use barrier with LOCAL_MEM_FENCE

[vchuravy]

Companion to JuliaGPU/OpenCL.jl#300

[github-actions]

Benchmark Results

	main	d268c0c...	main/d268c0c3de014b...
saxpy/default/Float16/1024	0.0513 ± 0.025 ms	0.043 ± 0.025 ms	1.19
saxpy/default/Float16/1048576	0.894 ± 0.023 ms	0.893 ± 0.026 ms	1
saxpy/default/Float16/16384	0.0647 ± 0.028 ms	0.061 ± 0.028 ms	1.06
saxpy/default/Float16/2048	0.0592 ± 0.023 ms	0.0454 ± 0.018 ms	1.3
saxpy/default/Float16/256	0.0605 ± 0.027 ms	0.0404 ± 0.028 ms	1.5
saxpy/default/Float16/262144	0.272 ± 0.024 ms	0.269 ± 0.025 ms	1.01
saxpy/default/Float16/32768	0.0846 ± 0.029 ms	0.074 ± 0.028 ms	1.14
saxpy/default/Float16/4096	0.0675 ± 0.026 ms	0.0513 ± 0.027 ms	1.32
saxpy/default/Float16/512	0.0578 ± 0.026 ms	0.0412 ± 0.026 ms	1.4
saxpy/default/Float16/64	0.0608 ± 0.027 ms	0.0391 ± 0.028 ms	1.55
saxpy/default/Float16/65536	0.108 ± 0.029 ms	0.101 ± 0.027 ms	1.07
saxpy/default/Float32/1024	0.0437 ± 0.026 ms	0.0418 ± 0.027 ms	1.04
saxpy/default/Float32/1048576	0.476 ± 0.028 ms	0.472 ± 0.05 ms	1.01
saxpy/default/Float32/16384	0.0563 ± 0.026 ms	0.0525 ± 0.026 ms	1.07
saxpy/default/Float32/2048	0.0469 ± 0.023 ms	0.0436 ± 0.019 ms	1.08
saxpy/default/Float32/256	0.0539 ± 0.028 ms	0.0398 ± 0.028 ms	1.35
saxpy/default/Float32/262144	0.166 ± 0.035 ms	0.151 ± 0.035 ms	1.1
saxpy/default/Float32/32768	0.062 ± 0.027 ms	0.0583 ± 0.028 ms	1.06
saxpy/default/Float32/4096	0.0623 ± 0.025 ms	0.0486 ± 0.024 ms	1.28
saxpy/default/Float32/512	0.0523 ± 0.027 ms	0.0408 ± 0.028 ms	1.28
saxpy/default/Float32/64	0.0571 ± 0.028 ms	0.0397 ± 0.029 ms	1.44
saxpy/default/Float32/65536	0.082 ± 0.029 ms	0.0723 ± 0.029 ms	1.13
saxpy/default/Float64/1024	0.0493 ± 0.026 ms	0.0407 ± 0.026 ms	1.21
saxpy/default/Float64/1048576	0.513 ± 0.051 ms	0.502 ± 0.043 ms	1.02
saxpy/default/Float64/16384	0.0593 ± 0.026 ms	0.0532 ± 0.026 ms	1.12
saxpy/default/Float64/2048	0.0523 ± 0.023 ms	0.0433 ± 0.02 ms	1.21
saxpy/default/Float64/256	0.0592 ± 0.026 ms	0.0394 ± 0.028 ms	1.5
saxpy/default/Float64/262144	0.174 ± 0.029 ms	0.17 ± 0.03 ms	1.03
saxpy/default/Float64/32768	0.0647 ± 0.026 ms	0.0615 ± 0.026 ms	1.05
saxpy/default/Float64/4096	0.0639 ± 0.025 ms	0.0483 ± 0.024 ms	1.32
saxpy/default/Float64/512	0.0608 ± 0.026 ms	0.04 ± 0.028 ms	1.52
saxpy/default/Float64/64	0.0607 ± 0.026 ms	0.039 ± 0.029 ms	1.56
saxpy/default/Float64/65536	0.0908 ± 0.027 ms	0.0809 ± 0.027 ms	1.12
saxpy/static workgroup=(1024,)/Float16/1024	0.0438 ± 0.025 ms	0.0416 ± 0.025 ms	1.05
saxpy/static workgroup=(1024,)/Float16/1048576	0.902 ± 0.028 ms	0.903 ± 0.034 ms	1
saxpy/static workgroup=(1024,)/Float16/16384	0.0618 ± 0.027 ms	0.0583 ± 0.026 ms	1.06
saxpy/static workgroup=(1024,)/Float16/2048	0.0536 ± 0.023 ms	0.0461 ± 0.022 ms	1.16
saxpy/static workgroup=(1024,)/Float16/256	0.0531 ± 0.025 ms	0.04 ± 0.026 ms	1.33
saxpy/static workgroup=(1024,)/Float16/262144	0.27 ± 0.028 ms	0.266 ± 0.027 ms	1.02
saxpy/static workgroup=(1024,)/Float16/32768	0.0789 ± 0.027 ms	0.0712 ± 0.026 ms	1.11
saxpy/static workgroup=(1024,)/Float16/4096	0.0602 ± 0.026 ms	0.0484 ± 0.027 ms	1.24
