wasmtime / issue #9351 List of future optimizations for W... · git-wasmtime

Stream: git-wasmtime

Topic: wasmtime / issue #9351 List of future optimizations for W...

Wasmtime GitHub notifications bot (Oct 01 2024 at 16:54):

fitzgen added the cranelift:goal:optimize-speed label to Issue #9351.

Wasmtime GitHub notifications bot (Oct 01 2024 at 16:54):

fitzgen added the performance label to Issue #9351.

Wasmtime GitHub notifications bot (Oct 01 2024 at 16:54):

fitzgen added the wasmtime:ref-types label to Issue #9351.

Wasmtime GitHub notifications bot (Oct 01 2024 at 16:54):

fitzgen opened issue #9351:

[ ] Speed up getting and setting funcrefs inside GC objects: https://github.com/bytecodealliance/wasmtime/issues/9347

[ ] Const fold through uadd_overflow_trap/trap{z,nz} insts that show up frequently in GC heap accesses: https://github.com/bytecodealliance/wasmtime/issues/5908

[ ] Type-based alias analysis for Wasm GC: https://github.com/bytecodealliance/wasmtime/issues/9348

[ ] Escape analysis for Wasm GC: https://github.com/bytecodealliance/wasmtime/issues/9349

[ ] Reuse linear memories and their optimized bounds checks for Wasm GC heaps: https://github.com/bytecodealliance/wasmtime/issues/9350

Wasmtime GitHub notifications bot (Oct 01 2024 at 16:54):

github-actions[bot] commented on issue #9351:

Subscribe to Label Action

cc @fitzgen

<details>
This issue or pull request has been labeled: "wasmtime:ref-types"

Thus the following users have been cc'd because of the following labels:

fitzgen: wasmtime:ref-types

To subscribe or unsubscribe from this label, edit the <code>.github/subscribe-to-label.json</code> configuration file.

Learn more.
</details>

Wasmtime GitHub notifications bot (Oct 01 2024 at 16:55):

fitzgen edited issue #9351:

[ ] Speed up getting and setting funcrefs inside GC objects: https://github.com/bytecodealliance/wasmtime/issues/9347

[ ] Const fold through uadd_overflow_trap/trap{z,nz} insts that show up frequently in GC heap accesses: https://github.com/bytecodealliance/wasmtime/issues/5908

[ ] Also add a mid-end rule like (uadd_overflow_trap (uextend x) (uextend y)) ==> (iadd (uextend x) (uextend y))

[ ] Type-based alias analysis for Wasm GC: https://github.com/bytecodealliance/wasmtime/issues/9348

[ ] Escape analysis for Wasm GC: https://github.com/bytecodealliance/wasmtime/issues/9349

[ ] Reuse linear memories and their optimized bounds checks for Wasm GC heaps: https://github.com/bytecodealliance/wasmtime/issues/9350

Wasmtime GitHub notifications bot (Oct 01 2024 at 19:27):

fitzgen edited issue #9351:

[ ] Speed up getting and setting funcrefs inside GC objects: https://github.com/bytecodealliance/wasmtime/issues/9347

[ ] Const fold through uadd_overflow_trap/trap{z,nz} insts that show up frequently in GC heap accesses: https://github.com/bytecodealliance/wasmtime/issues/5908

[ ] Also add a mid-end rule like (uadd_overflow_trap (uextend x) (uextend y)) ==> (iadd (uextend x) (uextend y))

[ ] Type-based alias analysis for Wasm GC: https://github.com/bytecodealliance/wasmtime/issues/9348

[ ] Escape analysis for Wasm GC: https://github.com/bytecodealliance/wasmtime/issues/9349

[ ] Reuse linear memories and their optimized bounds checks for Wasm GC heaps: https://github.com/bytecodealliance/wasmtime/issues/9350
```[tasklist]

Tasks
~~~

Wasmtime GitHub notifications bot (Oct 01 2024 at 19:28):

fitzgen edited issue #9351:

