diff options
| author | Sheri Zhang <sheri.zhang@arm.com> | 2021-09-24 16:02:57 +0100 |
|---|---|---|
| committer | Sheri Zhang <sheri.zhang@arm.com> | 2021-10-15 15:53:10 +0000 |
| commit | 6d9c982a5aec543d1f7f198f0fee10a7a3a78ddb (patch) | |
| tree | 43627d9cebdd9196b8cc350f81b4c2137db92177 /docs/user_guide/operator_list.dox | |
| parent | 273c28cc225d8edc7921397a754649529610c4e2 (diff) | |
| download | ComputeLibrary-6d9c982a5aec543d1f7f198f0fee10a7a3a78ddb.tar.gz | |
Conv3d support
* Add CpuDirectConv3d support for fp32 and fp16
* Dilation is not supported
* Need decouple
Partially resolve: COMPMID-4661
Signed-off-by: Sheri Zhang <sheri.zhang@arm.com>
Change-Id: Ib1865b9ff328b684d131512b1baf77bc2f10318f
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/6430
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Giorgio Arena <giorgio.arena@arm.com>
Diffstat (limited to 'docs/user_guide/operator_list.dox')
| -rw-r--r-- | docs/user_guide/operator_list.dox | 30 |
1 files changed, 30 insertions, 0 deletions
diff --git a/docs/user_guide/operator_list.dox b/docs/user_guide/operator_list.dox index 27ba52d72e..ebc970d8c1 100644 --- a/docs/user_guide/operator_list.dox +++ b/docs/user_guide/operator_list.dox @@ -600,6 +600,36 @@ where N = batches, C = channels, H = height, W = width <tr><td>QASYMM8_SIGNED<td>QSYMM8_PER_CHANNEL<td>S32<td>QASYMM8_SIGNED </table> <tr> + <td rowspan="2">Conv3D + <td rowspan="2" style="width:200px;"> Function to compute a 3d convolution layer. + <td rowspan="2"> + <ul> + <li>ANEURALNETWORKS_CONV_3D + </ul> + <td>NEConv3D + <td> + <ul> + <li>NDHWC + </ul> + <td> + <table> + <tr><th>src0<th>src1<th>src2<th>dst + <tr><td>F16<td>F16<td>F16<td>F16 + <tr><td>F32<td>F32<td>F32<td>F32 + </table> +<tr> + <td>CLConv3D + <td> + <ul> + <li>NDHWC + </ul> + <td> + <table> + <tr><th>src0<th>src1<th>src2<th>dst + <tr><td>F16<td>F16<td>F16<td>F16 + <tr><td>F32<td>F32<td>F32<td>F32 + </table> +<tr> <td rowspan="2">Copy <td rowspan="2" style="width:200px;"> Function to copy a tensor. <td rowspan="2"> |