saxpy/static workgroup=(1024,)/Float16/512	0.0514 ± 0.025 ms	0.0417 ± 0.026 ms	1.23
saxpy/static workgroup=(1024,)/Float16/64	0.0474 ± 0.026 ms	0.0397 ± 0.026 ms	1.19
saxpy/static workgroup=(1024,)/Float16/65536	0.111 ± 0.027 ms	0.0983 ± 0.026 ms	1.12
saxpy/static workgroup=(1024,)/Float32/1024	0.0413 ± 0.025 ms	0.0402 ± 0.026 ms	1.03
saxpy/static workgroup=(1024,)/Float32/1048576	0.464 ± 0.024 ms	0.456 ± 0.05 ms	1.02
saxpy/static workgroup=(1024,)/Float32/16384	0.0521 ± 0.026 ms	0.0507 ± 0.026 ms	1.03
saxpy/static workgroup=(1024,)/Float32/2048	0.0464 ± 0.023 ms	0.0419 ± 0.017 ms	1.11
saxpy/static workgroup=(1024,)/Float32/256	0.0608 ± 0.025 ms	0.0398 ± 0.026 ms	1.53
saxpy/static workgroup=(1024,)/Float32/262144	0.161 ± 0.035 ms	0.148 ± 0.035 ms	1.09
saxpy/static workgroup=(1024,)/Float32/32768	0.0578 ± 0.027 ms	0.0569 ± 0.028 ms	1.02
saxpy/static workgroup=(1024,)/Float32/4096	0.0573 ± 0.026 ms	0.0451 ± 0.026 ms	1.27
saxpy/static workgroup=(1024,)/Float32/512	0.0577 ± 0.025 ms	0.041 ± 0.026 ms	1.41
saxpy/static workgroup=(1024,)/Float32/64	0.0569 ± 0.025 ms	0.0418 ± 0.026 ms	1.36
saxpy/static workgroup=(1024,)/Float32/65536	0.0772 ± 0.029 ms	0.0694 ± 0.029 ms	1.11
saxpy/static workgroup=(1024,)/Float64/1024	0.0425 ± 0.025 ms	0.0402 ± 0.026 ms	1.06
saxpy/static workgroup=(1024,)/Float64/1048576	0.505 ± 0.046 ms	0.508 ± 0.048 ms	0.994
saxpy/static workgroup=(1024,)/Float64/16384	0.0548 ± 0.027 ms	0.0531 ± 0.025 ms	1.03
saxpy/static workgroup=(1024,)/Float64/2048	0.0451 ± 0.024 ms	0.0429 ± 0.021 ms	1.05
saxpy/static workgroup=(1024,)/Float64/256	0.0564 ± 0.025 ms	0.0399 ± 0.026 ms	1.41
saxpy/static workgroup=(1024,)/Float64/262144	0.173 ± 0.028 ms	0.166 ± 0.031 ms	1.04
saxpy/static workgroup=(1024,)/Float64/32768	0.0647 ± 0.027 ms	0.0619 ± 0.026 ms	1.04
saxpy/static workgroup=(1024,)/Float64/4096	0.0592 ± 0.026 ms	0.0457 ± 0.026 ms	1.3
saxpy/static workgroup=(1024,)/Float64/512	0.0569 ± 0.025 ms	0.0406 ± 0.025 ms	1.4
saxpy/static workgroup=(1024,)/Float64/64	0.059 ± 0.026 ms	0.0411 ± 0.026 ms	1.43
saxpy/static workgroup=(1024,)/Float64/65536	0.0894 ± 0.028 ms	0.0797 ± 0.028 ms	1.12
time_to_load	1.15 ± 0.0023 s	1.11 ± 0.015 s	1.04

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).

[codecov]

Codecov Report

Attention: Patch coverage is 0% with 1 line in your changes missing coverage. Please review.

Project coverage is 47.52%. Comparing base (de5dc99) to head (d268c0c).
Report is 1 commits behind head on main.

Files with missing lines	Patch %	Lines
src/pocl/backend.jl	0.00%	1 Missing ⚠️

Additional details and impacted files

@@           Coverage Diff           @@
##             main     #591   +/-   ##
=======================================
  Coverage   47.52%   47.52%           
=======================================
  Files          22       22           
  Lines        1715     1715           
=======================================
  Hits          815      815           
  Misses        900      900           

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

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.