cast.h¶
Functions to cast to/from FP8.
Functions
-
void nvte_fp8_quantize(const NVTETensor input, NVTETensor output, cudaStream_t stream)¶
Cast tensor to FP8.
- Parameters
input – [in] Input tensor to be cast.
output – [inout] Output FP8 tensor.
stream – [in] CUDA stream used for the operation.
-
void nvte_fp8_dequantize(const NVTETensor input, NVTETensor output, cudaStream_t stream)¶
Cast tensor from FP8.
- Parameters
input – [in] Input tensor to be cast.
output – [out] Output tensor.
stream – [in] CUDA stream used for the operation.