azwolski

Case A – the functionality is written directly in the code body. Case B – the function is not marked inline. Case C – the function is explicitly marked inline....

Regressions | target | benchmark | time_ref_ns | time_exp_ns | speedup | | --- | --- | --- | --- | --- | | avx2-i32x8 | popcnt_stdlib_varying_int32_even/8192 | 8279 |...

Reference: avx2-i64x4 ```ispc __declspec(safe) static inline int popcnt(int64 v) { int r; for (uniform int i = 0; i < programCount; ++i) r = insert(r, i, popcnt(extract(v, i))); return __mask...

## AVX512ICL-x8 | target | benchmark | time_ref_ns | time_exp_ns | speedup | |---------------|----------------------------------------------|-------------|-------------|---------| | avx512icl-x8 | popcnt_stdlib_varying_int64_all/8192 | 17272 | 4396 | 3.92903 | | avx512icl-x8 | popcnt_stdlib_varying_int64_even/8192 |...

| target | benchmark | time_ref_ns | time_exp_ns | speedup | |------------------|----------------------------------------------|-------------|-------------|----------| | avx512icl-x16 | popcnt_stdlib_varying_int64_all/8192 | 17298 | 2035 | 8.50025 | | avx512icl-x16 | popcnt_stdlib_varying_int64_even/8192 | 8656 |...