Skip to content
Snippets Groups Projects
Select Git revision
  • accl default
  • adegendt/accl_clone
  • dev
  • full_accl_support
  • georg/accl-build
  • georg/accl-complete
  • georg/accl-finn
  • georg/end2end
  • main protected
9 results
You can move around the graph by using the arrow keys.
Created with Raphaël 2.2.027Oct22212019181514131211108764130Sep292827242322212019171613121098731Aug30272320191716131210965426Jul23151312653230Jun282524232117151410987543130May28262520181716141110987653130Apr2927232221201918151412109826Mar25242317543226Feb242322181211108764131Jan29262120171517Dec161413983124Nov231996530Oct282726222120191514131211109876529Sep27262524[SWG] tentative new formula for 1D SWG URAM calc[SetFolding] work with 1D SWG too[Refactor] Concat -> StreamingConcat[DuplStreams] enable HLS conversion + test for n>2[DuplStreams] refactor DuplicateStreams to pave way for >2 dupls[Transform]: bug fix AbsorbTransposeIntoMultiThreshold to support non-linear graphs[Test] add rtlsim/cppsim sweep to concat test[Concat] fix rtlsim by doing no order reversal during packing[Test] add cppsim and rtlsim to concat fpgadataflow test[Concat] add InferConcatLayer transformation[Concat] fixes to concat op, add to registry[Concat] first sketch of HLSCustomOp for StreamingConcatSupport for multiple pixel unpacking in apintstream2npyUpdated Brevitas commit.Added support for new data type inference of MultiThreshold node in QONNX activation inference.Cleaned up open ToDos.Moved ExtractBiasFromConv transformation from FINN to finn-base.Added support for QONNX to test_brevitas_avg_pool_export test.Enabled non-scalar scales for the Trunc node and made bit width readout more robust.Enabled FINNManager export for test_end2end_cybsec_mlp_build.Merge remote-tracking branch 'origin/dev' into feature/qonnx_ingestionUpdated QONNX commitEnable QONNX ingestion for test_brevitas_act_export_relu_imagenet test.Enable QONNX ingestion for test_end2end_cybsec_mlp_* tests.Made name of infer_quant_avg_pool_2d module PEP8 conform.Refactored filter lambda for ConvertQuantActToMultiThreshold transformation to only use the function for filtering.Reuse bias tensor when extracting the bias from Conv nodes.Updated QONNX commit.Made QuantIdentityHandler and FoldQuantWeights compatible with DebugMarker nodes and added QONNX to the test_brevitas_debug test.Added Brevitas foward pass as a point of comparison for test_QONNX_to_FINN test.Merge pull request #401 from Xilinx/feature/lookupRenamed BinaryQuant to BipolarQuant.Enabled TestEnd2End for both FINN and QONNX export.Fixed test_brevitas_avg_pool_export to properly integrate the sign.Added QONNX_export test to test_brevitas_avg_pool_export test (required test modifications).Fixed off by one error in input bit width calculation.Added QONNX_export test to TestEnd2End test.Added QONNX_export test to test_brevitas_act_export_qhardtanh_scaled test.Added QONNX_export test to test_brevitas_act_export_relu test.Added QONNX_export test to test_brevitas_qlinear test.