Skip to content

optimize Disjoint Pool chunks #1147

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 2 commits into from
Mar 7, 2025
Merged

optimize Disjoint Pool chunks #1147

merged 2 commits into from
Mar 7, 2025

Conversation

bratpiorka
Copy link
Contributor

@bratpiorka bratpiorka commented Feb 28, 2025 •
edited
Loading

This PR replaces the bool array for Disjoint Pool chunks with bit fields.
Additionally, if there are fewer than 64 chunks in the slab, the array is not allocated at all, and a plain uint64_t is used to track all chunks.

It also includes a partial cleanup of the bit manipulation functions: ctzll/_BitScanForward64 and clzll/_BitScanReverse64.

With this change, the Disjoint Pool benchmarks show performance gains of 8% to 15%.

@bratpiorka bratpiorka force-pushed the rrudnick_dp_bit_chunk branch 2 times, most recently from 4948203 to be15237 Compare February 28, 2025 17:04
@github-actions GitHub Actions
Copy link

Compute Benchmarks run (with params: ):
https://github.com/oneapi-src/unified-memory-framework/actions/runs/13592912748

@bratpiorka bratpiorka force-pushed the rrudnick_dp_bit_chunk branch from be15237 to dbd931d Compare February 28, 2025 17:23
@github-actions GitHub Actions
Copy link

Compute Benchmarks run ():
https://github.com/oneapi-src/unified-memory-framework/actions/runs/13592912748
Job status: success. Test status: success.

Summary

(Emphasized values are the best results)

Improved 5 (threshold 2.00%)
Benchmark This PR baseline Change
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:4 scalable_pool<os_provider> 1904.180000 ns 2211.130 ns 16.12%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:1 disjoint_pool<os_provider> 974.179000 ns 1122.300 ns 15.20%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:128/granularity:8/iterations:500000/threads:1 disjoint_pool<os_provider> 788.576000 ns 862.505 ns 9.38%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:4 disjoint_pool<os_provider> 33102.100000 ns 35989.400 ns 8.72%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:4096/granularity:8/iterations:500000/threads:1 disjoint_pool<os_provider> 2412.720000 ns 2619.400 ns 8.57%
Regressed 7 (threshold 2.00%)
Benchmark This PR baseline Change
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:4096/granularity:8/iterations:500000/threads:4 scalable_pool<os_provider> 2197.670 ns 1643.510000 ns -25.22%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:128/granularity:8/iterations:500000/threads:4 scalable_pool<os_provider> 1743.110 ns 1320.200000 ns -24.26%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:4 glibc 1233.200 ns 1123.860000 ns -8.87%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:4096/granularity:8/iterations:500000/threads:4 glibc 2270.250 ns 2116.650000 ns -6.77%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:4 jemalloc_pool<os_provider> 47428.100 ns 45280.800000 ns -4.53%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:4096/granularity:8/iterations:500000/threads:4 jemalloc_pool<os_provider> 9679.350 ns 9416.900000 ns -2.71%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:4096/granularity:8/iterations:500000/threads:4 umfProxy 5158.370 ns 5052.000000 ns -2.06%

