root
/
MNN
mirror of https://github.com/alibaba/MNN.git
1
0
Fork 0
Code Issues Actions 1 Packages Projects Releases Wiki Activity
383 Commits 20 Branches 107 Tags 393 MiB
Commit Graph

114 Commits

Author SHA1 Message Date
hebin c0cb82d9ab [PATCH 22/28] [MNN:Speed] 8x8 Gemm and cache prefetch optimize 2020-03-31 11:12:40 +08:00
誉阳 cef64b2ea2 [PATCH 19/28] delete useless macro 2020-03-31 11:12:39 +08:00
誉阳 6dc7c7a95f [PATCH 18/28] close compile sdot default 2020-03-31 11:12:39 +08:00
xiaying f1cc216846 [PATCH 13/28] [MNN:Speed] Fix bug for winograd cost compute 2020-03-31 11:12:39 +08:00
akshaysharma01 593255ec87 Required library is added in source/backend/cpu/CPURuntime.hpp 2020-03-27 08:54:01 +05:30
xiaying 62279761af [PATCH 2/2] [ARM82:Bugfix] Remove log for not support op 2020-03-22 21:15:00 +08:00
xiaying 48c92a41e7 [MNN:Sync] Sync internal git for remain patch 2020-03-22 20:33:03 +08:00
liqing e287f7c89d [PATCH 158/160] [bugfix] fix metal softmax in half type 2020-03-22 19:02:15 +08:00
liqing 652e5c5979 [PATCH 157/160] [bugfix] fix softmax 2020-03-22 19:02:15 +08:00
liqing 327da9234e [PATCH 156/160] [Bugfix] fix matmul 2020-03-22 19:02:15 +08:00
xiaying aac271774a [PATCH 147/160] [Vulkan:Bugfix] Add config for high precision because adreno don't support 2020-03-22 19:02:14 +08:00
xiaying 34f7abce16 [PATCH 130/160] [Vulkan:Bugfix] Fix bug for other op use reshape 2020-03-22 19:02:12 +08:00
xiaying ff5d81ec22 [PATCH 129/160] [Vulkan:Refractor] Delete gemm16x16Half, replace by macro 2020-03-22 19:02:12 +08:00
xiaying 1c1ee5b2b3 [PATCH 128/160] [Vulkan:Feature] Support squeeze, expandim 2020-03-22 19:02:12 +08:00
xiaying cd8e4d17d9 [PATCH 126/160] [Vulkan:Speed] Optimize the pre/post treat of deconvolution 2020-03-22 19:02:12 +08:00
xiaying c4d8538862 [PATCH 125/160] [Vulkan:Speed] Optimize convolution's im2col and col2im's localsize 2020-03-22 19:02:12 +08:00
xiaying 7f12a224e5 [PATCH 124/160] [Vulkan:Refractor] Rename VulkanMatrixMultier as VulkanMatrixMultier4x4 2020-03-22 19:02:12 +08:00
xiaying 9131b1780f [PATCH 121/160] [Train:Refractor] Remove no use input of CPUConv2DBackPropFilter 2020-03-22 19:02:11 +08:00
xiaying c77d7c4ab3 [PATCH 120/160] [Vulkan:Feature] Support multi-input for deconvolution 2020-03-22 19:02:11 +08:00
xiaying 076194754f [PATCH 116/160] [Vulkan:Bugfix] Fix bug for pad error of batchtospacend and spacetobatchnd 2020-03-22 19:02:11 +08:00
xiaying 905d4332f2 [PATCH 115/160] [Vulkan:Refractor] Use common convolution and deconvolution pad 2020-03-22 19:02:11 +08:00
xiaying dbd3267c09 [PATCH 114/160] [MNN:Refractor] Seperate the pad compute for convolution and deconvolution 2020-03-22 19:02:11 +08:00
xiaying 2ef436377e [PATCH 113/160] [MNN:Bugfix] Fix bug for pool grad don't use pad 2020-03-22 19:02:11 +08:00
xiaying 0c18168e1b [PATCH 110/160] [Metal:Bugfix] Fix compile bug for metal 2020-03-22 19:02:10 +08:00
xiaying 4dec63d499 [PATCH 103/160] [Vulkan:Feature, MNN:Refract] Support Reduce for vulkan 2020-03-22 19:02:10 +08:00
