aboutsummaryrefslogtreecommitdiff
path: root/include
AgeCommit message (Collapse)Author
2020-12-06Add Runtime flag to indicate sparse inference preferenceMarat Dukhan
PiperOrigin-RevId: 346026748
2020-12-01Support ELU operator in Subgraph APIMarat Dukhan
PiperOrigin-RevId: 345167963
2020-12-01Implement ELU operatorMarat Dukhan
PiperOrigin-RevId: 345129548
2020-11-25NHWC version of Depth-To-Space operatorMarat Dukhan
PiperOrigin-RevId: 344304885
2020-11-24Unify parameters of DepthToSpace NCHW->NHWC operator with other operatorsMarat Dukhan
PiperOrigin-RevId: 344183834
2020-11-12Refactor Depth-To-Space operatorMarat Dukhan
PiperOrigin-RevId: 342119823
2020-11-12Make DepthToSpace X32 operator consistent with other X32 operatorsMarat Dukhan
PiperOrigin-RevId: 342102000
2020-11-12Rename CHW2HWC to NCHW2NCHW in DepthToSpace operatorMarat Dukhan
Force constistent naming with other operators PiperOrigin-RevId: 342098263
2020-11-05Introduce DEPTH_TO_SPACE operator and enable it for graph rewritingArtsiom Ablavatski
PiperOrigin-RevId: 340925373
2020-10-27Add operator implementation and tests for IBILINEAR CHW microkernelArtsiom Ablavatski
PiperOrigin-RevId: 339348062
2020-10-23Add WebAssembly SIMD IBILINEAR microkernels for CHW layoutXNNPACK Team
PiperOrigin-RevId: 338792392
2020-10-23Roll back the decision to split the packed weights for the CHW IBILINEAR ↵XNNPACK Team
microkernel interface PiperOrigin-RevId: 338785454
2020-10-23Split packed weights into horizontal and vertical in IBILINEAR CHW ↵XNNPACK Team
microkernel interface PiperOrigin-RevId: 338744798
2020-09-13Alphabetically order FP16 operatorsMarat Dukhan
PiperOrigin-RevId: 331473625
2020-09-11FP16 HardSwish operatorFrank Barchard
PiperOrigin-RevId: 331236366
2020-09-09FP16 binary multiply operatorFrank Barchard
PiperOrigin-RevId: 330810365
2020-09-03QS8 version of ND ADD operatorMarat Dukhan
PiperOrigin-RevId: 329993459
2020-08-07QS8 variant of NWC Global Average Pooling operatorMarat Dukhan
PiperOrigin-RevId: 325402397
2020-08-04QS8 version of NHWC Convolution operatorMarat Dukhan
- Operator implementation - Unit tests - Microbenchmarks PiperOrigin-RevId: 324916339
2020-07-23Support Resize Bilinear 2D in Subgraph APIMarat Dukhan
PiperOrigin-RevId: 322743689
2020-07-17Remove output_min and output_max arguments in ArgMax Pooling operatorMarat Dukhan
These arguments weren't actually used, but cost performance in WAsm builds PiperOrigin-RevId: 321727974
2020-07-15Rename Q8 microkernels and operators to QU8Marat Dukhan
- Indicate unsigned datatype - Make space for QS8 with signed quantization PiperOrigin-RevId: 321275708
2020-07-09Support Reshape Node in Subgraph APIMarat Dukhan
PiperOrigin-RevId: 320553774
2020-07-09Support Square Root node in Subgraph APIMarat Dukhan
PiperOrigin-RevId: 320364215
2020-06-30Square Root operatorMarat Dukhan
PiperOrigin-RevId: 319145965
2020-06-26FP16 Convolution NHWC operatorFrank Barchard
PiperOrigin-RevId: 318540097
2020-06-23FP16 binary add operatorFrank Barchard
PiperOrigin-RevId: 317999884
2020-06-22Group FP16 operators together in a section guarded by #ifndef ↵Frank Barchard
XNN_NO_F16_OPERATORS PiperOrigin-RevId: 317746266
2020-06-18Parametrise Convolution operators by strides along the channel dimensionMarat Dukhan
Unify parametrisation across NCHW and NHWC Convolution operators PiperOrigin-RevId: 317102226
2020-06-12FP16 Global Average Pooling operatorFrank Barchard
PiperOrigin-RevId: 316066562
2020-06-11Global Average Pooling operator in Subgraph APIMarat Dukhan
PiperOrigin-RevId: 315943883
2020-06-10Leaky ReLU operator in Subgraph APIMarat Dukhan
PiperOrigin-RevId: 315817869
2020-06-10FP32 Leaky ReLU operatorMarat Dukhan
PiperOrigin-RevId: 315806661
2020-06-10Six new unary elementwise operators in Subgraph APIMarat Dukhan
- Abs - Bankers' Rounding - Ceiling - Floor - Negate - Square PiperOrigin-RevId: 315649679
2020-06-09Rounding operatorsMarat Dukhan
- Bankers' Rounding - Ceiling - Floor - Truncation PiperOrigin-RevId: 315550942
2020-06-08Abs, Negate, and Square NC operatorsMarat Dukhan
PiperOrigin-RevId: 315341068
2020-06-08Remove remnants of Add NC operatorMarat Dukhan
PiperOrigin-RevId: 315336496
2020-06-05Divide/Subtract/Minimum2/Maximum2/SquaredDifference in Subgraph APIMarat Dukhan
PiperOrigin-RevId: 315020715
2020-06-05Squared Difference operatorMarat Dukhan
PiperOrigin-RevId: 314957945
2020-06-05Remove Channel Pad NC operatorMarat Dukhan
Pad ND provides a more flexible alternative PiperOrigin-RevId: 314947705
2020-06-04Copy NC operatorMarat Dukhan
PiperOrigin-RevId: 314845678
2020-05-22Rename Pad operator to Constant Pad in Operator APIMarat Dukhan
PiperOrigin-RevId: 312838474
2020-05-21Support Static Constant Pad operator in Subgraph APIMarat Dukhan
PiperOrigin-RevId: 312786589
2020-05-21N-dimensional Pad operatorMarat Dukhan
PiperOrigin-RevId: 312757022
2020-05-05ArgMax Pooling 2D and UnPooling 2D nodes in Subgraph APIMarat Dukhan
PiperOrigin-RevId: 310046015
2020-04-27Support Deconvolution in Subgraph APIMarat Dukhan
PiperOrigin-RevId: 308738695
2020-04-23Fully Connected node in Subgraph APIMarat Dukhan
PiperOrigin-RevId: 308156251
2020-02-29Support Average Pooling and Max Pooling in Subgraph APIMarat Dukhan
PiperOrigin-RevId: 298018572
2020-02-24Clarify input_padding_* arguments with XNN_FLAG_TENSORFLOW_SAME_PADDING flagMarat Dukhan
PiperOrigin-RevId: 296893373
2020-02-24Remove output_min & output_max arguments in PReLU operatorMarat Dukhan
- There arguments weren't really used (e.g. TFLite PReLU operator doesn't have built-in activation argument, and it is not common to have PReLU followed by ReLU in general). - Supporting them is a non-trivial cost on WebAssembly targets PiperOrigin-RevId: 296826639
generated by cgit v1.2.3 (git 2.25.1) at 2024-07-16 02:43:05 +0000