Performance change in benchmark groups

UMF
Relative perf in group multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:4 (5)
Benchmark This PR baseline Change
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:4 scalable_pool<os_provider> 1904.180000 ns 2211.130 ns 16.12%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:4 disjoint_pool<os_provider> 33102.100000 ns 35989.400 ns 8.72%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:4 umfProxy 5316.470 ns 5273.470000 ns -0.81%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:4 jemalloc_pool<os_provider> 47428.100 ns 45280.800000 ns -4.53%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:4 glibc 1233.200 ns 1123.860000 ns -8.87%
Relative perf in group multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:1 (5)
Benchmark This PR baseline Change
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:1 disjoint_pool<os_provider> 974.179000 ns 1122.300 ns 15.20%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:1 umfProxy 1051.950 ns 1049.970000 ns -0.19%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:1 scalable_pool<os_provider> 688.779 ns 684.613000 ns -0.60%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:1 glibc 368.543 ns 365.628000 ns -0.79%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:1 jemalloc_pool<os_provider> 1648.730 ns 1617.920000 ns -1.87%
Relative perf in group multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:4096/granularity:8/iterations:500000/threads:4 (4)
Benchmark This PR baseline Change
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:4096/granularity:8/iterations:500000/threads:4 umfProxy 5158.370 ns 5052.000000 ns -2.06%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:4096/granularity:8/iterations:500000/threads:4 jemalloc_pool<os_provider> 9679.350 ns 9416.900000 ns -2.71%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:4096/granularity:8/iterations:500000/threads:4 glibc 2270.250 ns 2116.650000 ns -6.77%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:4096/granularity:8/iterations:500000/threads:4 scalable_pool<os_provider> 2197.670 ns 1643.510000 ns -25.22%
Relative perf in group multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:4096/granularity:8/iterations:500000/threads:1 (5)
Benchmark This PR baseline Change
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:4096/granularity:8/iterations:500000/threads:1 disjoint_pool<os_provider> 2412.720000 ns 2619.400 ns 8.57%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:4096/granularity:8/iterations:500000/threads:1 umfProxy 1082.610000 ns 1085.020 ns 0.22%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:4096/granularity:8/iterations:500000/threads:1 jemalloc_pool<os_provider> 1136.000000 ns 1136.270 ns 0.02%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:4096/granularity:8/iterations:500000/threads:1 scalable_pool<os_provider> 632.267 ns 630.761000 ns -0.24%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:4096/granularity:8/iterations:500000/threads:1 glibc 845.042 ns 840.324000 ns -0.56%
Relative perf in group multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:128/granularity:8/iterations:500000/threads:4 (4)
Benchmark This PR baseline Change
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:128/granularity:8/iterations:500000/threads:4 jemalloc_pool<os_provider> 8932.650000 ns 8973.420 ns 0.46%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:128/granularity:8/iterations:500000/threads:4 glibc 952.177000 ns 953.113 ns 0.10%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:128/granularity:8/iterations:500000/threads:4 umfProxy 4919.710 ns 4821.430000 ns -2.00%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:128/granularity:8/iterations:500000/threads:4 scalable_pool<os_provider> 1743.110 ns 1320.200000 ns -24.26%
Relative perf in group multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:128/granularity:8/iterations:500000/threads:1 (5)
Benchmark This PR baseline Change
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:128/granularity:8/iterations:500000/threads:1 disjoint_pool<os_provider> 788.576000 ns 862.505 ns 9.38%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:128/granularity:8/iterations:500000/threads:1 umfProxy 708.293000 ns 714.140 ns 0.83%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:128/granularity:8/iterations:500000/threads:1 scalable_pool<os_provider> 378.976000 ns 379.301 ns 0.09%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:128/granularity:8/iterations:500000/threads:1 glibc 176.377 ns 174.965000 ns -0.80%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:128/granularity:8/iterations:500000/threads:1 jemalloc_pool<os_provider> 699.538 ns 692.916000 ns -0.95%
Relative perf in group multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:50000/threads:1 (2)
Benchmark This PR baseline Change
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:50000/threads:1 os_provider 9368.930000 ns 9443.800 ns 0.80%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:50000/threads:1 proxy_pool<os_provider> 11142.800 ns 11066.700000 ns -0.68%

Details

Benchmark details - environment, command...
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:4 glibc

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:1 glibc

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:4096/granularity:8/iterations:500000/threads:4 glibc

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:4096/granularity:8/iterations:500000/threads:1 glibc

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:128/granularity:8/iterations:500000/threads:4 glibc

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:128/granularity:8/iterations:500000/threads:1 glibc

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:50000/threads:1 proxy_pool

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:50000/threads:1 os_provider