[ ] Speed up getting and setting funcrefs inside GC objects: https://github.com/bytecodealliance/wasmtime/issues/9347

[ ] Const fold through uadd_overflow_trap/trap{z,nz} insts that show up frequently in GC heap accesses: https://github.com/bytecodealliance/wasmtime/issues/5908

[ ] Also add a mid-end rule like (uadd_overflow_trap (uextend x) (uextend y)) ==> (iadd (uextend x) (uextend y))

[ ] Type-based alias analysis for Wasm GC: https://github.com/bytecodealliance/wasmtime/issues/9348

[ ] Escape analysis for Wasm GC: https://github.com/bytecodealliance/wasmtime/issues/9349

[ ] Reuse linear memories and their optimized bounds checks for Wasm GC heaps: https://github.com/bytecodealliance/wasmtime/issues/9350
```[tasklist]

Tasks

[ ] https://github.com/bytecodealliance/wasmtime/issues/9352
~~~

Wasmtime GitHub notifications bot (Oct 01 2024 at 19:28):

fitzgen edited issue #9351:

[ ] Speed up getting and setting funcrefs inside GC objects: https://github.com/bytecodealliance/wasmtime/issues/9347

[ ] Const fold through uadd_overflow_trap/trap{z,nz} insts that show up frequently in GC heap accesses: https://github.com/bytecodealliance/wasmtime/issues/5908

[ ] Also add a mid-end rule like (uadd_overflow_trap (uextend x) (uextend y)) ==> (iadd (uextend x) (uextend y))

[ ] Type-based alias analysis for Wasm GC: https://github.com/bytecodealliance/wasmtime/issues/9348

[ ] Escape analysis for Wasm GC: https://github.com/bytecodealliance/wasmtime/issues/9349

[ ] Reuse linear memories and their optimized bounds checks for Wasm GC heaps: https://github.com/bytecodealliance/wasmtime/issues/9350
```[tasklist]

Tasks

[ ] https://github.com/bytecodealliance/wasmtime/issues/9352

[ ] https://github.com/bytecodealliance/wasmtime/issues/9350
~~~

Wasmtime GitHub notifications bot (Oct 01 2024 at 19:28):

fitzgen edited issue #9351:

[ ] Speed up getting and setting funcrefs inside GC objects: https://github.com/bytecodealliance/wasmtime/issues/9347

[ ] Const fold through uadd_overflow_trap/trap{z,nz} insts that show up frequently in GC heap accesses: https://github.com/bytecodealliance/wasmtime/issues/5908

[ ] Also add a mid-end rule like (uadd_overflow_trap (uextend x) (uextend y)) ==> (iadd (uextend x) (uextend y))

[ ] Type-based alias analysis for Wasm GC: https://github.com/bytecodealliance/wasmtime/issues/9348

[ ] Escape analysis for Wasm GC: https://github.com/bytecodealliance/wasmtime/issues/9349

[ ] Reuse linear memories and their optimized bounds checks for Wasm GC heaps: https://github.com/bytecodealliance/wasmtime/issues/9350
```[tasklist]

Tasks

[ ] https://github.com/bytecodealliance/wasmtime/issues/9352

[ ] https://github.com/bytecodealliance/wasmtime/issues/9350

[ ] https://github.com/bytecodealliance/wasmtime/issues/9349
~~~

Wasmtime GitHub notifications bot (Oct 01 2024 at 19:28):

fitzgen edited issue #9351:

[ ] Speed up getting and setting funcrefs inside GC objects: https://github.com/bytecodealliance/wasmtime/issues/9347

[ ] Const fold through uadd_overflow_trap/trap{z,nz} insts that show up frequently in GC heap accesses: https://github.com/bytecodealliance/wasmtime/issues/5908

[ ] Also add a mid-end rule like (uadd_overflow_trap (uextend x) (uextend y)) ==> (iadd (uextend x) (uextend y))

