Commits · 53da3c98563fcf12494b07c1ee83920976160bf7 · streichg / Finn · GitLab

Snippets Groups Projects

To receive notifications about scheduled maintenance, please subscribe to the mailing-list gitlab-operations@sympa.ethz.ch. You can subscribe to the mailing-list at https://sympa.ethz.ch

Mar 30, 2020
- [Registry] add data width converter to registry · 53da3c98
  auphelia authored 5 years ago
  
  53da3c98
- [HLSCustomOp] add a first draft of StreamingDataWidthConverter · c7f3f684
  Yaman Umuroglu authored 5 years ago
  
  c7f3f684
Mar 26, 2020
- [Transform] call shape&dt inference after HLS conversion · ee89e439
  Yaman Umuroglu authored 5 years ago
  
  ee89e439
- [StreamingFC] add shape and type inference to StreamingFC · 505a0a93
  Yaman Umuroglu authored 5 years ago
  
  505a0a93
- [ModelWrapper] helper fxns to get nodes by type, domain · e2c2a1bb
  Yaman Umuroglu authored 5 years ago
  
  e2c2a1bb
- [CustomOp] increase AP_INT_MAX_W to 16384 · af3ddc40
  Yaman Umuroglu authored 5 years ago
  
  af3ddc40
- [CustomOp] use -O3 in npysim compilation for faster runs · 01100692
  Yaman Umuroglu authored 5 years ago
  
  01100692
- [Transform] add InferStreamingMaxPool · 599d8497
  Yaman Umuroglu authored 5 years ago
  
  599d8497
- [CustomOp, Test] rewrite StreamingMaxPool and test suite · 909e645b
  Yaman Umuroglu authored 5 years ago
  
  909e645b
- [SWG] minor fix in num output elems calc · 608cc8fc
  Yaman Umuroglu authored 5 years ago
  
  608cc8fc
- [StreamingFC] Change streaming MVAU interface back to version before · fe7d83e8
  auphelia authored 5 years ago
  
  fe7d83e8
Mar 25, 2020
- [Transform] add a first v. of InferConvInpGen · cf2915da
  Yaman Umuroglu authored 5 years ago
  
  cf2915da
- [Transform] respect inner dim internal order while lowering · 08730af2
  Yaman Umuroglu authored 5 years ago
  
  08730af2
- [SWG] fix get_number_output_values · 826c9851
  Yaman Umuroglu authored 5 years ago
  
  826c9851
- [SWG] add an extra SIMD=IFM check · 46bf470c
  Yaman Umuroglu authored 5 years ago
  
  46bf470c
- [CustomOp] use standard data un/packing for ConvInpGen exec · b3740dd0
  Yaman Umuroglu authored 5 years ago
  
  b3740dd0
- [StreamingFC] Change streaming MVAU interface · 92e20e5b
  auphelia authored 5 years ago
  
  92e20e5b
- [CustomOp] change ConvInpGen layout, various shape fixes · fa5b11a7
  Yaman Umuroglu authored 5 years ago
  
  fa5b11a7
- [StreamingFC] Fixed bug in saving of weight .npy file · 3f1b888e
  auphelia authored 5 years ago
  
  3f1b888e
- [CustomOp] add shape and type inference to ConvInpGen · 7aed59b6
  Yaman Umuroglu authored 5 years ago
  
  7aed59b6
- [Test] npysim/rtlsim as param for test_fpgadataflow_slidingwindow · 346a5140
  Yaman Umuroglu authored 5 years ago
  
  346a5140
- [Transform] call InferDataTypes after bipolar2xnor · ca07070a
  Yaman Umuroglu authored 5 years ago
  
  ca07070a
- [StreamingFC] umReps = num inp vectors, reread decoupled weights · 338c651d
  Yaman Umuroglu authored 5 years ago
  
  338c651d
- [Util] support numReps in npy<>apintstream adapters · 52693fe4
  Yaman Umuroglu authored 5 years ago
  
  52693fe4
Mar 24, 2020
- [Transform] set numInpVectors when inferring StreamingFC · d5e7e8ac
  Yaman Umuroglu authored 5 years ago
  
  d5e7e8ac
- [StreamingFC] more i/o shape fixes to support mat-mat · faaf3cef
  Yaman Umuroglu authored 5 years ago
  
  faaf3cef
- [StreamingFC] consider num vecs when producing i/o folded shape · 177713ec
  Yaman Umuroglu authored 5 years ago
  
  177713ec
- [StreamingFC] use get_folded_shape fxns throughout · 72b81904
  Yaman Umuroglu authored 5 years ago
  
  72b81904
- [StreamingFC] Unflip weight tensor for .dat files (rtlsim) · 6475dea7
  auphelia authored 5 years ago
  
  6475dea7
- [Transform] relax matmul shape in ConvertToHLSLayers · bfd8818a
  Yaman Umuroglu authored 5 years ago
  
  bfd8818a
- [Transform] handle upstream MT finding in bipolar2xnor · 8bc6e3f2
  Yaman Umuroglu authored 5 years ago
  
  8bc6e3f2
- [DataTypeInf] infer odt=idt for certain node types · 4f363c3d
  Yaman Umuroglu authored 5 years ago
  
  4f363c3d
- [ModelWrapper] add find_upstream helper function · dd11cf2d
  Yaman Umuroglu authored 5 years ago
  
  dd11cf2d
- [Transform] call InferDataType after AbsorbTransposeIntoMultiT · 28474e96
  Yaman Umuroglu authored 5 years ago
  
  28474e96
- [CustomOp] support non-2D inputs for XnorPopcountMatMul · d2b2f7aa
  Yaman Umuroglu authored 5 years ago
  
  d2b2f7aa
- [CustomOp] add shape inference for MaxPoolNHWC · 10756cd8
  Yaman Umuroglu authored 5 years ago
  
  10756cd8
- [ShapeInf] pass model to CustomOp make_shape_compatible_op fxns · e9653644
  Yaman Umuroglu authored 5 years ago
  
  e9653644
Mar 23, 2020
- [Transform] partial conv support in bipolar2xnor · c91fbd6e
  Yaman Umuroglu authored 5 years ago
  
  c91fbd6e
- [CustomOp] add dt inference to MaxPoolNHWC, exception for SI · bae8394b
  Yaman Umuroglu authored 5 years ago
  
  bae8394b
- [CustomOp] add MaxPoolNHWC · e825879d
  Yaman Umuroglu authored 5 years ago
  
  e825879d