xiaying 9c697a4f46 [PATCH 100/160] [Vulkan:Feature] Support broadcast for binary, opt image add/sub 2020-03-22 19:02:09 +08:00
xiaying 2d71a17c47 [PATCH 099/160] [Vulkan:Bugfix, Demo:Bugfix] Fix bug for p20 can't run vulkan, Fix bug for android demo compile 2020-03-22 19:02:09 +08:00
xiaying bcf5b703b1 [PATCH 097/160] [Vulkan:Bugfix] fix deconvolution-dw crash bug, add print for copy buffer 2020-03-22 19:02:09 +08:00
xiaying eb132b63b0 [PATCH 096/160] [Vulkan:Feature] Support nc4hw4 unary 2020-03-22 19:02:09 +08:00
xiaying 6037692a07 [PATCH 095/160] [Vulkan:Bugfix] Fix bug for nchw convert, which cast kh != kw's convolution error 2020-03-22 19:02:09 +08:00
xiaying 956d528026 [PATCH 094/160] [MNN:Refractor] Sepearte compute dim from CPUBinary 2020-03-22 19:02:09 +08:00
xiaying 9697d157e0 [PATCH 093/160] [Vulkan:Feature] Support neg, sign, log, square 2020-03-22 19:02:09 +08:00
xiaying c024ac5243 [PATCH 091/160] [Vulkan:Bugfix] Fix bug for compute of MatMul 2020-03-22 19:02:08 +08:00
xiaying 86d7732ddc [PATCH 089/160] [Vulkan:Feature] Support Relu / Relu6 with buffer, add matmul 2020-03-22 19:02:08 +08:00
xiaying 1a49ff15f8 [PATCH 088/160] [Vulkan:Feature] Support multi-input convolution, remove unuseful shader 2020-03-22 19:02:08 +08:00
xiaying efcd54e695 [PATCH 083/160] [Vulkan:Bugfix] Fix compute bug for softmax buffer 2020-03-22 19:02:07 +08:00
xiaying 7f5fb2c414 [PATCH 079/160] [Vulkan:Feature] Support large feature map for convolution im2col+GEMM 2020-03-22 19:02:07 +08:00
xiaying d96b4d979a [PATCH 077/160] [Vulkan:Refractor] Seperate VulkanTensor 2020-03-22 19:02:07 +08:00
xiaying cc94cba6b3 [PATCH 072/160] [OpenCL:Bugfix] Fix compile bug for opencl use int8 load 2020-03-22 19:02:06 +08:00
xiaying 79640387e5 [PATCH 071/160] [Vulkan:Feature] Support idst quan and fp16 model 2020-03-22 19:02:06 +08:00
xiaying 010913aaaa [PATCH 070/160] [MNN:Refract] Seperate the load of quan and half to ConvolutionCommon 2020-03-22 19:02:06 +08:00
xiaying 161ee18693 [PATCH 069/160] [Vulkan:Bugfix] Fix bug for direct mode set error 2020-03-22 19:02:06 +08:00
xiaying 3a3d56ab46 [PATCH 067/160] [Vulkan:Bugfix] Fix compile bug for gcc don't link dl default 2020-03-22 19:02:06 +08:00
xiaying 091330447c [PATCH 063/160] [Vulkan:Feature] Support precision_high (use fp32) 2020-03-22 19:02:06 +08:00
xiaying 71bb5597b9 [PATCH 062/160] [Vulkan:Bugfix] Fix execution crash when not support, test will not crash 2020-03-22 19:02:05 +08:00
xiaying a243e4b98f [PATCH 061/160] [Vulkan:Bugfix] Support python3 for makeshader.py 2020-03-22 19:02:05 +08:00
xiaying 46c11b3128 [PATCH 060/160] [Vulkan:Feature] Support use system lib for vulkan 2020-03-22 19:02:05 +08:00
xiaying 0378386606 [PATCH 053/160] [OpenCL:Refract] Add hpp file in xcode project 2020-03-22 19:02:05 +08:00
xiaying f1aec2cf97 [PATCH 052/160] [OpenCL:Bugfix] Fix bug for deconv with two input 2020-03-22 19:02:04 +08:00
hebin d7660d1b5b [PATCH 051/160] opencl conv without bias 2020-03-22 19:02:04 +08:00