746 Commits (e73f4c73b9d857d2181092ca3f6034877953150f)

Author SHA1 Message Date
  Megvii Engine Team d977079212 feat(third_party): update cpuinfo 2 years ago
  Megvii Engine Team 217999b1fa feat(arm): add winograd F43 NCHW44 algo and winograd F43 44 algo 2 years ago
  Megvii Engine Team 1529bce525 perf(opencl): add opencl weight transpose kernel 2 years ago
  Megvii Engine Team 5ee0094322 fix(dnn/cuda): fix ptx mma algo compute bugs 2 years ago
  Megvii Engine Team 1404437a90 fix(mgb): fix the compatibility issue of cuda stub with older version drivers 2 years ago
  Megvii Engine Team a6a2646c10 feat(arm): add AlgoFP32Winograd F43, and add filter size into name of winograd-related algorithms 2 years ago
  Megvii Engine Team b8821edb3d perf(dnn/aarch64): optimize aarch64 sigmoid with asm 2 years ago
  Megvii Engine Team 2b99bfec4e feat(arm): supports weight pre-processing for winograd benchmark tests 2 years ago
  Megvii Engine Team 421bcfd3d8 style(mgb/tools): add format for tools, dnn and ci 3 years ago
  Megvii Engine Team 116781ba9c fix(mgb): fix megtee build errors 2 years ago
  Megvii Engine Team 54b5db1729 feat(x86/rvv): add AGENT_NCHW_NCHW44 algo 2 years ago
  Megvii Engine Team eaa180181a feat(x86/rvv): opt gi intrinsic helper 2 years ago
  Megvii Engine Team 399db31aab fix(dnn): fix build 2 years ago
  Megvii Engine Team f31e52d521 feat(mgb): warpperspective support multi src input 2 years ago
  Megvii Engine Team 669816e291 feat(dnn): warpperspective support multi src input 2 years ago
  Megvii Engine Team 1b94380794 fix(dnn): fix reduce sum/mean error when b is large 2 years ago
  Megvii Engine Team c7a9909839 feat(cuda): add int4 ptx 256x64 mma kernel 2 years ago
  Megvii Engine Team cf3ca1e9a2 feat(cuda): add int4 ptx 128x256 mma kernel 2 years ago
  Megvii Engine Team 1f8e930e28 feat(cuda): add int4 ptx 128x128 mma kernel 2 years ago
  Megvii Engine Team 1a2ed8c47b feat(cuda): add convbias ptx algo testcase 2 years ago
  Megvii Engine Team 64551105f9 feat(cuda): add convbias ptx algo 2 years ago
  Megvii Engine Team 8395a459b5 fix(dnn/fallback): fix naive shift multidefination error and optimize GiCvtFromInt32V4ToUint8 2 years ago
  Megvii Engine Team 23a3d13350 fix(dnn/softmax): create redcue and elemwise opr when get workspace size 2 years ago
  Megvii Engine Team b3a7d149a0 feat(dnn/fallback): add some new gi api 2 years ago
  Megvii Engine Team fac67e7c2b feat(gopt): support nchw44 global pooling with fuse_grain 2 years ago
  Megvii Engine Team 43bd949af0 fix(dnn): fix cudnn include 2 years ago
  Megvii Engine Team 8abc3ab8fc fix(imperative): fix convolution in rocm 2 years ago
  Megvii Engine Team 5f86368219 Revert "feat(dnn): add elemwise modes" 2 years ago
  Megvii Engine Team d2a1905ad5 Revert "feat(mgb): add cumprod opr" 2 years ago
  Megvii Engine Team 49e14f87b5 feat(mgb): add cumprod opr 3 years ago
  Megvii Engine Team 87aedc2991 feat(dnn): add elemwise modes 3 years ago
  Megvii Engine Team 25e89d68b0 feat(gi/rvv): remove winograd rvv do not use FIXLEN workaround 2 years ago
  Megvii Engine Team b3f46734e7 feat(megdnn/softmax): add softmax operator in fallback 3 years ago
  Megvii Engine Team c49d3070ba refactor(imperative/ops): extends DnnOprCaller with template 3 years ago
  Megvii Engine Team f5597d9a10 fix(mgb): make error infomation of input channel mismatch more readable 2 years ago
  Megvii Engine Team 38bd599911 fix(mgb): make error infomation of invalid MatMul more readable 2 years ago
  Megvii Engine Team e0d505e6bd fix(mgb/dnn): fix bug that some cutlass file compile very slowly on SM86 2 years ago
  Megvii Engine Team 58ba080d5f feat(x86/rvv): make gi conv algo adapt to vv and vf model 2 years ago
  Megvii Engine Team bd50e457ee feat(x86/rvv): make MATRIX_MUL_GI_F32_4x12 and FP32_GEMV_MK4_GI 2 years ago
  Megvii Engine Team 5c3b4e9584 feat(x86/rvv): opt AlgoFP32WinogradF63_4x4_NCHW44 2 years ago
  Megvii Engine Team fa59a7b061 feat(x86/rvv): opt AlgoF32DirectNCHWNCHW44 2 years ago
  Megvii Engine Team 0d82e9b72b feat(x86/rvv): opt FB_GI_F32_MK4_4x8 2 years ago
  Megvii Engine Team a54d9cb9cd feat(x86/rvv): opt FB_GI_F32_MK4_PACK_4x12 algo 2 years ago
  Megvii Engine Team 247e2f59a4 feat(mgb/dnn): add modes that the output type is bool in elemwise 3 years ago
  Megvii Engine Team 16ba05a81b fix(dnn): fix dnn nchwxx elemwise performance 3 years ago
  Megvii Engine Team 7b17c1180e refactor(dnn): make cudnn_frontend work 3 years ago
  Megvii Engine Team 35e9cc9845 feat(dnn/cuda): add cudnn frontend api 3 years ago
  Megvii Engine Team ab8f6398d9 fix(test): make test install 3 years ago
  Megvii Engine Team 99cfefbfe0 fix(test): fix test copybara 3 years ago
  Megvii Engine Team 0d7ace15c8 fix(mgb/dnn): suport fp16 for resize nhwc 3 years ago