[ ] Type-based alias analysis for Wasm GC: https://github.com/bytecodealliance/wasmtime/issues/9348

[ ] Escape analysis for Wasm GC: https://github.com/bytecodealliance/wasmtime/issues/9349

[ ] Reuse linear memories and their optimized bounds checks for Wasm GC heaps: https://github.com/bytecodealliance/wasmtime/issues/9350
```[tasklist]

Tasks

[ ] https://github.com/bytecodealliance/wasmtime/issues/9352

[ ] https://github.com/bytecodealliance/wasmtime/issues/9350

[ ] https://github.com/bytecodealliance/wasmtime/issues/9349

[ ] https://github.com/bytecodealliance/wasmtime/issues/9348
~~~

Wasmtime GitHub notifications bot (Oct 01 2024 at 19:28):

fitzgen edited issue #9351:

[ ] Speed up getting and setting funcrefs inside GC objects: https://github.com/bytecodealliance/wasmtime/issues/9347

[ ] Const fold through uadd_overflow_trap/trap{z,nz} insts that show up frequently in GC heap accesses: https://github.com/bytecodealliance/wasmtime/issues/5908

[ ] Also add a mid-end rule like (uadd_overflow_trap (uextend x) (uextend y)) ==> (iadd (uextend x) (uextend y))

[ ] Type-based alias analysis for Wasm GC: https://github.com/bytecodealliance/wasmtime/issues/9348

[ ] Escape analysis for Wasm GC: https://github.com/bytecodealliance/wasmtime/issues/9349

[ ] Reuse linear memories and their optimized bounds checks for Wasm GC heaps: https://github.com/bytecodealliance/wasmtime/issues/9350
```[tasklist]

Tasks

[ ] https://github.com/bytecodealliance/wasmtime/issues/9352

[ ] https://github.com/bytecodealliance/wasmtime/issues/9350

[ ] https://github.com/bytecodealliance/wasmtime/issues/9349

[ ] https://github.com/bytecodealliance/wasmtime/issues/9348

[ ] https://github.com/bytecodealliance/wasmtime/issues/5908
~~~

Wasmtime GitHub notifications bot (Oct 01 2024 at 19:32):

fitzgen edited issue #9351:

This is a tracking issue collecting all of our ideas for improving Wasm GC performance and potential optimizations we could implement.

```[tasklist]

Tasks

[ ] https://github.com/bytecodealliance/wasmtime/issues/9347

[ ] https://github.com/bytecodealliance/wasmtime/issues/9352

[ ] https://github.com/bytecodealliance/wasmtime/issues/9350

[ ] https://github.com/bytecodealliance/wasmtime/issues/9349

[ ] https://github.com/bytecodealliance/wasmtime/issues/9348

[ ] https://github.com/bytecodealliance/wasmtime/issues/5908

[ ] Const fold through uadd_overflow_trap/trap{z,nz} insts that show up frequently in GC heap accesses. Depends on https://github.com/bytecodealliance/wasmtime/issues/5908

[ ] Also add a mid-end rule like (uadd_overflow_trap (uextend x) (uextend y)) ==> (iadd (uextend x) (uextend y)). Depends on https://github.com/bytecodealliance/wasmtime/issues/5908
~~~

Wasmtime GitHub notifications bot (Oct 07 2024 at 19:00):

fitzgen edited issue #9351:

