Data type support#

The input and output data types supported by hipCUB are listed here:

Table 1 Supported Input/Output Types#

Input/Output Types

AMD Support

CUDA Support

int8

✅

✅

float8

❌

❌

bfloat8

❌

❌

int16

✅

✅

float16

✅

✅ 1

bfloat16

✅

✅ 2

int32

✅

✅

tensorfloat32

❌

❌

float32

✅

✅

float64

✅

✅

Table 1 Supported Input/Output Types#
Input/Output Types	AMD Support	CUDA Support
int8	✅	✅
float8	❌	❌
bfloat8	❌	❌
int16	✅	✅
float16	✅	✅ 1
bfloat16	✅	✅ 2
int32	✅	✅
tensorfloat32	❌	❌
float32	✅	✅
float64	✅	✅

Footnotes

1: NVIDIA backend can’t handle float16 with the following API calls: block_adjacent_difference, device_adjacenet_difference, device_reduce, device_scan, device_segmented_reduce and device_select.
2: NVIDIA backend can’t handle bfloat16 with the following API calls: block_adjacent_difference, device_adjacenet_difference, device_reduce, device_scan, device_segmented_reduce, device_select and device_histogram.