From 72cb5936f10f7fd611ba1eaa10d601e68084673f Mon Sep 17 00:00:00 2001 From: raver119 Date: Thu, 15 Aug 2019 18:24:29 +0300 Subject: [PATCH] no openmp for ClipByGlobalNorm Signed-off-by: raver119 --- libnd4j/include/ops/declarable/helpers/cuda/transforms.cu | 2 -- 1 file changed, 2 deletions(-) diff --git a/libnd4j/include/ops/declarable/helpers/cuda/transforms.cu b/libnd4j/include/ops/declarable/helpers/cuda/transforms.cu index a457d4c94..5406e8bbd 100644 --- a/libnd4j/include/ops/declarable/helpers/cuda/transforms.cu +++ b/libnd4j/include/ops/declarable/helpers/cuda/transforms.cu @@ -806,7 +806,6 @@ void clipByNormBP(nd4j::LaunchContext* context, const NDArray& input, const NDAr void clipByGlobalNorm_(nd4j::LaunchContext * context, std::vector const& inputs, double clipNorm, nd4j::memory::Workspace* workspace, std::vector& outputs, bool isInplace) { NDArray globalNorm = NDArrayFactory::create(0, inputs[0]->getContext()); //sqrt(sum([l2norm(t)**2 for t in t_list])) - PRAGMA_OMP_PARALLEL_FOR for (auto i = 0; i < inputs.size(); i++) { auto input = inputs[i]; auto l2norm = input->reduceNumber(reduce::Norm2); @@ -818,7 +817,6 @@ void clipByNormBP(nd4j::LaunchContext* context, const NDArray& input, const NDAr globalNorm.syncToHost(); const T factor = clipNorm / globalNorm.e(0); - PRAGMA_OMP_PARALLEL_FOR for (size_t e = 0; e < inputs.size(); e++) { // all-reduce auto input = inputs[e];