Commit Graph

1107 Commits

Author SHA1 Message Date
jxt1234 1b2e168d6e
Merge pull request #1678 from jokerz0624/acc/GridSample
improvement(GridSample): accelerate GridSample in CPU/Arm82/AVX2/AVX512
2021-10-08 19:34:24 +08:00
jxt1234 f5101c9b2b
Merge pull request #1712 from jun-lv-17/o-master
Fix CPUEltwiseInt8Add calculation issue.
2021-10-08 19:33:28 +08:00
jxt1234 f3bf2e2a3f
Merge pull request #1713 from jokerz0624/feat/support_A15
feat(ios): add Apple A15 support in CPU family
2021-10-08 19:32:54 +08:00
Joker 5feeb1beb9 feat(ios): add Apple A15 support in CPU family 2021-09-30 20:34:08 +08:00
jun.lv a5fa7eb446 Fix CPUEltwiseInt8Add calculation issue. 2021-09-30 14:58:01 +08:00
xiaying 823ae80e57 Fix bug for not change input for nhwc -> nc4hw4 2021-09-30 11:46:54 +08:00
Joker 640993bc56 improvement(GridSample): give areaRemain one better handle in AVX2/AVX512 2021-09-30 11:01:23 +08:00
xiaying bbfc5ecfbf [Onnx:Bugfix] Fix bug for Softmax Default axis set 2021-09-28 14:50:57 +08:00
tianbu.xsw 04bb665a00 fix raster op type print error 2021-09-28 11:35:24 +08:00
tianbu.xsw 18ba76bea6 file write bugfix 2021-09-28 11:35:18 +08:00
jxt1234 ec16cae757
Merge pull request #1705 from jokerz0624/fix/AVX512_detecting
fix(AVX512): fix detecting AVX512 features on Darwin
2021-09-26 17:17:44 +08:00
xiaying a867e23543 Fix compile bug for some ndk 2021-09-25 15:23:06 +08:00
xiaying b550871abb Fix bug for raw cpu winograd crash 2021-09-24 16:05:38 +08:00
周科 e4f0fd58cc fix(AVX512): fix detecting AVX512 features on Darwin 2021-09-24 10:06:47 +08:00
xiaying 55e4f4f110 Fix bug for FuseDupOp not check reduction 2021-09-23 10:23:09 +08:00
Joker c41503556d improvement(GridSample): accelerate GridSample in CPU/Arm82/AVX2/AVX512 2021-09-22 16:04:07 +08:00
jxt1234 69c2459c82
Merge pull request #1675 from alibaba/feature/sync
[MNN:Sync] Sync internal Gitlab
2021-09-18 15:58:24 +08:00
xiaying 03c7b5347b [MNN:Sync] Sync internal Gitlab 2021-09-18 15:52:30 +08:00
xiaying 72022880f0 Fix bug for converter may cause different model, add rank for not convert for rank 2021-09-17 15:22:44 +08:00
xiaying d4d040c57e Fix bug for single convert for NHWC <-> NC4HW4 don't care stride 2021-09-17 15:22:44 +08:00
xiaying 7b7384f515 Fix bug for add tensor convert 2021-09-17 15:22:44 +08:00
xiaying 3ef12f372b Run extra opt pass before graph optimization 2021-09-16 14:04:24 +08:00
xiaying e6db5c69fc Support fuse layernorm for Div Sqrt 2021-09-15 17:22:43 +08:00
xiaying 8894b81afb [Converter:Speed] Fix bug for RemoveInvalidCast invalid for not set op 2021-09-15 17:22:11 +08:00
jxt1234 bc355e84ca
Merge pull request #1635 from DaydreamCoding/patch-7
fix BF16 x86_64 Apple M1 failed
2021-09-15 14:13:53 +08:00
jxt1234 40526149f2
Merge pull request #1640 from jiuzhuanzhuan/master
fix bug of header file missing when build for aarch64 with open ARM82
2021-09-15 14:12:06 +08:00
xiaying 5cb6f8387c Fix bug for demo of segment for PackC8 2021-09-15 14:08:48 +08:00
xiaying 0c26d47a84 [MNN:Bugfix] Fix bug for Squeeze for axis < 0 2021-09-14 21:02:11 +08:00
xiaying a50cb93ad0 Add multithread_imgrecog 2021-09-13 13:26:42 +08:00
jxt1234 a87bd0e8d6
Merge pull request #1661 from szh6/dev_GridSample
feat(OpenCL):add GridSample support on OpenCL Backend
2021-09-09 11:33:40 +08:00
shufu 8e464d290c feat(OpenCL):add GridSample support on OpenCL Backend 2021-09-08 14:11:25 +08:00
xiaying 3a0fb480d2 Fix crash bug for origin quan model 2021-09-06 17:18:29 +08:00
xiaying 8aea8fd021 Fix bug for multi-axis onnxslice error 2021-09-06 17:18:29 +08:00
xiaying 55ce936d9c Support scale input for Interp 2021-09-06 17:18:29 +08:00
xiaying 0d33ceb04d Don't open CheckNAN for MSVC compile 2021-09-06 17:18:29 +08:00
jxt1234 d21fd2a910
Merge pull request #1650 from jokerz0624/vulkan_GridSample
Vulkan GridSample
2021-09-04 07:49:32 +08:00
tianbu.xsw 612199d0ee quant weight valid range issue 2021-09-03 17:06:24 +08:00
Joker 2c224c8ac8 fix(test): fix a bug of dataFormat in GridSampleTest.cpp 2021-09-01 17:37:51 +08:00
Joker 3461faa390 feat(Vulkan): add GridSample op support in Vulkan backend 2021-09-01 15:00:56 +08:00
xiaying 5dfe97e4c8 Fix bug for dim = 0's shape compute 2021-08-30 13:34:11 +08:00
xiaying 575a2c97dd Fix bug for CH Fused but W not fused fastblit error 2021-08-25 17:23:36 +08:00
qingzhu 4eda674234 fix bug of header file missing when build for aarch64 with open ARM82 2021-08-24 20:06:46 +08:00
DaydreamCoding 264a6039bb fix BF16 x86_64 Apple M1 failed 2021-08-19 18:39:51 +08:00
jxt1234 ff625ee97f
Merge pull request #1634 from alibaba/feature/bugfix
[MNN:Bugfix]fix illegal opcode(MV) bug for some GNU compiler
2021-08-18 19:55:59 +08:00
yuyang f1997b9a5f [MNN:Bugfix]fix illegal opcode(MV) bug for some GNU compiler 2021-08-18 18:13:00 +08:00
jxt1234 6ad2a632fd
Merge pull request #1609 from DaydreamCoding/patch-5
adapt MSVC
2021-08-12 13:00:14 +08:00
jxt1234 b8d8fc9d73
Merge pull request #1612 from MambaWong/master
fix libMNN.so: undefined symbol
2021-08-12 12:56:32 +08:00
xiaying e9d38acd6b Fix Prelu bug for multi-batch 2021-08-11 10:21:31 +08:00
xiaying 312b003f4c [MNN:Bugfix] Fix bug for ARM TopKV2 different 2021-08-05 19:41:37 +08:00
jason_w 6aca4ea6b9 fix symbol lookup error: libMNN.so: undefined symbol:
option(MNN_SUPPORT_TFLITE_QUAN "Enable MNN's tflite quantized op" OFF)
2021-08-05 11:08:13 +08:00