Commits · 7a5146c4aa415aee84f6eceedec676102fdf7dfe · 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

Nov 12, 2021
- [Concat] make both cppsim and rtl/stitched-ip rtlsim work · 7a5146c4
  Yaman Umuroglu authored 3 years ago
  
  7a5146c4
- [Stitch] gen hierarchical util report if synthesized dcp · 529bc249
  Yaman Umuroglu authored 3 years ago
  
  529bc249
- [Concat] allow indexing get_input_datatype, does nothing though · 14206f6f
  Yaman Umuroglu authored 3 years ago
  
  14206f6f
- [InsertFIFO] fixes when first node has multiple inputs · 8827ab97
  Yaman Umuroglu authored 3 years ago
  
  8827ab97
Nov 10, 2021
- [SWG] updates to ConvolutionInputGenerator1D resource estimates · 1c37c8d7
  Yaman Umuroglu authored 3 years ago
  
  1c37c8d7
- [Stitch] fix ext interface ordering for multi-I/O · c251fbc5
  Yaman Umuroglu authored 3 years ago
  
  c251fbc5
Nov 05, 2021
- [Refactor] remove print · a717451e
  Yaman Umuroglu authored 3 years ago
  
  a717451e
- [Driver] small fix for handling single output · a4ed5700
  Yaman Umuroglu authored 3 years ago
  
  a4ed5700
- [Docs] round of updates for the docs · c6b103a6
  Yaman Umuroglu authored 3 years ago
  
  c6b103a6
Nov 02, 2021
- [Build] change verification filename for step_qonnx_to_finn · 7822ecd0
  Yaman Umuroglu authored 3 years ago
  
  7822ecd0
Oct 29, 2021
- Bugfix: Added missing board_part definition for Pynq-Z2. · 6a22a11f
  Hendrik Borras authored 3 years ago
  
  6a22a11f
- Updated finn-base commit and moved _get_signed_from_upstream function from finn-base to finn. · 21e19d02
  Hendrik Borras authored 3 years ago
  
  21e19d02
- Updated finn-base commit and refactored changed function names. · 0fda042a
  Hendrik Borras authored 3 years ago
  
  0fda042a
Oct 28, 2021
- [InsertFIFO] handle multiple graph inputs correctly · d6fd1dee
  Yaman Umuroglu authored 3 years ago
  
  d6fd1dee
- [InsertDWC] handle StreamingConcat correctly · ea3522ba
  Yaman Umuroglu authored 3 years ago
  
  ea3522ba
- [DuplStreams] stitching fix · d966e8c8
  Yaman Umuroglu authored 3 years ago
  
  d966e8c8
- [Concat] shape and stitching bugfixes · ca8b97d6
  Yaman Umuroglu authored 3 years ago
  
  ca8b97d6
- Rename parameter · a975909c
  Felix Jentzsch authored 3 years ago
  
  a975909c
Oct 27, 2021
- [SWG] tentative new formula for 1D SWG URAM calc · bc1842f8
  Yaman Umuroglu authored 3 years ago
  
  bc1842f8
- [SetFolding] work with 1D SWG too · 1c5df0b8
  Yaman Umuroglu authored 3 years ago
  
  1c5df0b8
- [Refactor] Concat -> StreamingConcat · c67ecec4
  Yaman Umuroglu authored 3 years ago
  
  c67ecec4
- [DuplStreams] enable HLS conversion + test for n>2 · eda6054d
  Yaman Umuroglu authored 3 years ago
  
  eda6054d
- [DuplStreams] refactor DuplicateStreams to pave way for >2 dupls · 76c992de
  Yaman Umuroglu authored 3 years ago
  
  76c992de
Oct 22, 2021
- [Transform]: bug fix AbsorbTransposeIntoMultiThreshold to support non-linear graphs · 5ced9302
  mmrahorovic authored 3 years ago
  
  5ced9302
- [Concat] fix rtlsim by doing no order reversal during packing · 82360eda
  Yaman Umuroglu authored 3 years ago
  
  this may be a problem in the future
  82360eda
- [Concat] add InferConcatLayer transformation · 1c79b4ca
  Yaman Umuroglu authored 3 years ago
  
  1c79b4ca
- [Concat] fixes to concat op, add to registry · 572dc13f
  Yaman Umuroglu authored 3 years ago
  
  572dc13f
Oct 21, 2021
- [Concat] first sketch of HLSCustomOp for StreamingConcat · 26b762d2
  Yaman Umuroglu authored 3 years ago
  
  26b762d2
- Support for multiple pixel unpacking in apintstream2npy · a6fef4ce
  Felix Jentzsch authored 3 years ago
  
  a6fef4ce
Oct 20, 2021
- Added support for new data type inference of MultiThreshold node in QONNX activation inference. · 67888e0b
  Hendrik Borras authored 3 years ago
  
  67888e0b
- Cleaned up open ToDos. · f3245a81
  Hendrik Borras authored 3 years ago
  
  f3245a81
- Moved ExtractBiasFromConv transformation from FINN to finn-base. · a3ca45c0
  Hendrik Borras authored 3 years ago
  
  a3ca45c0
Oct 19, 2021
- Enabled non-scalar scales for the Trunc node and made bit width readout more robust. · 8a6fb6ff
  Hendrik Borras authored 3 years ago
  
  8a6fb6ff
- Made name of infer_quant_avg_pool_2d module PEP8 conform. · 658c1730
  Hendrik Borras authored 3 years ago
  
  658c1730
- Refactored filter lambda for ConvertQuantActToMultiThreshold transformation to... · 28f93d5a
  Hendrik Borras authored 3 years ago
  
  Refactored filter lambda for ConvertQuantActToMultiThreshold transformation to only use the function for filtering.
  28f93d5a
- Reuse bias tensor when extracting the bias from Conv nodes. · ff3d1413
  Hendrik Borras authored 3 years ago
  
  ff3d1413
Oct 18, 2021
- Made QuantIdentityHandler and FoldQuantWeights compatible with DebugMarker... · 23596d85
  Hendrik Borras authored 3 years ago
  
  Made QuantIdentityHandler and FoldQuantWeights compatible with DebugMarker nodes and added QONNX to the test_brevitas_debug test.
  23596d85
- Renamed BinaryQuant to BipolarQuant. · 828782f0
  Hendrik Borras authored 3 years ago
  
  828782f0
Oct 15, 2021
- Fixed off by one error in input bit width calculation. · 0c63afca
  Hendrik Borras authored 3 years ago
  
  0c63afca
- Fixed a bug in the QuantReluHandler, where the export would fail if the... · 078a4d74
  Hendrik Borras authored 3 years ago
  
  Fixed a bug in the QuantReluHandler, where the export would fail if the activation was at the top of the network.
  078a4d74