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:4 disjoint_pool

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:1 disjoint_pool

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:4096/granularity:8/iterations:500000/threads:1 disjoint_pool

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:128/granularity:8/iterations:500000/threads:1 disjoint_pool

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:4 jemalloc_pool

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:1 jemalloc_pool

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:4096/granularity:8/iterations:500000/threads:4 jemalloc_pool

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:4096/granularity:8/iterations:500000/threads:1 jemalloc_pool

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:128/granularity:8/iterations:500000/threads:4 jemalloc_pool

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:128/granularity:8/iterations:500000/threads:1 jemalloc_pool

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:4 scalable_pool

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:1 scalable_pool

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:4096/granularity:8/iterations:500000/threads:4 scalable_pool

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:4096/granularity:8/iterations:500000/threads:1 scalable_pool

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:128/granularity:8/iterations:500000/threads:4 scalable_pool

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:128/granularity:8/iterations:500000/threads:1 scalable_pool

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:4 umfProxy

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv --benchmark_filter=glibc

Environment Variables:

LD_PRELOAD=/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/lib/libumf_proxy.so

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:1 umfProxy

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv --benchmark_filter=glibc

Environment Variables:

LD_PRELOAD=/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/lib/libumf_proxy.so

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:4096/granularity:8/iterations:500000/threads:4 umfProxy

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv --benchmark_filter=glibc

Environment Variables:

LD_PRELOAD=/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/lib/libumf_proxy.so

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:4096/granularity:8/iterations:500000/threads:1 umfProxy

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv --benchmark_filter=glibc

Environment Variables:

LD_PRELOAD=/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/lib/libumf_proxy.so

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:128/granularity:8/iterations:500000/threads:4 umfProxy

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv --benchmark_filter=glibc

Environment Variables:

LD_PRELOAD=/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/lib/libumf_proxy.so

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:128/granularity:8/iterations:500000/threads:1 umfProxy

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv --benchmark_filter=glibc

Environment Variables:

LD_PRELOAD=/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/lib/libumf_proxy.so

@bratpiorka bratpiorka marked this pull request as ready for review March 3, 2025 13:36
@bratpiorka bratpiorka requested a review from a team as a code owner March 3, 2025 13:36
igchor
igchor reviewed Mar 3, 2025
View reviewed changes
@bratpiorka bratpiorka force-pushed the rrudnick_dp_bit_chunk branch 2 times, most recently from d9bf2b4 to a9fdb11 Compare March 3, 2025 16:02
igchor
igchor reviewed Mar 3, 2025
View reviewed changes
pbalcer
pbalcer reviewed Mar 4, 2025
View reviewed changes
@bratpiorka bratpiorka force-pushed the rrudnick_dp_bit_chunk branch 4 times, most recently from 855bf22 to f6c9611 Compare March 5, 2025 16:09
@github-actions GitHub Actions
Copy link

github-actions bot commented Mar 5, 2025

Compute Benchmarks run (with params: ):
https://github.com/oneapi-src/unified-memory-framework/actions/runs/13680399478

@github-actions GitHub Actions
Copy link

github-actions bot commented Mar 5, 2025

Compute Benchmarks run ():
https://github.com/oneapi-src/unified-memory-framework/actions/runs/13680399478
Job status: success. Test status: success.

Summary

(Emphasized values are the best results)

