Update supported data type table (#520)

* Update supported data type table

Use the type names from the precision landing page of the rocm documentation

* Update Changelog

CHANGELOG.md

@@ -83,6 +83,10 @@ Documentation for rocPRIM is available at
 
 * Build issues with `rmake.py` on Windows when using VS 2017 15.8 or later (due to a breaking fix with
  extended aligned storage)
+* Fix tests for `block_histogram`, `block_exchange`, `device_histogram` and `device_reduce_by_key` for various types
+
+### Known Issues
+* `device_run_length_encode`, `warp_exchange` and `warp_load` tests fail with `rocprim::half`
 
 ## rocPRIM-3.0.0 for ROCm 6.0.0
 

docs/reference/data-type-support.rst

@@ -15,52 +15,35 @@ The following table shows the supported input and output datatypes.
  :name: supported-input-output-types
 
  *
- - Input/Output Types
- - Library Data Type
+ - Type
  - Support
  *
  - int8
- - int8_t
- - ⚠️
+ - ✅
  *
- - float8
- - Not Supported
- - ❌
+ - int16
+ - ✅
  *
- - bfloat8
- - Not Supported
- - ❌
+ - int32
+ - ✅
  *
- - int16
- - int16_t
+ - int64
+ - ✅
+ *
+ - float8
  - ❌
  *
  - float16
- - rocprim::half
- - ⚠️
+ - ✅
  *
  - bfloat16 
- - rocprim::bfloat16
- - ⚠️
- *
- - int32
- - int
  - ✅
  *
  - tensorfloat32
- - Not Supported
  - ❌
  *
  - float32
- - float
  - ✅
  *
  - float64
- - double
  - ✅
-
-The ⚠️ means that the data type is mostly supported, but there are some API tests, that do not work.
-
- * The ``block_histogram`` test fails with ``int8``.
- * The ``device_histogram`` and ``device_reduce_by_key`` doesn't work with ``rocprim::half`` and ``rocprim::bfloat16``.
- * The ``device_run_length_encode``, ``warp_exchange`` and ``warp_load`` doesn't work with ``rocprim::half``.