Stream: git-wasmtime

Topic: wasmtime / PR #4514 x64: Shrink Inst from 72 to 48 bytes

Wasmtime GitHub notifications bot (Jul 22 2022 at 20:49):

elliottt opened PR #4514 from trevor/box-jump-table-vectors to main:

• Box jump table vectors as small vectors
• Switch to a small vec of 4 elements

<!--

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 (Jul 22 2022 at 21:39):

elliottt updated PR #4514 from trevor/box-jump-table-vectors to main.

Wasmtime GitHub notifications bot (Jul 22 2022 at 21:40):

elliottt edited PR #4514 from trevor/box-jump-table-vectors to main:

This PR pulls out the effective parts of #4496, boxing the tables in JmpTableSeq. The performance is roughly the same, but sequential runs show a very small improvement in memory usage. The parameter to the SmallVec used in BoxVecMachLabel could do with some tuning: 4 performed better than 16, but I haven't experimented further.

Sightglass spidermonkey benchmark

compilation :: nanoseconds :: benchmarks/spidermonkey/benchmark.wasm

  No difference in performance.

  [2850802574 2907543428.87 2955488180] branch.so
  [2849459537 2907931719.73 2970915233] main.so

compilation :: cycles :: benchmarks/spidermonkey/benchmark.wasm

  No difference in performance.

  [10263234120 10467497988.27 10640099020] branch.so
  [10258359048 10468881874.20 10695627810] main.so

Memory usage on the spidermonkey benchmark

Running the spidermonkey 10 times:

for i in $(seq 1 10); do
  /usr/bin/time -f "%M" -- taskset 0x1 ~/wasmtime/target/release/wasmtime compile benchmark.wasm
done

branch	main	trevor/box-jump-table-vectors
	105288	105288
	105600	105496
	105340	105396
	105508	105148
	105268	105340
	105596	105396
	105464	105428
	105368	105464
	105340	105248
	105296	105240
average(k)	105406	105344

<!--

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 (Jul 22 2022 at 21:40):

elliottt edited PR #4514 from trevor/box-jump-table-vectors to main:

This PR pulls out the effective parts of #4496, boxing the tables in JmpTableSeq. The performance is roughly the same, but sequential runs show a very small improvement in memory usage. The parameter to the SmallVec used in BoxVecMachLabel could do with some tuning: 4 performed better than 16, but I haven't experimented further.

Sightglass spidermonkey benchmark

compilation :: nanoseconds :: benchmarks/spidermonkey/benchmark.wasm

  Δ = 21407759.30 ± 19781778.63 (confidence = 99%)

  branch.so is 1.00x to 1.01x faster than main.so!
  main.so is 0.99x to 1.00x faster than branch.so!

  [2845686334 2898004956.07 2938721676] branch.so
  [2867347959 2919412715.37 2976040194] main.so

compilation :: cycles :: benchmarks/spidermonkey/benchmark.wasm

  Δ = 77065090.93 ± 71216617.01 (confidence = 99%)

  branch.so is 1.00x to 1.01x faster than main.so!
  main.so is 0.99x to 1.00x faster than branch.so!

  [10244830916 10433179969.40 10579750838] branch.so
  [10322803658 10510245060.33 10714117852] main.so

Memory usage on the spidermonkey benchmark

Running the spidermonkey 10 times:

for i in $(seq 1 10); do
  /usr/bin/time -f "%M" -- taskset 0x1 ~/wasmtime/target/release/wasmtime compile benchmark.wasm
done

branch	main	trevor/box-jump-table-vectors
	105288	105288
	105600	105496
	105340	105396
	105508	105148
	105268	105340
	105596	105396
	105464	105428
	105368	105464
	105340	105248
	105296	105240
average(k)	105406	105344

<!--

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 (Jul 22 2022 at 21:55):

cfallin submitted PR review.

Wasmtime GitHub notifications bot (Jul 22 2022 at 22:47):

elliottt updated PR #4514 from trevor/box-jump-table-vectors to main.

Wasmtime GitHub notifications bot (Jul 22 2022 at 22:59):

elliottt updated PR #4514 from trevor/box-jump-table-vectors to main.

Wasmtime GitHub notifications bot (Jul 22 2022 at 23:05):

elliottt updated PR #4514 from trevor/box-jump-table-vectors to main.

Wasmtime GitHub notifications bot (Jul 22 2022 at 23:47):

elliottt updated PR #4514 from trevor/box-jump-table-vectors to main.

Wasmtime GitHub notifications bot (Jul 22 2022 at 23:50):

elliottt updated PR #4514 from trevor/box-jump-table-vectors to main.

Wasmtime GitHub notifications bot (Jul 23 2022 at 00:30):

elliottt requested cfallin for a review on PR #4514.

Wasmtime GitHub notifications bot (Jul 25 2022 at 18:10):

elliottt has marked PR #4514 as ready for review.

Wasmtime GitHub notifications bot (Jul 25 2022 at 18:15):

elliottt updated PR #4514 from trevor/box-jump-table-vectors to main.

Wasmtime GitHub notifications bot (Jul 25 2022 at 18:18):

elliottt updated PR #4514 from trevor/box-jump-table-vectors to main.

Wasmtime GitHub notifications bot (Jul 27 2022 at 17:32):

cfallin submitted PR review.

Wasmtime GitHub notifications bot (Jul 27 2022 at 17:39):

elliottt merged PR #4514.

Last updated: Oct 23 2024 at 20:03 UTC