diff options
authorGeorgios Pinitas <georgios.pinitas@arm.com>2019-09-12 12:26:11 +0100
committerMichele Di Giorgio <michele.digiorgio@arm.com>2019-09-12 15:26:15 +0000
commitc370c754ebd133cfed19ad1773ef540e48bd4240 (patch)
parent9285adb5ac8e28a9cc82ce708bb2975dc5a074dd (diff)
COMPMID-2570: Add hybrid precision flag in GEMMKernelInfo.
Adds fp_mixed_precision flag in GEMMKernelInfo in order to indicate use of wider accumulators for FP16 workloads where supported. Change-Id: I8d69bc62d945fd40af77395ff673d07880b7aeb6 Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com> Reviewed-on: https://review.mlplatform.org/c/1914 Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
1 files changed, 1 insertions, 0 deletions
diff --git a/arm_compute/core/KernelDescriptors.h b/arm_compute/core/KernelDescriptors.h
index 3affc30f7..215f0f165 100644
--- a/arm_compute/core/KernelDescriptors.h
+++ b/arm_compute/core/KernelDescriptors.h
@@ -60,6 +60,7 @@ struct GEMMKernelInfo
unsigned int depth_output_gemm3d{ 0 }; /**< Depth of the output tensor in case is reinterpreted as 3D */
bool reinterpret_input_as_3d{ false }; /**< Flag used to reinterpret the input as 3D */
bool broadcast_bias{ false }; /**< Flag used to broadcase the bias addition */
+ bool fp_mixed_precision{ false }; /**< Flag used to indicate wider accumulators (32 bit instead of 16 for FP16). */
ActivationLayerInfo activation_info{}; /**< Activation function to perform after the matrix multiplication */