Megvii Engine Team
|
fa4883389a
|
feat(dnn,imperative): remove the restriction of tensor shape when using uint8 region mask
GitOrigin-RevId: 37d9900697
|
2 years ago |
Megvii Engine Team
|
0ebd4400d5
|
fix(dnn): fix the modulo of int
GitOrigin-RevId: 6f7280246b
|
2 years ago |
Megvii Engine Team
|
68d2710810
|
fix(build): remove build so many warning on windows
GitOrigin-RevId: 124b8975db
|
2 years ago |
Megvii Engine Team
|
582dd4ceb8
|
fix(dnn/sfotmax): call cpu dispatch for softmax opr
GitOrigin-RevId: a606e66101
|
2 years ago |
Megvii Engine Team
|
235d81ddb0
|
feat(dnn): add fp16 nchw88 im2col algo
GitOrigin-RevId: a6d6cb4fc7
|
2 years ago |
Megvii Engine Team
|
dbd9483993
|
feat(dnn,src,imperative): add groupnorm op
GitOrigin-RevId: de3c3d10e5
|
2 years ago |
Megvii Engine Team
|
0a52b2587e
|
fix(opencl/test): fix test weight preprocess filter UAF issue
GitOrigin-RevId: de5e71147b
|
2 years ago |
Megvii Engine Team
|
8fe8edf4d6
|
feat(dnn): add fp16 mk8 16x12 matmul algo
GitOrigin-RevId: d9978fa589
|
2 years ago |
Megvii Engine Team
|
f444d4fe4d
|
feat(dnn,imperative): region restricted conv support groups=1 even if
depthwise
GitOrigin-RevId: 950d2f4889
|
2 years ago |
Megvii Engine Team
|
fa9d719f7e
|
fix(gopt): fix global layout transform fold conv typecvt
GitOrigin-RevId: 66a23a927e
|
2 years ago |
Megvii Engine Team
|
ece454fd46
|
fix(third_party): fix cpuinfo related to sve2
GitOrigin-RevId: 51a3facd05
|
2 years ago |
Megvii Engine Team
|
6db4620e6d
|
feat(dnn): fix wgrad rrconv for compute capability
GitOrigin-RevId: ba8792d7a9
|
2 years ago |
Megvii Engine Team
|
4e9b1c4eee
|
feat(dnn): add rrconv wgrad, support int32 and uint8 region mask
GitOrigin-RevId: 0da9b3bca8
|
2 years ago |
Megvii Engine Team
|
977c207171
|
feat(dnn): add RegionRestrictedConv DGRAD support int32 and uint8
GitOrigin-RevId: 814b8a83f8
|
2 years ago |
Megvii Engine Team
|
543c9b77a8
|
feat(dnn): add RegionRestrictedConv cuda
GitOrigin-RevId: b9f2d34a13
|
2 years ago |
Megvii Engine Team
|
fdec82ece5
|
feat(dnn): add naive RegionRestrictedConv
GitOrigin-RevId: 5af93f7ccb
|
2 years ago |
Megvii Engine Team
|
e9cc523741
|
fix(mgb): format code
GitOrigin-RevId: 3e58539a2f
|
2 years ago |
huangxinda
|
a07fbf79f7
|
Merge pull request #484 from wangxiang9603:add-nchw44-deconv
GitOrigin-RevId: 444429a625
|
2 years ago |
Megvii Engine Team
|
ec234135a6
|
feat(lite): support discrete inputs
GitOrigin-RevId: 25ce8da275
|
2 years ago |
Megvii Engine Team
|
58b682ca00
|
feat(dnn/cuda): add naive bmm
GitOrigin-RevId: 4ba4b22e40
|
3 years ago |
Megvii Engine Team
|
edd3ee67ce
|
fix(mgb): add error infomation for old version load new elemwise mode
GitOrigin-RevId: d425425d2f
|
2 years ago |
Megvii Engine Team
|
a7e28ebe8c
|
fix(dnn): fix winograd load error and cpuinfo test error
GitOrigin-RevId: 32c9ee58d1
|
2 years ago |
Megvii Engine Team
|
41b9db85e2
|
fix(mgb): make error infomation of advanced indexing out of bound more readable
GitOrigin-RevId: 69e6f32e7e
|
2 years ago |
Megvii Engine Team
|
f0291883b6
|
fix(mgb): make error infomation of group conv input channel mismatch more readable
GitOrigin-RevId: d249408c26
|
2 years ago |
Megvii Engine Team
|
d977079212
|
feat(third_party): update cpuinfo
GitOrigin-RevId: 67c7b8e8cc
|
2 years ago |
wangxiang
