MCPcopy Create free account
hub / github.com/NVIDIA/TensorRT / serialize

Method serialize

samples/python/onnx_custom_plugin/plugin/customHardmaxPlugin.cpp:240–252  ·  view source on GitHub ↗

Source from the content-addressed store, hash-verified

238}
239
240void HardmaxPlugin::serialize(void* buffer) const noexcept
241{
242 // Same order as in deserialize()
243 uint8_t* d = static_cast<uint8_t*>(buffer);
244 uint8_t* const a = d;
245
246 writeToBuffer(d, mAxis);
247 writeToBuffer(d, mAxisSize);
248 writeToBuffer(d, mDimProductOuter);
249 writeToBuffer(d, mDimProductInner);
250
251 ASSERT(d == a + getSerializationSize());
252}
253
254bool HardmaxPlugin::supportsFormatCombination(
255 int32_t pos, nvinfer1::PluginTensorDesc const* inOut, int32_t nbInputs, int32_t nbOutputs) noexcept

Callers 6

create_engineMethod · 0.45
create_engineMethod · 0.45
create_engineMethod · 0.45
create_engineMethod · 0.45
saveEngineFunction · 0.45
updateTimingCacheFileFunction · 0.45

Calls 1

writeToBufferFunction · 0.70

Tested by

no test coverage detected