diff options
| author | Sang-Hoon Park <sang-hoon.park@arm.com> | 2020-05-21 20:34:19 +0100 |
|---|---|---|
| committer | Manuel Bottini <manuel.bottini@arm.com> | 2020-05-26 10:07:31 +0100 |
| commit | 2168a6809a9a59039530fcb02b9ac86f71fbd975 (patch) | |
| tree | f523c200d89f3782c2be7298c2949b652a517566 | |
| parent | e7019e8ff13a8e4e965cd811f90cf9e4e6a81592 (diff) | |
| download | ComputeLibrary-2168a6809a9a59039530fcb02b9ac86f71fbd975.tar.gz | |
COMPMID-3499: Fix integer overflow for large GEMM on NEON
Change-Id: Id9eef3abc8a902b52ba61772f716f2ba2b97f7d4
Signed-off-by: Sang-Hoon Park <sang-hoon.park@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3245
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Reviewed-by: Manuel Bottini <manuel.bottini@arm.com>
Reviewed-by: Michalis Spyrou <michalis.spyrou@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
| -rw-r--r-- | src/core/NEON/kernels/arm_gemm/transforms/a64_block16_interleave4_8bit.hpp | 6 |
1 files changed, 3 insertions, 3 deletions
diff --git a/src/core/NEON/kernels/arm_gemm/transforms/a64_block16_interleave4_8bit.hpp b/src/core/NEON/kernels/arm_gemm/transforms/a64_block16_interleave4_8bit.hpp index 6a15fc42e4..6b742c8776 100644 --- a/src/core/NEON/kernels/arm_gemm/transforms/a64_block16_interleave4_8bit.hpp +++ b/src/core/NEON/kernels/arm_gemm/transforms/a64_block16_interleave4_8bit.hpp @@ -38,7 +38,7 @@ void TransformImpl<4, 16, false, 1, 1, false>::Transform(T *out, const T *in, in uint8_t zerobuff[16] = { 0 }; - for (int y=y0; y<ymax; y+=4) { + for (uint64_t y = y0 ; y < static_cast<uint64_t>(ymax) ; y+=4) { const uint8_t *inptr0 = inptr + y * ldin + k0; const uint8_t *inptr1 = inptr0 + ldin; const uint8_t *inptr2 = inptr1 + ldin; @@ -52,7 +52,7 @@ void TransformImpl<4, 16, false, 1, 1, false>::Transform(T *out, const T *in, in int x=(kmax-k0); for (;x>15;x-=16) { /* Cope with ragged cases by copying from a buffer of zeroes instead */ - if ((y + 3) >= ymax) { + if ((y + 3) >= static_cast<uint64_t>(ymax)) { switch ((y + 3) - ymax) { /* Everything falls through in here */ case 2: @@ -90,7 +90,7 @@ void TransformImpl<4, 16, false, 1, 1, false>::Transform(T *out, const T *in, in if (x>0) { /* Need to duplicate this here, in case we didn't run the main loop. */ - if ((y + 3) >= ymax) { + if ((y + 3) >= static_cast<uint64_t>(ymax)) { switch ((y + 3) - ymax) { /* Everything falls through in here */ case 2: |