This is a tracking issue collecting all of our ideas for improving Wasm GC performance and potential optimizations we could implement.

```[tasklist]

Tasks

[ ] https://github.com/bytecodealliance/wasmtime/issues/9347

[ ] https://github.com/bytecodealliance/wasmtime/issues/9352

[ ] https://github.com/bytecodealliance/wasmtime/issues/9350

[ ] https://github.com/bytecodealliance/wasmtime/issues/9349

[ ] https://github.com/bytecodealliance/wasmtime/issues/9348

[ ] https://github.com/bytecodealliance/wasmtime/issues/5908

[ ] Const fold through uadd_overflow_trap/trap{z,nz} insts that show up frequently in GC heap accesses. Depends on https://github.com/bytecodealliance/wasmtime/issues/5908

[ ] Also add a mid-end rule like (uadd_overflow_trap (uextend x) (uextend y)) ==> (iadd (uextend x) (uextend y)). Depends on https://github.com/bytecodealliance/wasmtime/issues/5908

[ ] Use bulk memcpy-style APIs to implement array.copy when possible and do GC barriers inline when not possible
~~~

Wasmtime GitHub notifications bot (Oct 09 2024 at 20:17):

fitzgen edited issue #9351:

This is a tracking issue collecting all of our ideas for improving Wasm GC performance and potential optimizations we could implement.

```[tasklist]

Tasks

[ ] https://github.com/bytecodealliance/wasmtime/issues/9347

[ ] https://github.com/bytecodealliance/wasmtime/issues/9352

[ ] https://github.com/bytecodealliance/wasmtime/issues/9350

[ ] https://github.com/bytecodealliance/wasmtime/issues/9349

[ ] https://github.com/bytecodealliance/wasmtime/issues/9348

[ ] https://github.com/bytecodealliance/wasmtime/issues/5908

[ ] Const fold through uadd_overflow_trap/trap{z,nz} insts that show up frequently in GC heap accesses. Depends on https://github.com/bytecodealliance/wasmtime/issues/5908

[ ] Also add a mid-end rule like (uadd_overflow_trap (uextend x) (uextend y)) ==> (iadd (uextend x) (uextend y)). Depends on https://github.com/bytecodealliance/wasmtime/issues/5908

[ ] Use bulk memcpy-style APIs to implement array.copy when possible and do GC barriers inline when not possible

[ ] Expose type registry's supertypes arrays to Wasm, so it can do the O(1) subtype checks inline
~~~

Wasmtime GitHub notifications bot (Oct 10 2024 at 15:27):

fitzgen edited issue #9351:

This is a tracking issue collecting all of our ideas for improving Wasm GC performance and potential optimizations we could implement.

```[tasklist]

Tasks

[ ] https://github.com/bytecodealliance/wasmtime/issues/9347

[ ] https://github.com/bytecodealliance/wasmtime/issues/9352

[ ] https://github.com/bytecodealliance/wasmtime/issues/9350

[ ] https://github.com/bytecodealliance/wasmtime/issues/9349

[ ] https://github.com/bytecodealliance/wasmtime/issues/9348

[ ] https://github.com/bytecodealliance/wasmtime/issues/5908

[ ] Const fold through uadd_overflow_trap/trap{z,nz} insts that show up frequently in GC heap accesses. Depends on https://github.com/bytecodealliance/wasmtime/issues/5908

[ ] Also add a mid-end rule like (uadd_overflow_trap (uextend x) (uextend y)) ==> (iadd (uextend x) (uextend y)). Depends on https://github.com/bytecodealliance/wasmtime/issues/5908

[ ] Use bulk memcpy-style APIs to implement array.copy when possible and do GC barriers inline when not possible

[ ] Expose type registry's supertypes arrays to Wasm, so it can do the O(1) subtype checks inline

[ ] Propagate trapz backwards from ref.cast into the branch-y instance-of check that it emits. Similar for the jumps inside a br_on_cast[_fail]'s instance-of check. See https://github.com/bytecodealliance/wasmtime/pull/9437#discussion_r1795656924 for details. Alternatively, clean this stuff up in the mid-end: https://github.com/bytecodealliance/wasmtime/issues/6106
~~~

Wasmtime GitHub notifications bot (Oct 10 2024 at 21:55):

fitzgen edited issue #9351:

