p59230764
/
InfiniTensor
forked from jiuyuan/InfiniTensor
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity
239 Commits 70 Branches 0 Tags 141 MiB
Commit Graph

239 Commits

Author SHA1 Message Date
Liyan Zheng 1ee4a60af0 Add: convert expression to operator 2023-06-28 11:06:17 +08:00
Liyan Zheng 29071ddcac Rename: Expr construction helpers 2023-06-25 20:31:08 +08:00
Liyan Zheng c6c445991a Add: enable mutator search in python 2023-06-25 20:18:18 +08:00
Liyan Zheng d25b606e12 Add: TF32 supports and accurate timing for conv 2023-05-07 13:22:39 +08:00
Liyan Zheng abcfa76fb5 Add: efficient CUDA transpose for last two dims 2023-05-05 15:16:07 +08:00
Liyan Zheng 6a70555892 Add: TensorRT backend 2023-04-30 23:44:10 +08:00
Liyan Zheng f47a411095 Add: export with random weight 2023-04-30 22:25:07 +08:00
Liyan Zheng df2534d209 Fix: fuse Relu to GEMM/Conv 2023-04-30 16:16:16 +08:00
Liyan Zheng a1f02593d3 Add: export Flatten operator to ONNX 2023-04-30 16:15:24 +08:00
Liyan Zheng 65b4b42fa0 Merge remote-tracking branch 'origin/NNET_OpSearch' into NNET_e2e 
Fix: update of derivator.h is missing in NNET_OpSearch
 2023-04-30 15:48:48 +08:00
Liyan Zheng b068442bfb Add: print time in op evalution 2023-04-30 00:47:57 +08:00
Liyan Zheng c6e7748786 Add: rule of Any+Relu->Any 2023-04-28 21:46:54 +08:00
Liyan Zheng d0ae48d21d Add: CUDA Matmul selection 2023-04-28 19:13:19 +08:00
Liyan Zheng c875f3cbb8 Add: Matmul Transpose plans 2023-04-28 19:13:01 +08:00
Liyan Zheng 95a8b90fa7 Fix: add virutal method sync in Runtime 2023-04-28 00:44:54 +08:00
Liyan Zheng c58b67f743 Chore: suppress output 2023-04-26 14:11:53 +08:00
Liyan Zheng 75c9226164 Merge branch 'NNET_e2e' into NNET_op_test 2023-04-25 04:32:18 +08:00
Liyan Zheng f877eca517 Add: IT_ASSERT in checkCudaError 2023-04-25 04:25:10 +08:00
Liyan Zheng b13b799fbe All mdoels E2E 2023-04-25 04:24:43 +08:00
Liyan Zheng 350fc01d39 Add: Search Depth 2023-04-25 01:07:21 +08:00
whjthu 71f4f6e9d9 add op test for einnet 2023-04-24 21:48:29 +08:00
Liyan Zheng 1408d308cc Add: FCRS log 2023-04-24 21:20:29 +08:00
Liyan Zheng 11229a2baa Add: Figure 17 2023-04-24 21:07:30 +08:00
Liyan Zheng 2b85ac41ef Fix: CUDA Relu for 2D/1D tensor 2023-04-24 16:14:26 +08:00
Liyan Zheng 1e46750159 Add conv2bgemm and fix mutator::runtime 2023-04-24 13:12:40 +08:00
Liyan Zheng 079985bc8c Add: efficient transpose 2023-04-24 13:08:29 +08:00
Liyan Zheng c1275cddb6 Fix: conv2dreduce_kernel_ offset 2023-04-24 02:30:41 +08:00
Liyan Zheng 51cc042f56 Add: nchw to nhwc conversion 
Fix: conv parameter error in to_onnx
 2023-04-24 02:29:53 +08:00
Liyan Zheng 18d6ba4022 Merge branch 'NNET_e2e' into NNET_gcn 2023-04-23 23:20:46 +08:00
Liyan Zheng 4211fd1f32 Fix: matmul transpose in convNHWC2gemm rule 2023-04-23 22:54:50 +08:00
xxcclong 8409c1f9d4 tested fsrcnn 2023-04-23 22:19:51 +08:00
xxcclong 830b28913c better transposed convreduce 2023-04-23 21:36:25 +08:00
Liyan Zheng 1ba78d7f89 Add: reduce in Any 2023-04-23 21:36:12 +08:00
xxcclong 777aebafc9 fsrcnn 2023-04-23 20:56:19 +08:00
whjthu 131a679340 gcn optimization 2023-04-23 13:43:41 +08:00
Liyan Zheng 5df2524ff9 Merge branch 'NNET_eliminateOP' into NNET_e2e 2023-04-23 13:35:29 +08:00
Liyan Zheng f204866d93 Fix: reduce workspace size 2023-04-23 13:34:07 +08:00
Liyan Zheng b9819e65c1 Fix: allow eliminate and fusion failure in search 2023-04-23 13:15:34 +08:00
Liyan Zheng 7277356744 Add: Reshape/Transpose elimination 2023-04-23 02:10:05 +08:00
whjthu f820117acd fix unused code 2023-04-23 00:18:26 +08:00
whjthu 1ab2118716 add AnyOp and cuda kernel 2023-04-23 00:16:03 +08:00
huangshuhong ff97c879fb add ConvNHWC and FSRCNN graph 2023-04-23 00:02:22 +08:00
Liyan Zheng acc64fd32c Merge branch 'NNET_transpose' into NNET_e2e 
Fix: gridSize and blockSize in Reshape kernel
 2023-04-22 21:32:31 +08:00
Liyan Zheng 33ab5dcd3e Fix: gbmm kernel 2023-04-22 21:14:52 +08:00
Liyan Zheng e2f18272c9 Add: no malloc for reshape outputs 2023-04-22 21:13:57 +08:00
Liyan Zheng 40e6db6608 Add: tensor FUID in exported ONNX 2023-04-22 20:28:17 +08:00
Liyan Zheng c451918224 Fix: tensor size overflow 2023-04-22 20:28:00 +08:00
whjthu 34ed298725 fix format 2023-04-22 17:00:52 +08:00
whjthu 664f0dbe02 support cuda transpose 2023-04-22 16:57:27 +08:00
Liyan Zheng a732b6f176 Fix: ignore transpose in CudaGraph since no kernel 2023-04-22 16:08:40 +08:00