Skip to content

Implement a CPU backend using POCL #556

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.

Sign up for GitHub

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

Jump to bottom
Merged
merged 1 commit into from
Feb 17, 2025
Merged

Implement a CPU backend using POCL #556

merged 1 commit into from
Feb 17, 2025
+2,544 −411

Conversation

@vchuravy
Copy link
Member

@vchuravy vchuravy commented Jan 15, 2025 •
edited
Loading

TODO:

@github-actions
Copy link
Contributor

github-actions bot commented Jan 15, 2025 •
edited
Loading

Benchmark Results

main 0dcdc8b... main/0dcdc8b2f1e506...
saxpy/default/Float16/1024 0.0568 ± 0.026 ms 0.0527 ± 0.026 ms 1.08
saxpy/default/Float16/1048576 0.893 ± 0.025 ms 0.89 ± 0.025 ms 1
saxpy/default/Float16/16384 0.0675 ± 0.028 ms 0.0616 ± 0.026 ms 1.09
saxpy/default/Float16/2048 0.0599 ± 0.023 ms 0.0504 ± 0.023 ms 1.19
saxpy/default/Float16/256 0.0635 ± 0.027 ms 0.0473 ± 0.027 ms 1.34
saxpy/default/Float16/262144 0.272 ± 0.026 ms 0.268 ± 0.026 ms 1.01
saxpy/default/Float16/32768 0.0776 ± 0.028 ms 0.075 ± 0.027 ms 1.03
saxpy/default/Float16/4096 0.0647 ± 0.025 ms 0.0558 ± 0.025 ms 1.16
saxpy/default/Float16/512 0.0604 ± 0.026 ms 0.047 ± 0.027 ms 1.28
saxpy/default/Float16/64 0.064 ± 0.027 ms 0.0571 ± 0.028 ms 1.12
saxpy/default/Float16/65536 0.112 ± 0.028 ms 0.102 ± 0.026 ms 1.1
saxpy/default/Float32/1024 0.0535 ± 0.026 ms 0.0522 ± 0.026 ms 1.03
saxpy/default/Float32/1048576 0.478 ± 0.026 ms 0.47 ± 0.057 ms 1.02
saxpy/default/Float32/16384 0.0588 ± 0.027 ms 0.0558 ± 0.025 ms 1.05
saxpy/default/Float32/2048 0.0546 ± 0.024 ms 0.0474 ± 0.024 ms 1.15
saxpy/default/Float32/256 0.0616 ± 0.026 ms 0.052 ± 0.028 ms 1.18
saxpy/default/Float32/262144 0.168 ± 0.035 ms 0.157 ± 0.035 ms 1.07
saxpy/default/Float32/32768 0.0643 ± 0.027 ms 0.0614 ± 0.025 ms 1.05
saxpy/default/Float32/4096 0.0639 ± 0.025 ms 0.0516 ± 0.025 ms 1.24
saxpy/default/Float32/512 0.06 ± 0.026 ms 0.0484 ± 0.028 ms 1.24
saxpy/default/Float32/64 0.0631 ± 0.026 ms 0.048 ± 0.028 ms 1.31
saxpy/default/Float32/65536 0.0838 ± 0.029 ms 0.0753 ± 0.029 ms 1.11
saxpy/default/Float64/1024 0.0579 ± 0.026 ms 0.0465 ± 0.026 ms 1.24
saxpy/default/Float64/1048576 0.512 ± 0.051 ms 0.502 ± 0.045 ms 1.02
saxpy/default/Float64/16384 0.0604 ± 0.027 ms 0.0569 ± 0.026 ms 1.06
saxpy/default/Float64/2048 0.0523 ± 0.024 ms 0.0475 ± 0.023 ms 1.1
saxpy/default/Float64/256 0.0613 ± 0.026 ms 0.0553 ± 0.028 ms 1.11
saxpy/default/Float64/262144 0.176 ± 0.029 ms 0.174 ± 0.03 ms 1.01
saxpy/default/Float64/32768 0.0666 ± 0.027 ms 0.0647 ± 0.026 ms 1.03
saxpy/default/Float64/4096 0.0627 ± 0.025 ms 0.0506 ± 0.025 ms 1.24
saxpy/default/Float64/512 0.0606 ± 0.026 ms 0.0473 ± 0.028 ms 1.28
saxpy/default/Float64/64 0.0607 ± 0.027 ms 0.0469 ± 0.028 ms 1.3
saxpy/default/Float64/65536 0.0917 ± 0.027 ms 0.0835 ± 0.027 ms 1.1
saxpy/static workgroup=(1024,)/Float16/1024 0.0535 ± 0.026 ms 0.0459 ± 0.026 ms 1.17