This is a tracking issue collecting all of our ideas for improving Wasm GC performance and potential optimizations we could implement.

```[tasklist]

Tasks

[ ] https://github.com/bytecodealliance/wasmtime/issues/9347

[ ] https://github.com/bytecodealliance/wasmtime/issues/9352

[ ] https://github.com/bytecodealliance/wasmtime/issues/9350

[ ] https://github.com/bytecodealliance/wasmtime/issues/9349

[ ] https://github.com/bytecodealliance/wasmtime/issues/9348

[ ] https://github.com/bytecodealliance/wasmtime/issues/5908

[ ] Const fold through uadd_overflow_trap/trap{z,nz} insts that show up frequently in GC heap accesses. Depends on https://github.com/bytecodealliance/wasmtime/issues/5908

[ ] Also add a mid-end rule like (uadd_overflow_trap (uextend x) (uextend y)) ==> (iadd (uextend x) (uextend y)). Depends on https://github.com/bytecodealliance/wasmtime/issues/5908

[ ] Use bulk memcpy-style APIs to implement array.copy when possible and do GC barriers inline when not possible

[ ] Expose type registry's supertypes arrays to Wasm, so it can do the O(1) subtype checks inline

[ ] Propagate trapz backwards from ref.cast into the branch-y instance-of check that it emits. Similar for the jumps inside a br_on_cast[_fail]'s instance-of check. See https://github.com/bytecodealliance/wasmtime/pull/9437#discussion_r1795656924 for details. Alternatively, clean this stuff up in the mid-end: https://github.com/bytecodealliance/wasmtime/issues/6106

[ ] Speed up call_indirect in the face of subtyping. Much of this can be addressed by letting Wasm to the O(1) subtype checks inline, but it is still additional work to be done on every call_indirect vs MVP wasm. We should investigate inline caches and such again.
~~~

Wasmtime GitHub notifications bot (Oct 10 2024 at 21:59):

fitzgen commented on issue #9351:

See also https://github.com/WebAssembly/gc/issues/565 for potential benchmarks.

Wasmtime GitHub notifications bot (Feb 25 2025 at 17:37):

alexcrichton added the wasm-proposal:gc label to Issue #9351.

Wasmtime GitHub notifications bot (Mar 04 2025 at 23:21):

fitzgen edited issue #9351:

This is a tracking issue collecting all of our ideas for improving Wasm GC performance and potential optimizations we could implement.

[ ] https://github.com/bytecodealliance/wasmtime/issues/9347

[ ] https://github.com/bytecodealliance/wasmtime/issues/9352

[ ] https://github.com/bytecodealliance/wasmtime/issues/9350

[ ] https://github.com/bytecodealliance/wasmtime/issues/9349

[ ] https://github.com/bytecodealliance/wasmtime/issues/9348

[ ] https://github.com/bytecodealliance/wasmtime/issues/5908

[ ] Const fold through uadd_overflow_trap/trap{z,nz} insts that show up frequently in GC heap accesses. Depends on https://github.com/bytecodealliance/wasmtime/issues/5908

[ ] Also add a mid-end rule like (uadd_overflow_trap (uextend x) (uextend y)) ==> (iadd (uextend x) (uextend y)). Depends on https://github.com/bytecodealliance/wasmtime/issues/5908

[ ] Use bulk memcpy-style APIs to implement array.copy when possible and do GC barriers inline when not possible

[ ] Expose type registry's supertypes arrays to Wasm, so it can do the O(1) subtype checks inline

[ ] Propagate trapz backwards from ref.cast into the branch-y instance-of check that it emits. Similar for the jumps inside a br_on_cast[_fail]'s instance-of check. See https://github.com/bytecodealliance/wasmtime/pull/9437#discussion_r1795656924 for details. Alternatively, clean this stuff up in the mid-end: https://github.com/bytecodealliance/wasmtime/issues/6106

[ ] Speed up call_indirect in the face of subtyping. Much of this can be addressed by letting Wasm to the O(1) subtype checks inline, but it is still additional work to be done on every call_indirect vs MVP wasm. We should investigate inline caches and such again.

