salijagic a883b243b1 Merge branch 'linus/flopcount_inline_soa_avx2' into 'main' ...

flopcounts for basic avx2 with doubles

See merge request course-asl-fastcode26/team70!66

2026-06-14 16:03:19 +02:00

..

variant_AVX512_1.cpp

x86 measurements

2026-05-30 21:26:19 +02:00

variant_baseline.cpp

Baseline Flop Count Update, New Float Baseline

2026-05-23 20:50:59 +02:00

variant_branchless_sqrtfree.cpp

Add branchless w() variants sadly we had worse performance

2026-05-20 22:23:07 +02:00

variant_branchless.cpp

Add branchless w() variants sadly we had worse performance

2026-05-20 22:23:07 +02:00

variant_cutoff_callsite.cpp

Cutoff variant (skip far pedestrian pairs)

2026-05-21 19:16:40 +02:00

variant_cutoff.cpp

Cutoff variant (skip far pedestrian pairs)

2026-05-21 19:16:40 +02:00

variant_float_baseline.cpp

Baseline Flop Count Update, New Float Baseline

2026-05-23 20:50:59 +02:00

variant_float_inlining_SoA_AVX2_flopcount.cpp

flopcount for the initial float soa avx2 variant

2026-06-12 16:45:05 +02:00

variant_float_inlining_SoA_AVX2_polyexp_expfix.cpp

flopcount and expfix for soa_avx2_polyexp

2026-06-12 16:36:36 +02:00

variant_float_inlining_SoA_AVX2_polyexp_rsqrt_expfix.cpp

flopcount for the avx2_polyexp_rsqrt

2026-06-14 16:01:38 +02:00

variant_float_inlining_SoA_AVX2_polyexp_rsqrt_loopopt_expfix.cpp

flopcount of the best avx2 version

2026-06-14 16:01:17 +02:00

variant_float_inlining_SoA_AVX2_polyexp_rsqrt_loopopt.cpp

last minor loop optimization

2026-05-28 18:55:25 +02:00

variant_float_inlining_SoA_AVX2_polyexp_rsqrt.cpp

variant with rsqrt

2026-05-27 17:17:26 +02:00

variant_float_inlining_SoA_AVX2_polyexp.cpp

SoA implementation for float and AVX2 includind the polyexp fix

2026-05-25 12:07:46 +02:00

variant_float_inlining_SoA_AVX2.cpp

SoA implementation for float and AVX2 includind the polyexp fix

2026-05-25 12:07:46 +02:00

variant_float_inlining_SoA_AVX512_polyexp_fastinvsqrt.cpp

Added ILP after AVX-512

2026-05-28 22:07:18 +02:00

variant_float_inlining_SoA_AVX512_polyexp_moreavx_ilp_2_moreilp.cpp

Padding Fix + Float RSQRT Patch

2026-05-30 00:14:32 +02:00

variant_float_inlining_SoA_AVX512_polyexp_moreavx_ilp_2_rsqrt_morersqrt_expfix.cpp

Flop Count Correction

2026-06-09 00:01:08 +02:00

variant_float_inlining_SoA_AVX512_polyexp_moreavx_ilp_2_rsqrt_morersqrt.cpp

added comments to fastest version, maybe skim them

2026-06-01 14:28:14 +02:00

variant_float_inlining_SoA_AVX512_polyexp_moreavx_ilp_2_rsqrt.cpp

Padding Fix + Float RSQRT Patch

2026-05-30 00:14:32 +02:00

variant_float_inlining_SoA_AVX512_polyexp_moreavx_ilp_2.cpp

Padding Fix + Float RSQRT Patch

2026-05-30 00:14:32 +02:00

variant_float_inlining_SoA_AVX512_polyexp_moreavx_ilp_4.cpp

Padding Fix + Float RSQRT Patch

2026-05-30 00:14:32 +02:00

variant_float_inlining_SoA_AVX512_polyexp_moreavx.cpp

Added ILP after AVX-512

2026-05-28 22:07:18 +02:00

variant_float_inlining_SoA_AVX512_polyexp.cpp

Added ILP after AVX-512

2026-05-28 22:07:18 +02:00

variant_float_inlining_SoA_NEON_polyexp_moreavx_ilp_2_rsqrt_morrsqrt.cpp

Flop Count Correction

2026-06-09 00:01:08 +02:00

variant_float_inlining_SoA_symmetry_ILP_2.cpp

Added ILP after AVX-512

2026-05-28 22:07:18 +02:00

variant_float_inlining_SoA_symmetry_ILP_4.cpp

Added ILP after AVX-512

2026-05-28 22:07:18 +02:00

variant_float_inlining_SoA_symmetry_ILP_8.cpp

Added ILP after AVX-512

2026-05-28 22:07:18 +02:00

variant_float_inlining_SoA_symmetry_ILP_16.cpp

Added ILP after AVX-512

2026-05-28 22:07:18 +02:00

variant_float_inlining_SoA_symmetry.cpp

Symmetry

2026-05-27 09:33:36 +02:00

variant_float_inlining_SoA.cpp

SoA implementation for float and AVX2 includind the polyexp fix

2026-05-25 12:07:46 +02:00

variant_float_inlining.cpp

Baseline Flop Count Update, New Float Baseline

2026-05-23 20:50:59 +02:00

variant_inlining_7_SoA_AVX2_flopcount.cpp

flopcounts for basic avx2 with doubles

2026-06-12 17:27:54 +02:00

variant_inlining_7_SoA_AVX2_polyexp_expfix.cpp

flopcount for the double avx2 with polyexp

2026-06-14 16:03:03 +02:00

variant_inlining_7_SoA_AVX2_polyexp.cpp

AVX2 implementation

2026-05-25 11:43:58 +02:00

variant_inlining_7_SoA_AVX2.cpp

AVX2 implementation

2026-05-25 11:43:58 +02:00

variant_inlining_7_SoA.cpp

SoA implementation

2026-05-21 19:08:46 +02:00

variant_inlining_7_symmetry_1.cpp

implemented pair_symmetry that uses a LUT to calculate the distances

2026-05-20 10:29:17 +02:00

variant_inlining_7_symmetry_2_ilp_1.cpp

Ilp

2026-05-21 19:09:24 +02:00

variant_inlining_7_symmetry_2_ilp_2.cpp

Ilp

2026-05-21 19:09:24 +02:00

variant_inlining_7_symmetry_2.cpp

Ilp

2026-05-21 19:09:24 +02:00

variant_inlining_7.cpp

Inlining + Reorganization + Roofline

2026-05-19 23:17:44 +02:00

variant_inlining_8_SoA_AVX512_polyexp_fastinvsqrt_sr.cpp

Added ILP after AVX-512

2026-05-28 22:07:18 +02:00

variant_inlining_8_SoA_AVX512_polyexp_fastinvsqrt.cpp

Added ILP after AVX-512

2026-05-28 22:07:18 +02:00

variant_inlining_8_SoA_AVX512_polyexp_moreavx_ilp_2.cpp

Padding Fix + Float RSQRT Patch

2026-05-30 00:14:32 +02:00

variant_inlining_8_SoA_AVX512_polyexp_moreavx.cpp

Added ILP after AVX-512

2026-05-28 22:07:18 +02:00

variant_inlining_8.cpp

Baseline Flop Count Update, New Float Baseline

2026-05-23 20:50:59 +02:00

variant_lut_exp.cpp

Linus/variant lut

2026-05-20 22:55:47 +02:00