saxpy/static workgroup=(1024,)/Float16/1048576 0.901 ± 0.028 ms 0.901 ± 0.031 ms 1
saxpy/static workgroup=(1024,)/Float16/16384 0.0613 ± 0.026 ms 0.0596 ± 0.024 ms 1.03
saxpy/static workgroup=(1024,)/Float16/2048 0.0575 ± 0.024 ms 0.0501 ± 0.024 ms 1.15
saxpy/static workgroup=(1024,)/Float16/256 0.0631 ± 0.026 ms 0.049 ± 0.027 ms 1.29
saxpy/static workgroup=(1024,)/Float16/262144 0.272 ± 0.028 ms 0.267 ± 0.027 ms 1.02
saxpy/static workgroup=(1024,)/Float16/32768 0.0774 ± 0.026 ms 0.0721 ± 0.024 ms 1.07
saxpy/static workgroup=(1024,)/Float16/4096 0.0631 ± 0.027 ms 0.052 ± 0.027 ms 1.21
saxpy/static workgroup=(1024,)/Float16/512 0.057 ± 0.026 ms 0.0442 ± 0.026 ms 1.29
saxpy/static workgroup=(1024,)/Float16/64 0.0598 ± 0.026 ms 0.0494 ± 0.027 ms 1.21
saxpy/static workgroup=(1024,)/Float16/65536 0.107 ± 0.026 ms 0.101 ± 0.025 ms 1.07
saxpy/static workgroup=(1024,)/Float32/1024 0.0556 ± 0.026 ms 0.0545 ± 0.026 ms 1.02
saxpy/static workgroup=(1024,)/Float32/1048576 0.465 ± 0.028 ms 0.459 ± 0.048 ms 1.01
saxpy/static workgroup=(1024,)/Float32/16384 0.0551 ± 0.026 ms 0.0526 ± 0.024 ms 1.05
saxpy/static workgroup=(1024,)/Float32/2048 0.0511 ± 0.023 ms 0.0458 ± 0.023 ms 1.12
saxpy/static workgroup=(1024,)/Float32/256 0.0629 ± 0.025 ms 0.0506 ± 0.026 ms 1.24
saxpy/static workgroup=(1024,)/Float32/262144 0.161 ± 0.035 ms 0.151 ± 0.034 ms 1.07
saxpy/static workgroup=(1024,)/Float32/32768 0.0605 ± 0.027 ms 0.058 ± 0.025 ms 1.04
saxpy/static workgroup=(1024,)/Float32/4096 0.0598 ± 0.026 ms 0.0475 ± 0.026 ms 1.26
saxpy/static workgroup=(1024,)/Float32/512 0.0612 ± 0.025 ms 0.0467 ± 0.026 ms 1.31
saxpy/static workgroup=(1024,)/Float32/64 0.0617 ± 0.025 ms 0.0508 ± 0.026 ms 1.22
saxpy/static workgroup=(1024,)/Float32/65536 0.079 ± 0.029 ms 0.0714 ± 0.028 ms 1.11
saxpy/static workgroup=(1024,)/Float64/1024 0.0581 ± 0.026 ms 0.0465 ± 0.026 ms 1.25
saxpy/static workgroup=(1024,)/Float64/1048576 0.504 ± 0.041 ms 0.497 ± 0.041 ms 1.02
saxpy/static workgroup=(1024,)/Float64/16384 0.0566 ± 0.026 ms 0.0545 ± 0.025 ms 1.04
saxpy/static workgroup=(1024,)/Float64/2048 0.0525 ± 0.024 ms 0.0464 ± 0.024 ms 1.13
saxpy/static workgroup=(1024,)/Float64/256 0.0594 ± 0.025 ms 0.0492 ± 0.026 ms 1.21
saxpy/static workgroup=(1024,)/Float64/262144 0.173 ± 0.03 ms 0.169 ± 0.031 ms 1.02
saxpy/static workgroup=(1024,)/Float64/32768 0.0652 ± 0.026 ms 0.062 ± 0.025 ms 1.05
saxpy/static workgroup=(1024,)/Float64/4096 0.0596 ± 0.026 ms 0.0485 ± 0.026 ms 1.23
saxpy/static workgroup=(1024,)/Float64/512 0.0587 ± 0.026 ms 0.0453 ± 0.026 ms 1.3
saxpy/static workgroup=(1024,)/Float64/64 0.06 ± 0.025 ms 0.0445 ± 0.026 ms 1.35
saxpy/static workgroup=(1024,)/Float64/65536 0.0901 ± 0.028 ms 0.08 ± 0.028 ms 1.13
time_to_load 1.12 ± 0.012 s 1.12 ± 0.01 s 0.998

Benchmark Plots

A plot of the benchmark results have been uploaded as an artifact to the workflow run for this PR.
Go to "Actions"->"Benchmark a pull request"->[the most recent run]->"Artifacts" (at the bottom).

