__bfloat162half_rna
Applicability
Product |
Supported |
|---|---|
Atlas 350 Accelerator Card |
√ |
x |
|
x |
|
x |
|
x |
|
x |
|
x |
Function Usage
Converts bfloat16 data to the half type in line with the CAST_ROUND mode and returns the converted value.
Prototype
1 | inline half __bfloat162half_rna(const bfloat16_t x) |
Parameters
Parameter |
Input/Output |
Description |
|---|---|---|
x |
Input |
Source operand. |
Returns
The half data converted from the input in line with the CAST_ROUND mode. Special scenarios are described as follows:
- If x is nan, the return value is nan.
- If x is inf, the return value is inf.
- If x is -inf, the return value is -inf.
Restrictions
For SIMT programming, this API is not supported.
Header File to Be Included
To use this API, the simt_api/asc_bf16.h header file must be included.
1 | #include "simt_api/asc_bf16.h" |
Examples
For SIMD and SIMT programming:
1 2 3 4 5 | __simt_vf__ __launch_bounds__(1024) inline void kernel__bfloat162half_rna(__gm__ half* dst, __gm__ bfloat16_t* x) { int idx = threadIdx.x + blockIdx.x * blockDim.x; dst[idx] = __bfloat162half_rna(x[idx]); } |
Parent topic: Type Conversion