Improved 12 (threshold 2.00%)
Benchmark This PR baseline Change
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:4 scalable_pool<os_provider> 963.261000 ns 1942.030 ns 101.61%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:128/granularity:8/iterations:500000/threads:4 glibc 509.171000 ns 960.520 ns 88.64%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:128/granularity:8/iterations:500000/threads:1 disjoint_pool<os_provider> 586.828000 ns 908.131 ns 54.75%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:4096/granularity:8/iterations:500000/threads:4 glibc 1530.600000 ns 2318.180 ns 51.46%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:4 disjoint_pool<os_provider> 31349.400000 ns 36902.700 ns 17.71%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:1 disjoint_pool<os_provider> 976.839000 ns 1120.990 ns 14.76%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:4 jemalloc_pool<os_provider> 46375.300000 ns 52257.600 ns 12.68%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:1 scalable_pool<os_provider> 618.088000 ns 689.122 ns 11.49%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:4096/granularity:8/iterations:500000/threads:1 disjoint_pool<os_provider> 2354.220000 ns 2608.570 ns 10.80%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:4096/granularity:8/iterations:500000/threads:4 scalable_pool<os_provider> 2152.340000 ns 2360.380 ns 9.67%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:128/granularity:8/iterations:500000/threads:4 scalable_pool<os_provider> 1729.970000 ns 1875.200 ns 8.39%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:50000/threads:1 os_provider 9319.390000 ns 9524.580 ns 2.20%
Regressed 5 (threshold 2.00%)
Benchmark This PR baseline Change
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:4 glibc 2186.410 ns 1228.360000 ns -43.82%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:4 umfProxy 5959.370 ns 5108.020000 ns -14.29%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:4096/granularity:8/iterations:500000/threads:4 umfProxy 5718.480 ns 4984.990000 ns -12.83%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:128/granularity:8/iterations:500000/threads:4 umfProxy 5349.260 ns 4690.560000 ns -12.31%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:128/granularity:8/iterations:500000/threads:1 umfProxy 720.943 ns 699.311000 ns -3.00%