|
fb2329e9db
|
feat(dnn) add nchw44 deconv
|
2 years ago |
Megvii Engine Team
|
217999b1fa
|
feat(arm): add winograd F43 NCHW44 algo and winograd F43 44 algo
GitOrigin-RevId: a981b2f61b
|
2 years ago |
Megvii Engine Team
|
1529bce525
|
perf(opencl): add opencl weight transpose kernel
GitOrigin-RevId: 1102ecf4bb
|
2 years ago |
Megvii Engine Team
|
5ee0094322
|
fix(dnn/cuda): fix ptx mma algo compute bugs
GitOrigin-RevId: 19628d0c94
|
2 years ago |
Megvii Engine Team
|
1404437a90
|
fix(mgb): fix the compatibility issue of cuda stub with older version drivers
GitOrigin-RevId: 628afbf3cf
|
2 years ago |
Megvii Engine Team
|
a6a2646c10
|
feat(arm): add AlgoFP32Winograd F43, and add filter size into name of winograd-related algorithms
GitOrigin-RevId: 909503a90d
|
2 years ago |
Megvii Engine Team
|
b8821edb3d
|
perf(dnn/aarch64): optimize aarch64 sigmoid with asm
GitOrigin-RevId: 7d54d67669
|
2 years ago |
Megvii Engine Team
|
2b99bfec4e
|
feat(arm): supports weight pre-processing for winograd benchmark tests
GitOrigin-RevId: 1797f3b91c
|
2 years ago |
Megvii Engine Team
|
421bcfd3d8
|
style(mgb/tools): add format for tools, dnn and ci
GitOrigin-RevId: 5684e5ea43
|
3 years ago |
Megvii Engine Team
|
116781ba9c
|
fix(mgb): fix megtee build errors
GitOrigin-RevId: b351dd3994
|
2 years ago |
Megvii Engine Team
|
54b5db1729
|
feat(x86/rvv): add AGENT_NCHW_NCHW44 algo
GitOrigin-RevId: 8cf6c3fac0
|
2 years ago |
Megvii Engine Team
|
eaa180181a
|
feat(x86/rvv): opt gi intrinsic helper
for rvv, detail: https://github.com/riscv-collab/riscv-gnu-toolchain/issues/1106
GitOrigin-RevId: 27615584c0
|
2 years ago |
Megvii Engine Team
|
399db31aab
|
fix(dnn): fix build
GitOrigin-RevId: d91077248a
|
2 years ago |
Megvii Engine Team
|
f31e52d521
|
feat(mgb): warpperspective support multi src input
GitOrigin-RevId: 0887656864
|
2 years ago |
Megvii Engine Team
|
669816e291
|
feat(dnn): warpperspective support multi src input
GitOrigin-RevId: 8a4789852e
|
2 years ago |
Megvii Engine Team
|
1b94380794
|
fix(dnn): fix reduce sum/mean error when b is large
GitOrigin-RevId: d1bae619b1
|
2 years ago |
Megvii Engine Team
|
c7a9909839
|
feat(cuda): add int4 ptx 256x64 mma kernel
GitOrigin-RevId: 8f7475b0f6
|
2 years ago |
Megvii Engine Team
|
cf3ca1e9a2
|
feat(cuda): add int4 ptx 128x256 mma kernel
GitOrigin-RevId: 1ae7c9f034
|
2 years ago |
Megvii Engine Team
|
1f8e930e28
|
feat(cuda): add int4 ptx 128x128 mma kernel
GitOrigin-RevId: 5a8b9c3f8e
|
2 years ago |
Megvii Engine Team
|
1a2ed8c47b
|
feat(cuda): add convbias ptx algo testcase
GitOrigin-RevId: 9ad6d4561f
|
2 years ago |
Megvii Engine Team
|
64551105f9
|
feat(cuda): add convbias ptx algo
GitOrigin-RevId: 08e9f66641
|
2 years ago |
Megvii Engine Team
|
8395a459b5
|
fix(dnn/fallback): fix naive shift multidefination error and optimize GiCvtFromInt32V4ToUint8
GitOrigin-RevId: 6660c35214
|
2 years ago |
Megvii Engine Team
|
23a3d13350
|
fix(dnn/softmax): create redcue and elemwise opr when get workspace size
GitOrigin-RevId: 476a39bdd3
|
2 years ago |
Megvii Engine Team
|
b3a7d149a0
|
feat(dnn/fallback): add some new gi api
GitOrigin-RevId: 4aede0ac6a
|
2 years ago |
Megvii Engine Team
|
fac67e7c2b
|
feat(gopt): support nchw44 global pooling with fuse_grain
GitOrigin-RevId: 4c43a149f8
|
2 years ago |