diff options
| author | Narumol Prangnawarat <narumol.prangnawarat@arm.com> | 2020-03-06 14:45:57 +0000 |
|---|---|---|
| committer | Jan Eilers <jan.eilers@arm.com> | 2020-03-09 16:13:56 +0000 |
| commit | 8832522f47b701f5f042069e7bf8deae9b75d449 (patch) | |
| tree | f217ab7fbda860a947eba88c9508eb1ac1b1d670 /include | |
| parent | 97da5e2e6c8aaaf4249af60e8305431315226f15 (diff) | |
| download | armnn-8832522f47b701f5f042069e7bf8deae9b75d449.tar.gz | |
IVGCVSW-4517 Implement BFloat16 Encoder and Decoder
* Add ConvertFloat32ToBFloat16
* Add ConvertBFloat16ToFloat32
* Add BFloat16Encoder
* Add BFloat16Decoder
* Unit tests
Signed-off-by: Narumol Prangnawarat <narumol.prangnawarat@arm.com>
Change-Id: I198888384c923aba28cfbed09a02edc6f8194b3e
Diffstat (limited to 'include')
| -rw-r--r-- | include/armnnUtils/FloatingPointConverter.hpp | 6 |
1 files changed, 6 insertions, 0 deletions
diff --git a/include/armnnUtils/FloatingPointConverter.hpp b/include/armnnUtils/FloatingPointConverter.hpp index cf573a2ee8..a2244735dc 100644 --- a/include/armnnUtils/FloatingPointConverter.hpp +++ b/include/armnnUtils/FloatingPointConverter.hpp @@ -18,6 +18,12 @@ public: static void ConvertFloat32To16(const float *srcFloat32Buffer, size_t numElements, void *dstFloat16Buffer); static void ConvertFloat16To32(const void *srcFloat16Buffer, size_t numElements, float *dstFloat32Buffer); + + // Converts a buffer of FP32 values to BFloat16, and stores in the given dstBFloat16Buffer. + static void ConvertFloat32ToBFloat16(const float* srcFloat32Buffer, size_t numElements, void* dstBFloat16Buffer); + + // Converts a buffer of BFloat16 to FP32 value, and stores in the given dstFloat32Buffer. + static void ConvertBFloat16ToFloat32(const void* srcBFloat16Buffer, size_t numElements, float* dstFloat32Buffer); }; } // namespace armnnUtils |