experiments/STT-tensorflow
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Disable multi-threading for quantized depthwise_conv

Browse Source 
PiperOrigin-RevId: 243815763
This commit is contained in:
Jared Duke 2019-04-16 08:53:47 -07:00 committed by TensorFlower Gardener
parent b8206b4279
commit e350128d64
1 changed files with 3 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
tensorflow/lite/kernels/internal/optimized/depthwiseconv_uint8.h
View File

@ -2128,8 +2128,9 @@ inline void DepthwiseConv(
thread_count = thread_count_row;
}
const int max_threads = gemm_context ? gemm_context->max_num_threads() : 1;
thread_count = std::max(1, std::min(thread_count, max_threads));
// TODO(b/130555917): Allow multi-threading after fixing ARM accuracy issues.
constexpr int kMaxThreads = 1;
thread_count = std::max(1, std::min(thread_count, kMaxThreads));
if (thread_count == 1) {
DepthwiseConvImpl(params, input_shape, input_data, filter_shape,