Skip to content

[EXPERIMENT] Keep ARM integer widening loops compact - #11050

Closed
connortsui20 wants to merge 1 commit into
ct/map-into-widening-benchmarksfrom
ct/map-into-widening
Closed

[EXPERIMENT] Keep ARM integer widening loops compact#11050
connortsui20 wants to merge 1 commit into
ct/map-into-widening-benchmarksfrom
ct/map-into-widening

Keep ARM integer widening loops compact

ee129e8
Select commit
Loading
Failed to load commit list.
CodSpeed / CodSpeed Performance Analysis failed Oct 30, 2026 in 1m 1s

Performance Regression: -11.44%

1 warning

⚠️ 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

⚡ 13 improved benchmarks
❌ 31 regressed benchmarks
✅ 2465 untouched benchmarks
⏩ 1 skipped benchmark1

Warning

Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

Mode Benchmark BASE HEAD Efficiency
❌ Walltime map_into_u16_u32_neon[1] 1 ns 2 ns -50%
❌ CPU Simulation filter_powerlaw_by_random[10000] 583.5 ns 1,074 ns -45.67%
❌ Walltime map_into_widening_u16_u32_neon[64] 5 ns 8 ns -37.5%
❌ Walltime map_into_u16_u32_neon[8192] 541 ns 796 ns -32.04%
❌ Walltime map_into_u16_u32_neon[64] 5 ns 7 ns -28.57%
❌ Walltime map_into_i32_i64_neon[128] 18 ns 25 ns -28%
❌ Walltime map_into_widening_i32_i64_neon[128] 18 ns 25 ns -28%
❌ Walltime map_into_i32_i64_neon[8192] 1.2 µs 1.6 µs -23.49%
❌ Walltime map_into_i32_i64_neon[129] 20 ns 26 ns -23.08%
❌ Walltime map_into_i32_i64_neon[64] 10 ns 13 ns -23.08%
❌ Walltime cast_i32_to_i64_dense_avx2[129] 685 ns 869 ns -21.17%
❌ Walltime map_into_i32_i64_neon[16384] 2.6 µs 3.2 µs -19.73%
❌ Walltime map_into_i32_i64_neon[16385] 2.6 µs 3.2 µs -19.45%
❌ CPU Simulation bench_compare_varbin[(50000, 2)] 133.4 µs 162.6 µs -17.95%
❌ CPU Simulation bench_compare_varbin[(50000, 64)] 135.2 µs 164.4 µs -17.75%
❌ Walltime map_into_u16_u32_neon[128] 11 ns 13 ns -15.38%
❌ Walltime map_into_widening_i32_i64_neon[64] 11 ns 13 ns -15.38%
❌ Walltime map_into_widening_u16_u32_neon[128] 11 ns 13 ns -15.38%
❌ Walltime map_into_u16_u32_neon[16384] 1.4 µs 1.6 µs -14.85%
❌ Walltime map_into_u16_u32_neon[16385] 1.4 µs 1.6 µs -14.79%
... ... ... ... ... ...

ℹ️ Only the first 20 benchmarks are displayed. Go to the app to view all benchmarks.

Tip

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

Comparing ct/map-into-widening (ee129e8) with ct/map-into-widening-benchmarks (dbfa820)

Open in CodSpeed

Footnotes

  1. 1 benchmark was skipped, so the baseline result was used instead. If it was deleted from the codebase, click here and archive it to remove it from the performance reports. ↩

View more details on CodSpeed