From 6011f2444eacd8e549c74bdbb10e1c3cd2b29e77 Mon Sep 17 00:00:00 2001 From: Georgios Pinitas Date: Fri, 15 Nov 2019 14:26:44 +0000 Subject: COMPMID-2853: Fix VGG16 for fp16. Half the number of granules for fp16 compaired to fp32. Change-Id: Id9a39d29a5b08295fade8271b93ef80a19a536ce Signed-off-by: Georgios Pinitas Reviewed-on: https://review.mlplatform.org/c/2298 Tested-by: Arm Jenkins Reviewed-by: Michalis Spyrou Comments-Addressed: Arm Jenkins --- src/runtime/NEON/functions/NEGEMMAssemblyDispatch.cpp | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/src/runtime/NEON/functions/NEGEMMAssemblyDispatch.cpp b/src/runtime/NEON/functions/NEGEMMAssemblyDispatch.cpp index 88e060109a..24254eb9e5 100644 --- a/src/runtime/NEON/functions/NEGEMMAssemblyDispatch.cpp +++ b/src/runtime/NEON/functions/NEGEMMAssemblyDispatch.cpp @@ -390,10 +390,10 @@ void Fallback::run() // Schedule assembly kernel IScheduler::Hints scheduling_hint = IScheduler::Hints(Window::DimX); - if(_kernel_info.method == arm_gemm::GemmMethod::GEMM_INTERLEAVED) + if(_kernel_info.method == arm_gemm::GemmMethod::GEMM_INTERLEAVED && _d->info()->data_type() == DataType::F32) { - constexpr int granule_threshold = 200; - scheduling_hint = IScheduler::Hints(Window::DimX, IScheduler::StrategyHint::DYNAMIC, granule_threshold); + const int granule_threshold = 200; + scheduling_hint = IScheduler::Hints(Window::DimX, IScheduler::StrategyHint::DYNAMIC, granule_threshold); } NEScheduler::get().schedule(_optimised_kernel.get(), scheduling_hint); } -- cgit v1.2.1