MCPcopy Create free account
hub / github.com/DeepRec-AI/DeepRec / Compile

Method Compile

tensorflow/lite/delegates/gpu/cl/kernels/conv_buffer_1x1.cc:234–250  ·  view source on GitHub ↗

Source from the content-addressed store, hash-verified

232}
233
234Status ConvBuffer1x1::Compile(const CreationContext& creation_context) {
235 std::string code_flt4 = GenerateConvBuffer1x1(
236 definition_.src_tensors[0], definition_.dst_tensors[0],
237 definition_.precision, flt4_x_count_, flt4_y_count_, 1,
238 linked_operations_);
239 RETURN_IF_ERROR(creation_context.cache->GetOrCreateCLKernel(
240 code_flt4, "main_function", *creation_context.context,
241 *creation_context.device, &kernel_flt4_));
242 std::string code_flt8 = GenerateConvBuffer1x1(
243 definition_.src_tensors[0], definition_.dst_tensors[0],
244 definition_.precision, flt8_x_count_, flt8_y_count_, 2,
245 linked_operations_);
246 RETURN_IF_ERROR(creation_context.cache->GetOrCreateCLKernel(
247 code_flt8, "main_function", *creation_context.context,
248 *creation_context.device, &kernel_flt8_));
249 return OkStatus();
250}
251
252CLKernel* ConvBuffer1x1::GetKernel(int width) {
253 if (width % 2 == 0) {

Callers

nothing calls this directly

Calls 3

GenerateConvBuffer1x1Function · 0.85
OkStatusFunction · 0.85
GetOrCreateCLKernelMethod · 0.80

Tested by

no test coverage detected