gemma.cpp/ops
Jan Wassenberg cf4d7ceb82 1.16x decode speedup: remove last MatVec in Attention
Precompute row pointers.
Remove no longer used MHA support; QStride -> qkv_dim.
Remove RowPtr from MatMul interface, use only MatPtrT.
Require opt-in define for NUQ to speed up builds.
Also fix io.cc on Windows.

PiperOrigin-RevId: 766228108
2025-06-02 09:40:29 -07:00
..
bench_matmul.cc 1.16x decode speedup: remove last MatVec in Attention 2025-06-02 09:40:29 -07:00
dot-inl.h Huge refactor of weight handling and model loading. 2025-05-06 04:44:21 -07:00
dot_test.cc Remove backprop/ 2025-05-28 07:01:17 -07:00
fp_arith-inl.h Decouple MatMul from gemma-inl: precompile for all input types 2025-05-27 07:08:58 -07:00
gemma_matvec_test.cc Huge refactor of weight handling and model loading. 2025-05-06 04:44:21 -07:00
matmul-inl.h 1.16x decode speedup: remove last MatVec in Attention 2025-06-02 09:40:29 -07:00
matmul.cc Fix RowT issue and improve Griffin (currently still broken) 2025-05-19 07:02:10 -07:00
matmul.h 1.16x decode speedup: remove last MatVec in Attention 2025-06-02 09:40:29 -07:00
matmul_static-inl.h 1.16x decode speedup: remove last MatVec in Attention 2025-06-02 09:40:29 -07:00
matmul_static.h 1.16x decode speedup: remove last MatVec in Attention 2025-06-02 09:40:29 -07:00
matmul_static_bf16.cc Decouple MatMul from gemma-inl: precompile for all input types 2025-05-27 07:08:58 -07:00
matmul_static_f32.cc Decouple MatMul from gemma-inl: precompile for all input types 2025-05-27 07:08:58 -07:00
matmul_static_nuq.cc Decouple MatMul from gemma-inl: precompile for all input types 2025-05-27 07:08:58 -07:00
matmul_static_sfp.cc Decouple MatMul from gemma-inl: precompile for all input types 2025-05-27 07:08:58 -07:00
matmul_test.cc 1.16x decode speedup: remove last MatVec in Attention 2025-06-02 09:40:29 -07:00
matvec-inl.h Replace last ConstMat with MatPtr 2025-05-13 10:55:22 -07:00
ops-inl.h Remove backprop/ 2025-05-28 07:01:17 -07:00
ops.h Remove backprop/ 2025-05-28 07:01:17 -07:00
ops_test.cc Remove backprop/ 2025-05-28 07:01:17 -07:00
sum-inl.h Minor cleanup, Windows+Bazel build fixes 2024-10-10 09:05:06 -07:00