diff options
author | Louis Verhaard <louis.verhaard@arm.com> | 2021-02-09 16:08:26 +0100 |
---|---|---|
committer | Louis Verhaard <louis.verhaard@arm.com> | 2021-02-17 16:29:20 +0100 |
commit | 1a92f78e14f31f1423824228deb0628b7a9a9071 (patch) | |
tree | 4e0de9a2e6b5b7d9159b25cbfead4a625c134a3c /SUPPORTED_OPS.md | |
parent | 8d0f4890aa0ceae92a33ebb789701ff644a6fcaa (diff) | |
download | ethos-u-vela-1a92f78e14f31f1423824228deb0628b7a9a9071.tar.gz |
MLBEDSW-4022: support PAD followed by pool operator
PAD followed by max/average pool is run on NPU if NPU
padding can be used. Average pool is converted to depthwise.
Change-Id: Icc3652e6d9ecff5ac3dc7d92080313d90c245404
Signed-off-by: Louis Verhaard <louis.verhaard@arm.com>
Diffstat (limited to 'SUPPORTED_OPS.md')
-rw-r--r-- | SUPPORTED_OPS.md | 5 |
1 files changed, 3 insertions, 2 deletions
diff --git a/SUPPORTED_OPS.md b/SUPPORTED_OPS.md index dfa24d0a..20134ccc 100644 --- a/SUPPORTED_OPS.md +++ b/SUPPORTED_OPS.md @@ -1,7 +1,7 @@ # Supported Ops This file was automatically generated by Vela using the `--supported-ops-report` parameter. -Vela version: `2.0.2.dev49+gda756aa` +Vela version: `2.0.2.dev69+g83e3bb3.d20210212` This file complies with [**Gitiles Markdown syntax**](https://github.com/google/gitiles/blob/master/Documentation/markdown.md) @@ -63,6 +63,7 @@ This is a list of constraints that all NPU operators must satisfy in order to be - Input(s), Output and Weight tensors with quantization scales must be finite - Per-axis quantization is only supported for the following op types: CONV_2D, DEPTHWISE_CONV_2D, TRANSPOSE_CONV - The fused activation function (if present) must be one of type: LOGISTIC, RELU, RELU6, RELU_N1_TO_1, TANH +- If a fused activation function is present, the Output tensor must be one of type: int16, int8, uint8 - Input and Output tensors must have quantization scales that fit within float32 precision ## ABS Constraints @@ -221,7 +222,7 @@ This is a list of constraints that the PAD operator must satisfy in order to be - The pad tensor can only pad width and height - Pad tensor must be of type: int32, int64 - The padding tensor must be constant -- Must be followed by one of the following operator types: CONV_2D, DEPTHWISE_CONV_2D +- Must be followed by one of the following operator types: AVERAGE_POOL_2D, CONV_2D, DEPTHWISE_CONV_2D, MAX_POOL_2D - Padding must be at most kernel size divided by 2 ## RESIZE_BILINEAR Constraints |