diff options
| author | Samuel Yap <samuel.yap@arm.com> | 2022-07-06 15:36:03 +0100 |
|---|---|---|
| committer | Nikhil Raj <nikhil.raj@arm.com> | 2022-07-27 15:58:31 +0100 |
| commit | 6b47809e7d6c55d20a05d863ce2f09159f381f85 (patch) | |
| tree | c33e5820f89e359c80d8773288e8adb075735039 /docs | |
| parent | 919ec71ea7f44bb2d284eb88cda511c2424358b2 (diff) | |
| download | armnn-6b47809e7d6c55d20a05d863ce2f09159f381f85.tar.gz | |
IVGCVSW-7109: Add Batch MatMul front end support - Reference
* Descriptors added for BatchMatMul
* Layer definition added
* Input validation added (will likely change when opt. param support comes in)
* Ref workload implementation for BatchMatMul added (will also change with opt. param support)
* Ref layer tests made for BatchMatMul
* CMake and other build files updated
Signed-off-by: Samuel Yap <samuel.yap@arm.com>
Change-Id: Ic885301da543ee0fbe7922b85e7f9658c4efc617
Diffstat (limited to 'docs')
| -rw-r--r-- | docs/02_operator_list.dox | 42 |
1 files changed, 42 insertions, 0 deletions
diff --git a/docs/02_operator_list.dox b/docs/02_operator_list.dox index 960428999e..658aa07d1d 100644 --- a/docs/02_operator_list.dox +++ b/docs/02_operator_list.dox @@ -268,6 +268,48 @@ where N = batches, C = channels, H = height, W = width <tr><td>FLOAT32 </table> <tr> + <td rowspan="3">BatchMatMulLayer + <td rowspan="3" style="width:200px;"> Layer to perform batch matrix multiplication. + <td rowspan="3"> + <ul> + <li>N/A + </ul> + <td>CpuRef + <td> + <ul> + <li>All + </ul> + <td> + <table> + <tr><th> + <tr><td>BFLOAT16 + <tr><td>FLOAT16 + <tr><td>FLOAT32 + <tr><td>QASYMMS8 + <tr><td>QASYMMU8 + <tr><td>QSYMMS16 + </table> +<tr> + <td>CpuAcc + <td> + <ul> + <li>N/A + </ul> + <td> + <ul> + <li>N/A + </ul> +<tr> + <td>GpuAcc + <td> + <ul> + <li>N/A + </ul> + <td> + <ul> + <li>N/A + </ul> +<tr> <td rowspan="3">BatchNormalizationLayer <td rowspan="3" style="width:200px;"> Layer to perform batch normalization. <td rowspan="3"> |