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

Class Name

tensorflow/core/framework/op_kernel.h:1452–1466  ·  view source on GitHub ↗

Source from the content-addressed store, hash-verified

1450namespace register_kernel {
1451
1452class Name : public KernelDefBuilder {
1453 public:
1454 // With selective registration, kernels whose implementation class is not used
1455 // by any kernel are disabled with the SHOULD_REGISTER_OP_KERNEL call in
1456 // REGISTER_KERNEL_BUILDER_UNIQ. However, an unused kernel that shares an
1457 // implementation class with a used kernel would get through that mechanism.
1458 //
1459 // This mechanism stops that registration by changing the name of the kernel
1460 // for the unused op to one that is ignored by
1461 // OpKernelRegistrar::InitInternal. Note that this method alone is
1462 // not sufficient - the compiler can't evaluate the entire KernelDefBuilder at
1463 // compilation time, so this method doesn't actually reduce code size.
1464 explicit Name(const char* op)
1465 : KernelDefBuilder(SHOULD_REGISTER_OP(op) ? op : "_no_register") {}
1466};
1467
1468namespace system {
1469

Callers 15

AddNodeInputsMethod · 0.70
DepMethod · 0.70
NameMethod · 0.70
op_kernel_test.ccFile · 0.70
dataset.ccFile · 0.70
lookup_kernels.ccFile · 0.50
cuda_op_kernel.ccFile · 0.50

Calls

no outgoing calls

Tested by

no test coverage detected