Megvii Engine Team
|
0447574446
|
feat(opencl): add OpenCL cache compat level api
GitOrigin-RevId: e7561e6879
|
3 years ago |
Megvii Engine Team
|
6fb5a34360
|
build(flatbuffer/cx2): fix cx2 build and fix uclibc build flatbuffer
GitOrigin-RevId: af851e155f
|
3 years ago |
Megvii Engine Team
|
87de704a46
|
feat(gopt): fuse conv h_swish
GitOrigin-RevId: a3d12991fb
|
3 years ago |
Megvii Engine Team
|
4adba37867
|
feat(lite): add example script and some small change for lar
GitOrigin-RevId: a28ed2f27a
|
3 years ago |
Megvii Engine Team
|
87f00232f2
|
fix(mge/gm): fix missing dtype checking while attach tensors
GitOrigin-RevId: f0aaea99b9
|
3 years ago |
Megvii Engine Team
|
3726f5cc92
|
feat(gopt): merger consecutive relayout and dimshuffle to one relayout to optimize CD4 performarce
GitOrigin-RevId: a058776be3
|
3 years ago |
Megvii Engine Team
|
1fead9b6b0
|
feat(gopt): merge consecutive dimshuffle and relayout to one relayout to optimize CD4 performace
GitOrigin-RevId: 16f22baa80
|
3 years ago |
Megvii Engine Team
|
26d1e4f7ed
|
feat(gopt): optimize cd4 pass rule for elemwise and typecvt to let cd4 start as soon as possible
GitOrigin-RevId: 6580dedca7
|
3 years ago |
Megvii Engine Team
|
ac26bdcef5
|
fix(cuda): fix direct conv speed and memory problem
GitOrigin-RevId: 6faeeff3b8
|
3 years ago |
Megvii Engine Team
|
f7994683bd
|
feat(cuda): add large kernel direct conv to heuristic algo chooser
GitOrigin-RevId: bc927b6df7
|
3 years ago |
Megvii Engine Team
|
6dc0c0b9cc
|
fix(dnn): fix the sync problem in some kernels
GitOrigin-RevId: df3f7dc51b
|
3 years ago |
Megvii Engine Team
|
04193e3bd1
|
feat(dnn): add nearest mode for remap and resize
GitOrigin-RevId: 31e7b72a78
|
3 years ago |
Megvii Engine Team
|
69b89388e8
|
docs(mge/functional): fix debug_param set_execution_strategy docstring
GitOrigin-RevId: 434929c998
|
3 years ago |
Megvii Engine Team
|
93c7e45188
|
feat(arm): delete the reduant implement
GitOrigin-RevId: ff32a3dc8b
|
3 years ago |
Megvii Engine Team
|
e34a642b31
|
feat(fallback): reduce support general intrinsic
GitOrigin-RevId: f250aa7b2a
|
3 years ago |
Megvii Engine Team
|
10f23778a8
|
feat(fallback): add simd general intrinsic
GitOrigin-RevId: ad78ba689f
|
3 years ago |
Megvii Engine Team
|
286051ede1
|
feat(dnn): differentiate sass kernel with cuda version
GitOrigin-RevId: 40bb4423b8
|
3 years ago |
Megvii Engine Team
|
f78b60ec10
|
feat(bazel): make bazel gensass depend on cuda toolchain version automatically
GitOrigin-RevId: 9433f21a91
|
3 years ago |
Megvii Engine Team
|
f48227c07d
|
feat(mgb): show more details for cuda driver api call
GitOrigin-RevId: 40e63d9dac
|
3 years ago |
Megvii Engine Team
|
bb5af9b475
|
feat(lite): hidden lar gflags symbols for static link
GitOrigin-RevId: 28823da644
|
3 years ago |
Megvii Engine Team
|
d8bb3ff5b4
|
fix(cuda): fix fp16 tensorcore gemm split k workspace
GitOrigin-RevId: d04a0e0985
|
3 years ago |
Megvii Engine Team
|
597efed40b
|
feat(lite): add get last error code interface in lite c
GitOrigin-RevId: 280cc88092
|
3 years ago |
Megvii Engine Team
|
90c8a58cca
|
docs(docstring): add pad docstring
GitOrigin-RevId: eaf6a87456
|
3 years ago |
Megvii Engine Team
|
5f4501e0f3
|
fix(gopt): fix conv bias fuse 2 noline
GitOrigin-RevId: a6ab9f4e5e
|
3 years ago |
Megvii Engine Team
|
ac2f548c9a
|
docs(imperative/dataloader): update preload description
