86 Commits (c269a690e596ac7bbc69dda6c9b197aacf3efabc)

Author SHA1 Message Date
  Megvii Engine Team 36b1ba052f fix(mgb/dnn): fix cudnn8.0.4 convbias with z 4 years ago
  Megvii Engine Team 40bab1ed66 feat(log): opt log, enable mgb sdk log at opt build 4 years ago
  Megvii Engine Team 2d18074a70 fix(mgb): fix spell error 4 years ago
  Megvii Engine Team a437ec8e88 fix(src/gopt): add replace func of typecvt opr for nhwcd4 pass 4 years ago
  Megvii Engine Team 04b1a45af4 fix(dnn): fix cudnn crash when finalize called after cudnn dtor 4 years ago
  Megvii Engine Team 14a089c49d fix(dnn): change ci to cudnn804, reopen testcase 4 years ago
  Megvii Engine Team ba2ad46e54 feat(gopt): add deconv nchw4 int8 opt pass, add deconv nchw int8 4 years ago
  Megvii Engine Team 9da26407ea feat(ci): and model to test algo policy compatible 4 years ago
  Megvii Engine Team a3ea1f153c feat(mgb/opr): add fast profile and combined Execution strategy 5 years ago
  Megvii Engine Team c82d88751a fix(dnn/cuda): add cuda nchw int8 conv impl with nchw4 to fix cu111 compatibility 4 years ago
  Megvii Engine Team 51868533c8 fix(mgb/gopt): fix opt pass elementwise operation shape issue at tranform to NCHW4 4 years ago
  Megvii Engine Team cf27dd642c fix(cuda): use cudnn8.0.4 as cu111 default libs 5 years ago
  Megvii Engine Team 649e4dd750 test(cuda): fix test for cu111 5 years ago
  Megvii Engine Team 2e4b9a42f7 fix(mgb/gopt): fix folding conv dimshuffle opt pass 5 years ago
  Megvii Engine Team e82fa4ec23 fix(gopt): using new_inp for build_chain in DelayBroadcast pass 5 years ago
  Megvii Engine Team 364afec033 chore(mge): update copyright years 5 years ago
  Megvii Engine Team 1b1ad56a82 fix(mgb/gopt): fix warp fusion opt pass 5 years ago
  Megvii Engine Team 0f739c1118 fix(mgb/gopt): fix NormalizeArithChainPass to process sub-div chains 5 years ago
  Megvii Engine Team 55042195d4 chore(winograd): add Convolutionv2 param 5 years ago
  Megvii Engine Team 5063a2063b feat(mgb): add fastrun support for matmul/batchedmatrixmul 5 years ago
  Megvii Engine Team d07cfdcbe9 refactor(mgb): move convolution mixin to search_policy 5 years ago
  Megvii Engine Team 7f3f9a94ae feat(mgb/core): add shape hint for graph optimization 5 years ago
  Megvii Engine Team 4e9be159f7 feat(mgb/gopt): add opt pass for fusing convolution and reformat 5 years ago
  Megvii Engine Team 61f917fb8e feat(dnn/cuda): add impl for fusing warp perspective and dimshuffle 5 years ago
  Megvii Engine Team fc0fcd2f7f chore(winograd): remove winograd transform code 5 years ago
  Megvii Engine Team 3bf73ff16f feat(dnn): add cuda preprocess fusion 5 years ago
  Megvii Engine Team 5f171298aa feat(mgb/gopt): add AxisAddRemove opr support for cd4 opt pass 5 years ago
  Megvii Engine Team 6f5d0febf1 perf(dnn/cuda): enhance performance for pooling forward 5 years ago
  Megvii Engine Team 7cd71c3102 fix(mgb/gopt): fix cd4 elewise transform 5 years ago
  Megvii Engine Team cab46f2a00 fix(dnn): fix relayout format when group conv group=1 5 years ago
  Megvii Engine Team 4afa4b72c3 fix(mgb/oppass): fix fuse conv bias pass when bias is full bias and bias 5 years ago
  Megvii Engine Team 03d0cc0281 feat(gopt): add remove redundant copy pass 5 years ago
  Megvii Engine Team 40e778fbd1 fix(mgb): fix-deconv-io16c32-pass-fail 5 years ago
  Megvii Engine Team ee2e2b3c7b fix(mgb/gopt): fix nchwxx optpass of no handle conv_bias opr which with no bias 5 years ago
  Megvii Engine Team 59a9275c66 fix(mge): fix optimize_for_inference during trace.dump 5 years ago
  Megvii Engine Team 8f44d6ea60 fix(mgb): fix optpass fail at transform NCHW to NCHW4 when input dtype is float 5 years ago
  Megvii Engine Team fade97d4ef fix(mgb/gopt): fix convert batchnorm to elemwise pass issue 5 years ago
  Megvii Engine Team 78d7d400d1 feat(opr): add a constant flavor for SharedDeviceTensor 5 years ago
  Megvii Engine Team c20d4cc6dc feat(dnn): fix opt pass nchw44 can not dump resnet 5 years ago
  Megvii Engine Team 56381f808b fix(dnn/arm): use vcvtq_f32_s32 for all arm code 5 years ago
  Megvii Engine Team 1173205726 fix(gopt): nchw_nchwxx useable and opt pass use nchw_nchwxx_valid 5 years ago
  Megvii Engine Team eb18eba87d fix(gopt): fix nchw44 nchw44_dot gopt test 5 years ago
  Megvii Engine Team eab7ab0530 fix(gopt): gen nchw_nchw44 when kernel is optimized 5 years ago
  Megvii Engine Team 777f3ea970 refactor(gopt): format code 5 years ago
  Megvii Engine Team 80c4705317 perf(mgb): use midout in megbrain to reduce binary size 5 years ago
  Megvii Engine Team 4348960c40 fix(mge/gopt): fix fp16 compute mode 5 years ago
  Megvii Engine Team 9f4060b050 fix(mgb/gopt): add ShuffleShuffleRemovePass assert 5 years ago
  Megvii Engine Team c1fb3c8c1a fix(mgb/gopt): fix ConvertF32ToF16Pass endpoints 5 years ago
  Megvii Engine Team ab3efab066 fix(mgb): add compute capability check 5 years ago
  Megvii Engine Team 8070f40aa1 fix(mgb/gopt): fix gopt nchwxx convert elemwise and reshape 5 years ago