diff options
| author | Gian Marco <gianmarco.iodice@arm.com> | 2017-11-21 10:57:50 +0000 |
|---|---|---|
| committer | Anthony Barbier <anthony.barbier@arm.com> | 2018-11-02 16:41:17 +0000 |
| commit | 05288a2b871ef99f544771621c3bba409b2f70df (patch) | |
| tree | 21e3d2a9927ef31f6d5bcdd5523c4c8e933047a6 /src/core/NEON/kernels/NEGEMMLowpReductionKernel.cpp | |
| parent | c82799003fbfdc5bb9526ff944e41eaae23e3f03 (diff) | |
| download | ComputeLibrary-05288a2b871ef99f544771621c3bba409b2f70df.tar.gz | |
COMPMID-697 - Rework GEMMLowp interface on OpenCL
Reworked the interface of GemmLowp in order to make easy the integration
in Android NN
- Added support for different output stage
- Added validation for both matrix multiplication and output stage
- Added bounded relu support in the output stage
- Added in32_t bias support
- Added optimized path for vector by matrix case
This rework is required for:
- Convolution quantized
- Fully connected quantized
Change-Id: I512283d406099cf8c614dd89d0a97ed411143afc
Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/110625
Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
Tested-by: BSG Visual Compute Jenkins server to access repositories on http://mpd-gerrit.cambridge.arm.com <bsgcomp@arm.com>
Diffstat (limited to 'src/core/NEON/kernels/NEGEMMLowpReductionKernel.cpp')
| -rw-r--r-- | src/core/NEON/kernels/NEGEMMLowpReductionKernel.cpp | 2 |
1 files changed, 1 insertions, 1 deletions
diff --git a/src/core/NEON/kernels/NEGEMMLowpReductionKernel.cpp b/src/core/NEON/kernels/NEGEMMLowpReductionKernel.cpp index a8395a15cb..81094f8743 100644 --- a/src/core/NEON/kernels/NEGEMMLowpReductionKernel.cpp +++ b/src/core/NEON/kernels/NEGEMMLowpReductionKernel.cpp @@ -209,7 +209,7 @@ void NEGEMMLowpMatrixAReductionKernel::run(const Window &window, const ThreadInf uint32x4_t sum_row_u32 = vdupq_n_u32(0); uint32_t sum_row = 0; - const uint8_t *matrix_a = (in.ptr() + id.x() * _input->info()->strides_in_bytes()[1] + +id.y() * _input->info()->strides_in_bytes()[2]); + const uint8_t *matrix_a = (in.ptr() + id.x() * _input->info()->strides_in_bytes()[1] + id.y() * _input->info()->strides_in_bytes()[2]); #if __arm__ asm volatile("PLD [%0, #128*4]" ::"r"(matrix_a)); |