Performance change in benchmark groups

UMF
Relative perf in group multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:4 (9)
Benchmark This PR baseline Change
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:4 scalable_pool<os_provider> 963.261000 ns 1942.030 ns 101.61%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:4 disjoint_pool<os_provider> 31349.400000 ns 36902.700 ns 17.71%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:4 jemalloc_pool<os_provider> 46375.300000 ns 52257.600 ns 12.68%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:4 umfProxy 5959.370 ns 5108.020000 ns -14.29%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:4 glibc 2186.410 ns 1228.360000 ns -43.82%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:4 disjoint_pool<fixed_provider> 31221.400000 ns -
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:4 scalable_pool<fixed_provider> 1766.800000 ns -
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:4 jemalloc 480.397000 ns -
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:4 tbbProxy 1876.430000 ns -
Relative perf in group multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:1 (11)
Benchmark This PR baseline Change
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:1 disjoint_pool<os_provider> 976.839000 ns 1120.990 ns 14.76%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:1 scalable_pool<os_provider> 618.088000 ns 689.122 ns 11.49%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:1 jemalloc_pool<os_provider> 1649.500000 ns 1671.160 ns 1.31%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:1 glibc 366.379000 ns 370.501 ns 1.13%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:1 umfProxy 1053.360 ns 1047.630000 ns -0.54%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:1 proxy_pool<fixed_provider> 3806.520000 ns -
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:1 fixed_provider 2106.600000 ns -
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:1 disjoint_pool<fixed_provider> 929.142000 ns -
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:1 scalable_pool<fixed_provider> 599.999000 ns -
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:1 jemalloc 72.564200 ns -
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:1 tbbProxy 498.765000 ns -
Relative perf in group multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:4096/granularity:8/iterations:500000/threads:4 (8)
Benchmark This PR baseline Change
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:4096/granularity:8/iterations:500000/threads:4 glibc 1530.600000 ns 2318.180 ns 51.46%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:4096/granularity:8/iterations:500000/threads:4 scalable_pool<os_provider> 2152.340000 ns 2360.380 ns 9.67%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:4096/granularity:8/iterations:500000/threads:4 jemalloc_pool<os_provider> 9459.200 ns 9435.310000 ns -0.25%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:4096/granularity:8/iterations:500000/threads:4 umfProxy 5718.480 ns 4984.990000 ns -12.83%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:4096/granularity:8/iterations:500000/threads:4 disjoint_pool<os_provider> 24864.900000 ns -
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:4096/granularity:8/iterations:500000/threads:4 scalable_pool<fixed_provider> 2076.050000 ns -
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:4096/granularity:8/iterations:500000/threads:4 jemalloc 571.814000 ns -
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:4096/granularity:8/iterations:500000/threads:4 tbbProxy 2158.970000 ns -
Relative perf in group multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:4096/granularity:8/iterations:500000/threads:1 (9)
Benchmark This PR baseline Change
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:4096/granularity:8/iterations:500000/threads:1 disjoint_pool<os_provider> 2354.220000 ns 2608.570 ns 10.80%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:4096/granularity:8/iterations:500000/threads:1 jemalloc_pool<os_provider> 1126.140000 ns 1145.630 ns 1.73%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:4096/granularity:8/iterations:500000/threads:1 glibc 842.207000 ns 842.870 ns 0.08%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:4096/granularity:8/iterations:500000/threads:1 scalable_pool<os_provider> 633.643 ns 632.682000 ns -0.15%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:4096/granularity:8/iterations:500000/threads:1 umfProxy 1086.340 ns 1082.810000 ns -0.32%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:4096/granularity:8/iterations:500000/threads:1 disjoint_pool<fixed_provider> 1499.870000 ns -
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:4096/granularity:8/iterations:500000/threads:1 scalable_pool<fixed_provider> 634.158000 ns -
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:4096/granularity:8/iterations:500000/threads:1 jemalloc 155.797000 ns -
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:4096/granularity:8/iterations:500000/threads:1 tbbProxy 553.952000 ns -
Relative perf in group multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:128/granularity:8/iterations:500000/threads:4 (8)
Benchmark This PR baseline Change
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:128/granularity:8/iterations:500000/threads:4 glibc 509.171000 ns 960.520 ns 88.64%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:128/granularity:8/iterations:500000/threads:4 scalable_pool<os_provider> 1729.970000 ns 1875.200 ns 8.39%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:128/granularity:8/iterations:500000/threads:4 jemalloc_pool<os_provider> 8912.240000 ns 9030.120 ns 1.32%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:128/granularity:8/iterations:500000/threads:4 umfProxy 5349.260 ns 4690.560000 ns -12.31%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:128/granularity:8/iterations:500000/threads:4 disjoint_pool<os_provider> 10337.000000 ns -
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:128/granularity:8/iterations:500000/threads:4 scalable_pool<fixed_provider> 1679.500000 ns -
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:128/granularity:8/iterations:500000/threads:4 jemalloc 512.560000 ns -
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:128/granularity:8/iterations:500000/threads:4 tbbProxy 1293.550000 ns -
Relative perf in group multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:128/granularity:8/iterations:500000/threads:1 (9)
Benchmark This PR baseline Change
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:128/granularity:8/iterations:500000/threads:1 disjoint_pool<os_provider> 586.828000 ns 908.131 ns 54.75%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:128/granularity:8/iterations:500000/threads:1 jemalloc_pool<os_provider> 697.594000 ns 703.540 ns 0.85%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:128/granularity:8/iterations:500000/threads:1 scalable_pool<os_provider> 378.899 ns 378.787000 ns -0.03%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:128/granularity:8/iterations:500000/threads:1 glibc 178.975 ns 177.026000 ns -1.09%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:128/granularity:8/iterations:500000/threads:1 umfProxy 720.943 ns 699.311000 ns -3.00%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:128/granularity:8/iterations:500000/threads:1 disjoint_pool<fixed_provider> 587.563000 ns -
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:128/granularity:8/iterations:500000/threads:1 scalable_pool<fixed_provider> 379.233000 ns -
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:128/granularity:8/iterations:500000/threads:1 jemalloc 95.267100 ns -
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:128/granularity:8/iterations:500000/threads:1 tbbProxy 338.755000 ns -
Relative perf in group multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:50000/threads:1 (2)
Benchmark This PR baseline Change
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:50000/threads:1 os_provider 9319.390000 ns 9524.580 ns 2.20%
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:50000/threads:1 proxy_pool<os_provider> 11064.000 ns 11032.500000 ns -0.28%

