[0ax1]

Problem

Pinned-buffer reclamation allocates an extra completion event for each transfer.

Solution

Reuse the buffer owner's event to track inflight transfers.

Assessment

Preserves the source-retention guarantees and event flags from #10712, which must merge first.

[codspeed]

Merging this PR will regress 4 benchmarks

4 warnings, 1 note

⚠️ Walltime results may be unreliable

Some benchmarks used the Walltime instrument outside CodSpeed Macro Runners, which can make the results inconsistent.

Learn more about Macro Runners

See affected benchmarks

⚠️ Unable to measure the execution of 1 benchmark

This benchmark measured no execution time, so its result is not comparable and counts as unchanged. This usually happens when the compiler removes the code under test.

Prevent compiler optimizations

See affected benchmarks

⚠️ 20 benchmarks are dominated by system calls

System calls cannot be simulated consistently and are left out of the measure. The reported value understates the real cost. Switch to the Walltime instrument to measure them.

Learn more about system calls

See affected benchmarks

⚠️ Base and head ran on different runner types

Some benchmarks with significant performance changes were compared across different runner types, so the changes might not come from the code. Run base and head on the same runner type, or use CodSpeed Macro Runners for more stable results.

Learn more about Macro Runners

Open the report in CodSpeed to investigate

ℹ️ 40 benchmarks have no flamegraph

The flamegraph wasn't generated: the symbols of the profiling trace could not be resolved, because no information about the language frames was collected.

See affected benchmarks

⚡ 15 improved benchmarks
❌ 4 regressed benchmarks
✅ 2389 untouched benchmarks
⏩ 12 skipped benchmarks¹
🗄️ 43 archived benchmarks run²

Warning

Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
❌	CPU Simulation	bench_compare_sliced_dict_primitive[(2000, 10000)]	13.1 µs	68.9 µs	-80.97%
❌	CPU Simulation	density_sweep_dense_runs[0.001]	44.5 µs	61.9 µs	-28%
❌	Walltime	starts_with_avx512[64, VarBin]	53.6 µs	62.2 µs	-13.83%
❌	CPU Simulation	random_i8[0.01]	103.5 µs	116.6 µs	-11.27%
⚡	CPU Simulation	prepare[1000]	430.3 µs	10.5 µs	×41
⚡	CPU Simulation	prepare[100]	74.8 µs	10.5 µs	×7.1
⚡	CPU Simulation	decompress[u32, (1000, 256)]	52.2 µs	31.5 µs	+65.73%
⚡	CPU Simulation	prepare[1]	16.5 µs	10.5 µs	+57.2%
⚡	CPU Simulation	scan[1]	95.6 µs	76.2 µs	+25.39%
⚡	CPU Simulation	prepare_and_compile[100]	323 µs	260.7 µs	+23.9%
⚡	CPU Simulation	prepare_and_compile[1000]	2.7 ms	2.2 ms	+21.38%
⚡	CPU Simulation	test_repeated_scan_scalar_at[int_float]	450.6 µs	371.3 µs	+21.36%
⚡	CPU Simulation	test_scan_filter_to_arrow_array[int_float]	4.1 ms	3.6 ms	+13.57%
⚡	CPU Simulation	test_scan_filter[int]	3.4 ms	3 ms	+12.87%
⚡	CPU Simulation	test_scan_filter_to_arrow_array[int]	2.8 ms	2.5 ms	+11.15%
⚡	CPU Simulation	test_repeated_scan_filter[int]	2.7 ms	2.5 ms	+11.03%
⚡	CPU Simulation	decompress[u64, (1000, 256)]	51 µs	46.2 µs	+10.52%
⚡	CPU Simulation	prepare_and_compile[1]	52.7 µs	47.8 µs	+10.34%
⚡	CPU Simulation	test_repeated_scan_filter[int_float]	3.9 ms	3.5 ms	+10.06%

Tip

Investigate this regression by commenting @codspeedbot fix this regression on this PR, or directly use the CodSpeed MCP with your agent.

---

_{Comparing ad/cuda-pinned-event-reuse (81e4221) with ad/cuda-pinned-transfer-safety (40f13e2)³}

Footnotes

1. 12 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩

2. 43 benchmarks were run, but are now archived. If they were deleted in another branch, consider rebasing to remove them from the report. Instead if they were added back, click here to restore them. ↩

3. No successful run was found on ad/cuda-pinned-transfer-safety (cd9ea47) during the generation of this report, so 159322e was used instead as the comparison base. There might be some changes unrelated to this pull request in this report. ↩