jameysharp also points out that we can use a combination of bmask and splat to get similar behavior as well, so there are many ways to work around bmask not supporting vector results.

Co-authored-by: Jamey Sharp <jsharp@fastly.com>

<!--

Please ensure that the following steps are all taken care of before submitting
the PR.

• [ ] This has been discussed in issue #..., or if not, please tell us why
  here.

• [ ] A short description of what this does, why it is needed; if the
  description becomes long, the matter should probably be discussed in an issue
  first.

• [ ] This PR contains test cases, if meaningful.

• [ ] A reviewer from the core maintainer team has been assigned for this PR.
  If you don't know who could review this, please indicate so. The list of
  suggested reviewers on the right can help you.

Please ensure all communication adheres to the code of conduct.
-->

Wasmtime GitHub notifications bot (Mar 07 2023 at 12:02):

afonso360 submitted PR review.

Wasmtime GitHub notifications bot (Mar 07 2023 at 18:14):

elliottt merged PR #5948.

Last updated: Nov 22 2024 at 17:03 UTC