Wasmtime GitHub notifications bot (Apr 08 2025 at 18:30):

fitzgen edited issue #9351:

This is a tracking issue collecting all of our ideas for improving Wasm GC performance and potential optimizations we could implement.

[ ] https://github.com/bytecodealliance/wasmtime/issues/9347

[ ] https://github.com/bytecodealliance/wasmtime/issues/9352

[ ] https://github.com/bytecodealliance/wasmtime/issues/9350

[ ] https://github.com/bytecodealliance/wasmtime/issues/9349

[ ] https://github.com/bytecodealliance/wasmtime/issues/9348

[x] https://github.com/bytecodealliance/wasmtime/issues/5908

[ ] Const fold through uadd_overflow_trap/trap{z,nz} insts that show up frequently in GC heap accesses. Depends on https://github.com/bytecodealliance/wasmtime/issues/5908

[ ] Also add a mid-end rule like (uadd_overflow_trap (uextend x) (uextend y)) ==> (iadd (uextend x) (uextend y)). Depends on https://github.com/bytecodealliance/wasmtime/issues/5908

[ ] Use bulk memcpy-style APIs to implement array.copy when possible and do GC barriers inline when not possible

[ ] Expose type registry's supertypes arrays to Wasm, so it can do the O(1) subtype checks inline

[ ] Propagate trapz backwards from ref.cast into the branch-y instance-of check that it emits. Similar for the jumps inside a br_on_cast[_fail]'s instance-of check. See https://github.com/bytecodealliance/wasmtime/pull/9437#discussion_r1795656924 for details. Alternatively, clean this stuff up in the mid-end: https://github.com/bytecodealliance/wasmtime/issues/6106

[ ] Speed up call_indirect in the face of subtyping. Much of this can be addressed by letting Wasm to the O(1) subtype checks inline, but it is still additional work to be done on every call_indirect vs MVP wasm. We should investigate inline caches and such again.

Wasmtime GitHub notifications bot (Apr 11 2025 at 21:39):

fitzgen edited issue #9351:

This is a tracking issue collecting all of our ideas for improving Wasm GC performance and potential optimizations we could implement.

[ ] https://github.com/bytecodealliance/wasmtime/issues/9347

[ ] https://github.com/bytecodealliance/wasmtime/issues/9352

[x] https://github.com/bytecodealliance/wasmtime/issues/9350

[ ] https://github.com/bytecodealliance/wasmtime/issues/9349

[ ] https://github.com/bytecodealliance/wasmtime/issues/9348

[x] https://github.com/bytecodealliance/wasmtime/issues/5908

[ ] Const fold through uadd_overflow_trap/trap{z,nz} insts that show up frequently in GC heap accesses. Depends on https://github.com/bytecodealliance/wasmtime/issues/5908

[ ] Also add a mid-end rule like (uadd_overflow_trap (uextend x) (uextend y)) ==> (iadd (uextend x) (uextend y)). Depends on https://github.com/bytecodealliance/wasmtime/issues/5908

[ ] Use bulk memcpy-style APIs to implement array.copy when possible and do GC barriers inline when not possible

[ ] Expose type registry's supertypes arrays to Wasm, so it can do the O(1) subtype checks inline

[ ] Propagate trapz backwards from ref.cast into the branch-y instance-of check that it emits. Similar for the jumps inside a br_on_cast[_fail]'s instance-of check. See https://github.com/bytecodealliance/wasmtime/pull/9437#discussion_r1795656924 for details. Alternatively, clean this stuff up in the mid-end: https://github.com/bytecodealliance/wasmtime/issues/6106

[ ] Speed up call_indirect in the face of subtyping. Much of this can be addressed by letting Wasm to the O(1) subtype checks inline, but it is still additional work to be done on every call_indirect vs MVP wasm. We should investigate inline caches and such again.

Last updated: Apr 17 2025 at 16:03 UTC