GitOrigin-RevId: 523618b1e1
|
3 years ago |
Megvii Engine Team
|
73b518b718
|
feat(lite): add get physic addr interface in lite
GitOrigin-RevId: e5a9eb1999
|
3 years ago |
Megvii Engine Team
|
f67086adde
|
fix(lite): fix lite global layout transform symvar replace error
GitOrigin-RevId: 7ac74a596a
|
3 years ago |
megvii-mge
|
4462953fba
|
feat(mge/third_party): update cutlass version
|
3 years ago |
Megvii Engine Team
|
d7b0994a3e
|
feat(cuda): add fp16 compute 16 kernel
GitOrigin-RevId: e03435be02
|
3 years ago |
Megvii Engine Team
|
8a2e92bd6c
|
refactor(cuda): depthwish large kernel
GitOrigin-RevId: dade8710b4
|
3 years ago |
Megvii Engine Team
|
6b8a69d5b6
|
feat(cuda): float16 depthwise large kernel conv compute fp32
GitOrigin-RevId: 3050d48f26
|
3 years ago |
Megvii Engine Team
|
bc385b5374
|
feat(cuda): support float16 depthwise large kernel conv
GitOrigin-RevId: fdc1b15fbc
|
3 years ago |
Megvii Engine Team
|
7d2063e35a
|
perf(cuda): speedup conv backward data with small feature map and large filter size
GitOrigin-RevId: 85592bca6b
|
3 years ago |
Megvii Engine Team
|
72403e8929
|
perf(cuda): speedup chanwise conv with small feature map and large filter size
GitOrigin-RevId: e65b2ce856
|
3 years ago |
Megvii Engine Team
|
28d48f2f7a
|
fix(mgb/src): fix megbrain cmake unsupport android_nn
GitOrigin-RevId: 037c197912
|
3 years ago |
Megvii Engine Team
|
ab6d12caff
|
feat(mge): add conv padding mode
GitOrigin-RevId: 147ced856e
|
3 years ago |
Megvii Engine Team
|
177001d5e5
|
refactor(dispatch): allow dynamic type creation
GitOrigin-RevId: 27dde05cff
|
3 years ago |
Megvii Engine Team
|
150a6a6151
|
perf(dispatch/trace): remove unnecessary h2d for constant
GitOrigin-RevId: d00de3fc1f
|
3 years ago |
Megvii Engine Team
|
81d8c73a41
|
perf(dispatch/trace): serval tricks to speed up trace
GitOrigin-RevId: 2bdd70cde2
|
3 years ago |
Megvii Engine Team
|
4fa6162027
|
perf(dispatch): improve performance of dispatch system
GitOrigin-RevId: 860028e1af
|
3 years ago |
Megvii Engine Team
|
ca00177719
|
perf(dispatch): speed up dispatch system
GitOrigin-RevId: eabbe3e021
|
3 years ago |
Megvii Engine Team
|
187c1dc081
|
fix(jit): copy aux var when shallow copying JITExecutor
GitOrigin-RevId: 3b331e1c17
|
3 years ago |
Megvii Engine Team
|
7bd848ce04
|
fix(subgraph): fix hand-written backward for serval jit-elemwise ops
GitOrigin-RevId: ea3a40d96e
|
3 years ago |
Megvii Engine Team
|
7be7656c9f
|
fix(imperative): explicitly manage global structures
GitOrigin-RevId: 0f910c34b6
|
3 years ago |
Megvii Engine Team
|
62034fb262
|
fix(imperative): make CompNode finalize happens before global object destructor
GitOrigin-RevId: 9a1f507c69
|
3 years ago |
Megvii Engine Team
|
59cbf9583d
|
fix(subgraph): use CompiledOp in cpu to avoid workspace error
GitOrigin-RevId: 104dd982ef
|
3 years ago |
Megvii Engine Team
|
b6ce02a152
|
fix(subgraph): fallback back to cg if jit unsupported
GitOrigin-RevId: 853a00a402
|
3 years ago |
Megvii Engine Team
|
21f5a7fcc0
|
fix(subgraph): fix device recognition and scalar propagate
GitOrigin-RevId: fd2fe8bec9
|
3 years ago |
Megvii Engine Team
|
27346b0b65
|
test(opr): add scalar check for opr_test
GitOrigin-RevId: dcfd7ad5d6
|
3 years ago |
Megvii Engine Team
|
225045236b
|
perf(imperative): improve shape inference
GitOrigin-RevId: 98b4d7e9af
|
3 years ago |