Details

Benchmark details - environment, command...
multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:4 glibc

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:1 glibc

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:4096/granularity:8/iterations:500000/threads:4 glibc

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:4096/granularity:8/iterations:500000/threads:1 glibc

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:128/granularity:8/iterations:500000/threads:4 glibc

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:128/granularity:8/iterations:500000/threads:1 glibc

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:50000/threads:1 proxy_pool

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:50000/threads:1 os_provider

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:4 disjoint_pool

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:1 disjoint_pool

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:4096/granularity:8/iterations:500000/threads:4 disjoint_pool

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:4096/granularity:8/iterations:500000/threads:1 disjoint_pool

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:128/granularity:8/iterations:500000/threads:4 disjoint_pool

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:128/granularity:8/iterations:500000/threads:1 disjoint_pool

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:4 jemalloc_pool

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:1 jemalloc_pool

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:4096/granularity:8/iterations:500000/threads:4 jemalloc_pool

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:4096/granularity:8/iterations:500000/threads:1 jemalloc_pool

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:128/granularity:8/iterations:500000/threads:4 jemalloc_pool

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:128/granularity:8/iterations:500000/threads:1 jemalloc_pool

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:4 scalable_pool

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:1 scalable_pool

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:4096/granularity:8/iterations:500000/threads:4 scalable_pool

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:4096/granularity:8/iterations:500000/threads:1 scalable_pool

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:128/granularity:8/iterations:500000/threads:4 scalable_pool

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:128/granularity:8/iterations:500000/threads:1 scalable_pool

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:1 proxy_pool

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:1 fixed_provider

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:4 disjoint_pool

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:1 disjoint_pool

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:4096/granularity:8/iterations:500000/threads:1 disjoint_pool

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:128/granularity:8/iterations:500000/threads:1 disjoint_pool

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:4 scalable_pool

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:1 scalable_pool

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:4096/granularity:8/iterations:500000/threads:4 scalable_pool

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:4096/granularity:8/iterations:500000/threads:1 scalable_pool

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:128/granularity:8/iterations:500000/threads:4 scalable_pool

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:128/granularity:8/iterations:500000/threads:1 scalable_pool

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:4 umfProxy

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv --benchmark_filter=glibc

Environment Variables:

LD_PRELOAD=/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/lib/libumf_proxy.so

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:1 umfProxy

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv --benchmark_filter=glibc

Environment Variables:

LD_PRELOAD=/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/lib/libumf_proxy.so

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:4096/granularity:8/iterations:500000/threads:4 umfProxy

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv --benchmark_filter=glibc

Environment Variables:

LD_PRELOAD=/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/lib/libumf_proxy.so

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:4096/granularity:8/iterations:500000/threads:1 umfProxy

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv --benchmark_filter=glibc

Environment Variables:

LD_PRELOAD=/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/lib/libumf_proxy.so

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:128/granularity:8/iterations:500000/threads:4 umfProxy

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv --benchmark_filter=glibc

Environment Variables:

LD_PRELOAD=/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/lib/libumf_proxy.so

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:128/granularity:8/iterations:500000/threads:1 umfProxy

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv --benchmark_filter=glibc

Environment Variables:

LD_PRELOAD=/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/lib/libumf_proxy.so

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:4 jemalloc

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv --benchmark_filter=glibc

Environment Variables:

LD_PRELOAD=libjemalloc.so

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:1 jemalloc

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv --benchmark_filter=glibc

Environment Variables:

LD_PRELOAD=libjemalloc.so

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:4096/granularity:8/iterations:500000/threads:4 jemalloc

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv --benchmark_filter=glibc

Environment Variables:

LD_PRELOAD=libjemalloc.so

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:4096/granularity:8/iterations:500000/threads:1 jemalloc

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv --benchmark_filter=glibc

Environment Variables:

