Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Reland "[spirv][vulkan] Enable device query generation and execution" #16075

Merged
merged 5 commits into from
Jan 9, 2024

Conversation

antiagainst
Copy link
Contributor

@antiagainst antiagainst commented Jan 9, 2024

This commit relands #15977
with the following changes:

  • Disable the test requiring 16bit uniform buffer access on Adreno.
    Moto Edge X30 supports VK_KHR_16bit_storage for only storage
    buffers, but not uniform buffers, see
    https://vulkan.gpuinfo.org/displayreport.php?id=14481#features_extensions
    We request both bits.
  • Avoid requesting dot product to be accelerated. As long as the
    shaderIntegerDotProduct bit is true, we can use all the integer
    dot product instructions in SPIR-V. Whether it's accelerated or not
    does not matter for the purpose of deciding whether the SPIR-V blob
    is runnable.

@antiagainst antiagainst force-pushed the spirv-fix-adreno-f16 branch from ffe572c to 8a710e5 Compare January 9, 2024 04:48
As long as the `shaderIntegerDotProduct` bit is true, we can
use all the integer dot product instructions in SPIR-V.
Whether it's accelerated or not does not matter for the purpose
of deciding whether the SPIR-V blob is runnable.
@antiagainst
Copy link
Contributor Author

@antiagainst antiagainst added the benchmarks:android-gpu Run default Android GPU benchmarks label Jan 9, 2024
@antiagainst antiagainst changed the title [vulkan] Disable 16bit storage tests on Adreno GPUs Reland "[spirv][vulkan] Enable device query generation and execution" Jan 9, 2024
Copy link

github-actions bot commented Jan 9, 2024

Abbreviated Benchmark Summary

@ commit 8a7b30a73d838504f956288f3faca02aea8f01f9 (vs. base 282ab7758324d175179988d315edf6a2f38d5b0c)

Improved Latencies 🎉

Benchmark Name Average Latency (ms) Median Latency (ms) Latency Standard Deviation (ms)
MobileBertSquad\_int8(tflite) [arm-valhall-vulkan\_android31-vulkan\_spirv][experimental-flags,fuse-padding,max-concurrency] vulkan(none)[full-inference,default-flags] with default @ pixel-6-pro[gpu] 70.017 (vs. 74.318, 5.79%↓) 69.843 0.651

No improved or regressed compilation metrics 🏖️

For more information:

Source Workflow Run

@antiagainst antiagainst marked this pull request as ready for review January 9, 2024 16:30
@antiagainst
Copy link
Contributor Author

This is just relanding #15977, with cb36ec1 and 80b87bf to fix tests.

@antiagainst antiagainst requested review from kuhar and hanhanW January 9, 2024 16:32
@antiagainst antiagainst merged commit c8ecc1c into iree-org:main Jan 9, 2024
61 of 68 checks passed
@antiagainst antiagainst deleted the spirv-fix-adreno-f16 branch January 9, 2024 17:09
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
benchmarks:android-gpu Run default Android GPU benchmarks hal/vulkan Runtime Vulkan GPU HAL backend
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants