aboutsummaryrefslogtreecommitdiff
path: root/src/core/NEON/kernels/arm_gemm/gemm_fp16.cpp
diff options
context:
space:
mode:
authorAnthony Barbier <anthony.barbier@arm.com>2018-07-03 16:22:02 +0100
committerAnthony Barbier <anthony.barbier@arm.com>2018-11-02 16:54:10 +0000
commit5f707736413aeac77818c42838296966f8dc6761 (patch)
treeb829ed3243ea5f3085f288836132416c78bc2e72 /src/core/NEON/kernels/arm_gemm/gemm_fp16.cpp
parent7485d5a62685cb745ab50e970adb722cb71557ac (diff)
downloadComputeLibrary-5f707736413aeac77818c42838296966f8dc6761.tar.gz
COMPMID-1369: Revert accidental formatting of RSH's repo
Pulled latest fixes from David's repo: commit f43ebe932c84083332b0b1a0348241b69dda63a7 Author: David Mansell <David.Mansell@arm.com> Date: Tue Jul 3 18:09:01 2018 +0100 Whitespace tidying, fixed comment in gemv_batched imported from ACL. Change-Id: Ie37a623f44e90d88072236cb853ac55ac82d5f51 Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/138530 Tested-by: Jenkins <bsgcomp@arm.com> Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com> Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com> Reviewed-by: David Mansell <david.mansell@arm.com> Reviewed-by: Anthony Barbier <anthony.barbier@arm.com>
Diffstat (limited to 'src/core/NEON/kernels/arm_gemm/gemm_fp16.cpp')
-rw-r--r--src/core/NEON/kernels/arm_gemm/gemm_fp16.cpp20
1 files changed, 9 insertions, 11 deletions
diff --git a/src/core/NEON/kernels/arm_gemm/gemm_fp16.cpp b/src/core/NEON/kernels/arm_gemm/gemm_fp16.cpp
index d1180b13cb..fa12942829 100644
--- a/src/core/NEON/kernels/arm_gemm/gemm_fp16.cpp
+++ b/src/core/NEON/kernels/arm_gemm/gemm_fp16.cpp
@@ -30,33 +30,31 @@
#include "gemm_common.hpp"
#include "gemm_interleaved.hpp"
-#include "kernels/a32_sgemm_8x6.hpp"
#include "kernels/a64_hgemm_24x8.hpp"
#include "kernels/a64_sgemm_12x8.hpp"
+#include "kernels/a32_sgemm_8x6.hpp"
+
+namespace arm_gemm {
-namespace arm_gemm
-{
-template <>
+template<>
UniqueGemmCommon<__fp16, __fp16> gemm(const CPUInfo &ci, const unsigned int M, const unsigned int N, const unsigned int K,
const unsigned int nbatches, const unsigned int nmulti,
const bool trA, const bool trB, const __fp16 alpha, const __fp16 beta,
- const int maxthreads, const bool pretransposed_hint)
-{
+ const int maxthreads, const bool pretransposed_hint) {
#ifdef __aarch64__
- // Only consider the native FP16 kernel if it will get built.
+// Only consider the native FP16 kernel if it will get built.
#if defined(__ARM_FEATURE_FP16_VECTOR_ARITHMETIC) || defined(FP16_KERNELS)
#ifdef __ARM_FEATURE_FP16_VECTOR_ARITHMETIC
// If the compiler is configured to enable this feature always, then assume it is available at runtime too.
- const bool use_fp16 = true;
+ const bool use_fp16=true;
#else
// Otherwise, detect at runtime via CPUInfo.
- const bool use_fp16 = ci.has_fp16();
+ const bool use_fp16=ci.has_fp16();
#endif
// If FP16 is supported, use it.
- if(use_fp16)
- {
+ if (use_fp16) {
return UniqueGemmCommon<__fp16, __fp16>(new GemmInterleaved<hgemm_24x8, __fp16, __fp16>(&ci, M, N, K, nbatches, nmulti, trA, trB, alpha, beta, maxthreads, pretransposed_hint));
}
#endif