[connortsui20]

Tracking Issue: #10701
Based on develop, following #10989. Merge this before #11050 and let the develop benchmark runs finish so the new cases have a recorded baseline.

Summary

The new widening cases need baseline results before the loop optimization can be compared in CI. Add the benchmarks separately, running both named variants through the existing map_into implementation.

Changes

• Add stable widening benchmark names and sizes around the loop boundary, plus dense and nullable primitive casts.
• Return the output from the existing widening benchmark so its stores remain observable.
• Leave the implementation change to the stacked [EXPERIMENT] Keep ARM integer widening loops compact #11050.

API Changes

None.

[codspeed]

Merging this PR will not alter performance

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

✅ 2317 untouched benchmarks
🆕 192 new benchmarks
⏩ 1 skipped benchmark¹

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
🆕	Walltime	cast_i32_to_i64_dense_avx2[128]	N/A	861 ns	N/A
🆕	Walltime	cast_i32_to_i64_dense_avx2[129]	N/A	685 ns	N/A
🆕	Walltime	cast_i32_to_i64_dense_avx2[16384]	N/A	4.4 µs	N/A
🆕	Walltime	cast_i32_to_i64_dense_avx2[65]	N/A	837 ns	N/A
🆕	Walltime	cast_i32_to_i64_dense_avx2[8192]	N/A	2.5 µs	N/A
🆕	Walltime	cast_i32_to_i64_nullable_avx2[128]	N/A	776 ns	N/A
🆕	Walltime	cast_i32_to_i64_nullable_avx2[129]	N/A	779 ns	N/A
🆕	Walltime	cast_i32_to_i64_nullable_avx2[16384]	N/A	4.5 µs	N/A
🆕	Walltime	cast_i32_to_i64_nullable_avx2[65]	N/A	946 ns	N/A
🆕	Walltime	cast_i32_to_i64_nullable_avx2[8192]	N/A	2.5 µs	N/A
🆕	Walltime	cast_u16_to_u32_avx2[128]	N/A	751 ns	N/A
🆕	Walltime	cast_u16_to_u32_avx2[129]	N/A	753 ns	N/A
🆕	Walltime	cast_u16_to_u32_avx2[16384]	N/A	2.5 µs	N/A
🆕	Walltime	cast_u16_to_u32_avx2[65]	N/A	917 ns	N/A
🆕	Walltime	cast_u16_to_u32_avx2[8192]	N/A	1.6 µs	N/A
🆕	Walltime	cast_u16_to_u32_dense_avx2[128]	N/A	651 ns	N/A
🆕	Walltime	cast_u16_to_u32_dense_avx2[129]	N/A	654 ns	N/A
🆕	Walltime	cast_u16_to_u32_dense_avx2[16384]	N/A	2.4 µs	N/A
🆕	Walltime	cast_u16_to_u32_dense_avx2[65]	N/A	650 ns	N/A
🆕	Walltime	cast_u16_to_u32_dense_avx2[8192]	N/A	1.5 µs	N/A
...	...	...	...	...	...

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

---

_{Comparing ct/map-into-widening-benchmarks (dbfa820) with develop (fad0b7e)}

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. ↩