Data type support#
Supported input and output types.
Table 1 Supported Input/Output Types# Input/Output Types
Library Data Type
AMD Supports
CUDA Supports
int8
HIPSPARSELT_R_8I
✅
✅
float8
HIPSPARSELT_R_8F
❌
❌
bfloat8
HIPSPARSELT_R_8BF
❌
❌
int16
Not Supported
❌
❌
float16
HIPSPARSELT_R_16F
✅
✅
bfloat16
HIPSPARSELT_R_16BF
✅
✅
int32
Not Supported
❌
❌
tensorfloat32
Not Supported
❌
❌
float32
HIPSPARSELT_R_32F
❌
✅
float64
Not Supported
❌
❌
Supported accumulator types.
Table 2 Supported Compute Types# Accumulator Types
Library Data Type
AMD Supports
CUDA Supports
int8
Not Supported
❌
❌
float8
Not Supported
❌
❌
bfloat8
Not Supported
❌
❌
int16
Not Supported
❌
❌
float16
HIPSPARSELT_COMPUTE_16F
❌
✅
bfloat16
Not Supported
❌
❌
int32
HIPSPARSELT_COMPUTE_32I
✅
✅
tensorfloat32
Not Supported
❌
✅
float32
HIPSPARSELT_COMPUTE_32F
✅
❌
float64
Not Supported
❌
❌
List of supported compute types at specific input and output types:
Input
Output
Compute type
Backend
HIPSPARSELT_R_16F
HIPSPARSELT_R_16F
HIPSPARSELT_COMPUTE_32F
HIP
HIPSPARSELT_R_16BF
HIPSPARSELT_R_16BF
HIPSPARSELT_COMPUTE_32F
HIP
HIPSPARSELT_R_8I
HIPSPARSELT_R_8I
HIPSPARSELT_COMPUTE_32I
HIP / CUDA
HIPSPARSELT_R_8I
HIPSPARSELT_R_16F
HIPSPARSELT_COMPUTE_32I
HIP / CUDA
HIPSPARSELT_R_16F
HIPSPARSELT_R_16F
HIPSPARSELT_COMPUTE_16F
CUDA
HIPSPARSELT_R_16BF
HIPSPARSELT_R_16BF
HIPSPARSELT_COMPUTE_16F
CUDA
HIPSPARSELT_R_32F
HIPSPARSELT_R_32F
HIPSPARSELT_COMPUTE_TF32
CUDA
HIPSPARSELT_R_32F
HIPSPARSELT_R_32F
HIPSPARSELT_COMPUTE_TF32_FAST
CUDA