Element types encoded in a tensormap descriptor.
Declared in <llvm/IR/NVVMIntrinsicUtils.h>
enum class TensormapElemType : uint8_t;
| Name | Description |
|---|---|
U8 | Unsigned 8-bit integer. |
U16 | Unsigned 16-bit integer. |
U32 | Unsigned 32-bit integer. |
S32 | Signed 32-bit integer. |
U64 | Unsigned 64-bit integer. |
S64 | Signed 64-bit integer. |
F16 | Half-precision floating-point. |
F32 | Single-precision floating-point. |
F32_FTZ | Single-precision with flush-to-zero. |
F64 | Double-precision floating-point. |
BF16 | Brain floating-point 16. |
TF32 | TensorFloat-32. |
TF32_FTZ | TensorFloat-32 with flush-to-zero. |
B4x16 | Packed 4-bit elements in 16-bit containers. |
B4x16_p64 | Packed 4-bit elements with 64-bit packing. |
B6x16_p32 | Packed 6-bit elements with 32-bit packing. |