@VarLad VarLad mentioned this pull request Jan 27, 2025
Closed
@codecov
Copy link

codecov bot commented Jan 28, 2025 •
edited
Loading

Codecov Report

Attention: Patch coverage is 0% with 879 lines in your changes missing coverage. Please review.

Project coverage is 0.00%. Comparing base (da32c2b) to head (0dcdc8b).
Report is 2 commits behind head on main.

Files with missing lines Patch % Lines
src/pocl/nanoOpenCL.jl 0.00% 517 Missing ⚠️
src/pocl/device/array.jl 0.00% 101 Missing ⚠️
src/pocl/backend.jl 0.00% 93 Missing ⚠️
src/pocl/compiler/execution.jl 0.00% 43 Missing ⚠️
src/pocl/compiler/compilation.jl 0.00% 32 Missing ⚠️
src/pocl/device/quirks.jl 0.00% 24 Missing ⚠️
src/pocl/compiler/reflection.jl 0.00% 23 Missing ⚠️
src/pocl/pocl.jl 0.00% 20 Missing ⚠️
src/pocl/device/runtime.jl 0.00% 13 Missing ⚠️
src/macros.jl 0.00% 6 Missing ⚠️
... and 2 more
Additional details and impacted files 
@@          Coverage Diff           @@
##            main    #556    +/-   ##
======================================
  Coverage   0.00%   0.00%            
======================================
  Files         12      21     +9     
  Lines        796    1584   +788     
======================================
- Misses       796    1584   +788

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

@vchuravy vchuravy mentioned this pull request Feb 3, 2025
Draft
@vchuravy vchuravy changed the base branch from main to vc/barriers February 4, 2025 15:18
@vchuravy vchuravy changed the base branch from vc/barriers to main February 5, 2025 12:29
@vchuravy Graphite App
Copy link
Member Author

vchuravy commented Feb 5, 2025 •
edited
Loading

This stack of pull requests is managed by Graphite. Learn more about stacking.

@vchuravy vchuravy mentioned this pull request Feb 5, 2025
Closed
@vchuravy vchuravy marked this pull request as ready for review February 5, 2025 12:43
github-actions[bot]
github-actions bot reviewed Feb 5, 2025
View reviewed changes
@vchuravy vchuravy force-pushed the vc/pocl branch 8 times, most recently from 6fb1cea to 8098378 Compare February 6, 2025 14:48
github-actions[bot]
github-actions bot reviewed Feb 6, 2025
View reviewed changes
github-actions[bot]
github-actions bot reviewed Feb 6, 2025
View reviewed changes
@vchuravy vchuravy force-pushed the vc/pocl branch 3 times, most recently from ba4eee7 to a6ae55b Compare February 7, 2025 11:31
This was referenced Feb 7, 2025
Merged
Merged
@vchuravy vchuravy force-pushed the vc/pocl branch 2 times, most recently from 90a10d7 to f038d8c Compare February 10, 2025 15:08
vchuravy
vchuravy commented Feb 11, 2025
View reviewed changes
src/pocl/device/quirks.jl
end

# math.jl
@device_override @noinline Base.Math.throw_complex_domainerror(f::Symbol, x) =
Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@maleadt this will be fun.

@vchuravy vchuravy force-pushed the vc/pocl branch 4 times, most recently from 0121280 to ed2ee63 Compare February 14, 2025 15:31
@vchuravy Graphite App
Copy link
Member Author

vchuravy commented Feb 17, 2025 •
edited
Loading

Merge activity

  • Feb 17, 7:42 AM EST: A user started a stack merge that includes this pull request via Graphite.
  • Feb 17, 7:43 AM EST: A user merged this pull request with Graphite.

@vchuravy vchuravy merged commit 45b74d9 into main Feb 17, 2025
8 of 23 checks passed
@vchuravy vchuravy deleted the vc/pocl branch February 17, 2025 12:43
@vchuravy
Copy link
Member Author

Welp... This got merged by graphite to soon.

Now we have a release-0.9 branch and main is now for 0.10

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@github-actions github-actions[bot] github-actions[bot] left review comments

Assignees

No one assigned

Labels

None yet

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

2 participants

@vchuravy