LD_PRELOAD=libjemalloc.so

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:128/granularity:8/iterations:500000/threads:4 jemalloc

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv --benchmark_filter=glibc

Environment Variables:

LD_PRELOAD=libjemalloc.so

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:128/granularity:8/iterations:500000/threads:1 jemalloc

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv --benchmark_filter=glibc

Environment Variables:

LD_PRELOAD=libjemalloc.so

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:4 tbbProxy

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv --benchmark_filter=glibc

Environment Variables:

LD_PRELOAD=libtbbmalloc_proxy.so

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/size:4096/iterations:500000/threads:1 tbbProxy

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv --benchmark_filter=glibc

Environment Variables:

LD_PRELOAD=libtbbmalloc_proxy.so

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:4096/granularity:8/iterations:500000/threads:4 tbbProxy

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv --benchmark_filter=glibc

Environment Variables:

LD_PRELOAD=libtbbmalloc_proxy.so

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:4096/granularity:8/iterations:500000/threads:1 tbbProxy

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv --benchmark_filter=glibc

Environment Variables:

LD_PRELOAD=libtbbmalloc_proxy.so

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:128/granularity:8/iterations:500000/threads:4 tbbProxy

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv --benchmark_filter=glibc

Environment Variables:

LD_PRELOAD=libtbbmalloc_proxy.so

multiple_malloc_free/max_allocs:10000/thread_local_allocations:1/min_size:8/max_size:128/granularity:8/iterations:500000/threads:1 tbbProxy

Command:

/home/test-user/actions-runners/umf-perf-runner/_work/unified-memory-framework/unified-memory-framework/umf-repo/build/benchmark/umf-benchmark --benchmark_format=csv --benchmark_filter=glibc

Environment Variables:

LD_PRELOAD=libtbbmalloc_proxy.so

@bratpiorka bratpiorka requested a review from pbalcer March 6, 2025 08:37
@bratpiorka
Copy link
Contributor Author

@lplewa please check the benchmark results

pbalcer
pbalcer reviewed Mar 6, 2025
View reviewed changes
src/pool/pool_disjoint.c Outdated
size_t free_chunk = i * CHUNK_BITMAP_SIZE + bit_index;
if (free_chunk < slab->num_chunks_total) {
return free_chunk;
}
Copy link
Contributor

Choose a reason for hiding this comment

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

else return SIZE_MAX ? otherwise this would iterate over the remaining bitmap that we know is past slab->num_chunks_total.

Copy link
Contributor

Choose a reason for hiding this comment

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

In this case it is last iteration anyway, so it should fine anyway.

Copy link
Contributor

Choose a reason for hiding this comment

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

It may not be though...
I spotted this: https://github.com/oneapi-src/unified-memory-framework/pull/1147/files/f6c9611edcb835cb6dad69c61556fecccceebfdb#diff-7d792b84303f5dbbfba19a0b300b3645a0455df415f4246568dc0b1a5b846c73R103-R107

So we shouldn't need this if at all. if the bit is set, it can be allocated.

Copy link
Contributor Author

Choose a reason for hiding this comment

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

ok, removed

pbalcer
pbalcer approved these changes Mar 6, 2025
View reviewed changes
Copy link
Contributor

@pbalcer pbalcer left a comment

Choose a reason for hiding this comment

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

lgtm, 1 nit

@bratpiorka bratpiorka force-pushed the rrudnick_dp_bit_chunk branch 2 times, most recently from eb7e17b to d693070 Compare March 7, 2025 11:24
@bratpiorka bratpiorka merged commit 92da225 into oneapi-src:main Mar 7, 2025
236 of 239 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@pbalcer pbalcer pbalcer approved these changes

@igchor igchor igchor approved these changes

@vinser52 vinser52 Awaiting requested review from vinser52

@ldorau ldorau Awaiting requested review from ldorau

@lplewa lplewa Awaiting requested review from lplewa

Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
4 participants
@bratpiorka @pbalcer @lplewa @igchor