x86 only supports FP32 precision on AVX512/FMA.
- ONNX
Op Type | Op Set | Linux/Windows/Darwin X86-64 |
---|---|---|
Add | 7~16 | ✓ |
And | 7~16 | ✓ |
ArgMax | 1~11 | ✓ |
AveragePool | 1~16 | ✓ |
BatchNormalization | 9~13 | ✓ |
Cast | 9~16 | ✓ |
Ceil | 6~16 | ✓ |
Clip | 6~16 | ✓ |
Concat | 4~16 | ✓ |
Constant | 9~16 | ✓ |
ConstantOfShape | 9~16 | ✓ |
Conv | 1~16 | ✓ |
ConvTranspose | 1~16 | ✓ |
Cos | 7~16 | ✓ |
CumSum | 11~16 | ✓ |
DepthToSpace | 1~16 | ✓ |
Dropout | 1~16 | ✓ |
Div | 7~16 | ✓ |
Equal | 7~16 | ✓ |
Erf | 9~16 | ✓ |
Exp | 6~16 | ✓ |
Expand | 8~16 | ✓ |
Flatten | 1~16 | ✓ |
Floor | 6~16 | ✓ |
Gather | 1~16 | ✓ |
GatherND | 11 | ✓ |
Gemm | 9~16 | ✓ |
GlobalAveragePool | 1~16 | ✓ |
Greater | 7~16 | ✓ |
Identity | 1~13 | ✓ |
If | 1~12 | ✓ |
LeakyRelu | 6~16 | ✓ |
Less | 7~16 | ✓ |
Log | 6~16 | ✓ |
Loop | 1~12 | ✓ |
LSTM | 7~13 | ✓ |
MatMul | 1~16 | ✓ |
Max | 6~16 | ✓ |
MaxPool | 1~16 | ✓ |
MaxUnpool | 9~16 | ✓ |
Min | 6~16 | ✓ |
Mul | 7~16 | ✓ |
NonMaxSuppression | 10~16 | ✓ |
NonZero | 9~16 | ✓ |
Not | 1~16 | ✓ |
Pad | 2~16 | ✓ |
Pow | 7~16 | ✓ |
PRelu | 6~16 | ✓ |
Range | 11~16 | ✓ |
ReduceMax | 1~16 | ✓ |
ReduceMean | 1~16 | ✓ |
ReduceMin | 1~16 | ✓ |
ReduceProd | 1~16 | ✓ |
ReduceSum | 1~16 | ✓ |
Relu | 6~16 | ✓ |
Reshape | 5~13 | ✓ |
Resize | 11~16 | ✓ |
RoiAlign | 10~15 | ✓ |
ScatterElements | 11~15 | ✓ |
ScatterND | 11~15 | ✓ |
SequenceAt | 11~16 | ✓ |
Shape | 1~14 | ✓ |
Sigmoid | 6~16 | ✓ |
Sin | 7~16 | ✓ |
Slice | 1~16 | ✓ |
Softmax | 1~12 | ✓ |
Split | 2~12 | ✓ |
SplitToSequence | 11~16 | ✓ |
Sqrt | 6~16 | ✓ |
Squeeze | 1~12 | ✓ |
Sub | 7~16 | ✓ |
Sum | 6~16 | ✓ |
Tanh | 6~16 | ✓ |
Tile | 6~16 | ✓ |
TopK | 1~16 | ✓ |
Transpose | 1~16 | ✓ |
Unsqueeze | 1~12 | ✓ |
Where | 9~16 | ✓ |
- MMCV
Op Type | Op Set | Linux/Windows/Darwin X86-64 |
---|---|---|
NonMaxSuppression | 1 | ✓ |
RoiAlign | 1 | ✓ |
grid_sample | 1 | ✓ |
- PPL
Op Type | Op Set | Linux/Windows/Darwin X86-64 |
---|---|---|
ChannelShuffle | 1 | ✓ |
ShapeOperation | 1 | ✓ |
Swish | 1 | ✓ |