Megvii Engine Team
b6c0e8d093
fix(mgb/gopt): update profiler cache of global layout transform test
GitOrigin-RevId: fe8a712599
3 years ago
Megvii Engine Team
f2db7b0ded
feat(mgb/gopt): global layout transform support cuda fp16
GitOrigin-RevId: 1449c54215
3 years ago
Megvii Engine Team
ca7cec7a5d
fix(mgb/gopt): minor fixes for global layout transform
Merge Target::ARM and Target::X86 into Target::CPU to make global layout transform easier to use
GitOrigin-RevId: cc9363fa38
3 years ago
Megvii Engine Team
93152dfa14
fix(mgb/gopt): fix global layout transform
deduplicate the states of the DP problem due to different layout config(NCHW44 & NCHW44_HYBRID) will produce tensors with same layout
GitOrigin-RevId: 7f77efd21b
3 years ago
Megvii Engine Team
fe93013a6e
feat(mgb/gopt): global layout transform support nchw_nchwxx hybrid mode
GitOrigin-RevId: 6d5b55d7fc
3 years ago
Megvii Engine Team
3d45d35241
feat(mgb/gopt): profiler support checking algo availability
GitOrigin-RevId: 39cad612cc
3 years ago
Megvii Engine Team
5f15f75984
test(mgb/gopt): add a testcase for SubGraphExtractor with multiple outputs
GitOrigin-RevId: 7785bdc8c0
3 years ago
Megvii Engine Team
a6230ba95a
feat(mgb/gopt): global layout transform support arm
GitOrigin-RevId: db50b33c11
3 years ago
Megvii Engine Team
0be6ca8899
fix(src/core): fix record change ptr bug on comp node copy
GitOrigin-RevId: 0f68966211
3 years ago
Megvii Engine Team
84baf3df1b
feat(mgb): add tensorrt plugin support
GitOrigin-RevId: 5428b4f665
3 years ago
Megvii Engine Team
b59e8ccf24
fix(mgb): fix cambricon bangc copybara
GitOrigin-RevId: 22b57364e8
3 years ago
Megvii Engine Team
fe6299c52d
Revert "feat(mgb): allow output tensor's ptr change when record"
This reverts commit c610c8bf9a
.
GitOrigin-RevId: 2eb0504c5d
3 years ago
Megvii Engine Team
d90cb7763c
feat(src/core): record support change ptr basic
GitOrigin-RevId: b12f1c4a66
3 years ago
Megvii Engine Team
68cdabd288
feat(opr): indexing_multi_axis_vec support nd index
GitOrigin-RevId: 07b1248bdc
3 years ago
Megvii Engine Team
0a56a5b110
feat(cuda/comp_node): gets (maximum) GPU memory allocated/reserved
GitOrigin-RevId: da2cc22436
3 years ago
Megvii Engine Team
13c7c572df
fix(mgb): fix shape infer's condition in lite
GitOrigin-RevId: 550eaff4cd
3 years ago
Megvii Engine Team
0ddabb06cc
fix(imperative/astype): fix astype when target dtype is int4
GitOrigin-RevId: 1c5195176d
3 years ago
Megvii Engine Team
34773ba37b
fix(mgb/gopt): tensorcore pass replace BatchConvBias inputs to nchw4
GitOrigin-RevId: 3ff3c422fb
3 years ago
Megvii Engine Team
481007812e
fix(mgb): support rc version
GitOrigin-RevId: aa5fe5198d
3 years ago
Megvii Engine Team
f00741dad5
feat(build): bazel build support flatbuffer serialization
GitOrigin-RevId: 4755c27e06
3 years ago
Megvii Engine Team
8b7649345a
feat(lite): lite support output var no copy option
GitOrigin-RevId: 5b9488cb93
3 years ago
Megvii Engine Team
7642f66df7
feat(core): support output var no copy
GitOrigin-RevId: fd68a1735b
3 years ago
Megvii Engine Team
7fa5f6f4e2
fix(mgb): correct version of fbs serialization
GitOrigin-RevId: 2a71d9afc2
3 years ago
Megvii Engine Team
25ec2530ba
feat(whl/api/lar): enable megengine dll on Windows
1: reduce python whl package size
2: unify api link logic on all OS
3: add option: MGE_WINDOWS_BUILD_WITH_STATIC_CRT
--- default OFF
--- if build CRT(vc runtime) with STATIC with megengine.dll
some CRT api will crash, for example, flush, so if you
build with static megengine, and do not want to install CRT
you can set MGE_WINDOWS_BUILD_WITH_STATIC_CRT TRUE
--- how to install CRT:
https://docs.microsoft.com/en-us/cpp/windows/latest-supported-vc-redist?view=msvc-160
install VC_redist.x64.exe
4: rename megengine_export to megengine_shared(only export needed symbols ),
caused by runtime symbols conflict with pytorch
GitOrigin-RevId: 93d8d80f29
3 years ago
Megvii Engine Team
c17d603847
feat(mgb): allow output tensor's ptr change when record
GitOrigin-RevId: c610c8bf9a
3 years ago
Megvii Engine Team
d9a9d9d49e
fix(imperative/fastrun): set workspace limit for imperative rt
GitOrigin-RevId: 474dc691a3
3 years ago
Megvii Engine Team
a09a2b730d
fix(mgb/opr): fix fastrun workspace limit for imperative rt
GitOrigin-RevId: bd69a82d4c
3 years ago
Megvii Engine Team
5207154e5c
Revert "fix(mgb): fix fastrun no_profiling_on_shape_change"
This reverts commit 2735536424
.
GitOrigin-RevId: 6a641808f0
3 years ago
Megvii Engine Team
bd42cb06cf
refactor(mgb/lite): refactor lite InfilePersistentCache with core impl
GitOrigin-RevId: 64b7825c34
3 years ago
Megvii Engine Team
d898838e18
fix(src/gopt/test): remove device info in the cache_data
GitOrigin-RevId: 8e9758971d
3 years ago
Megvii Engine Team
a4ac5e7e8f
fix(mgb/gopt): fix profiler cache when build without opr type info name
GitOrigin-RevId: 0eba678f2b
3 years ago
Megvii Engine Team
012de7695d
feat(mgb/gopt): add profiler cache
In order to improve performance of the profiling procedure. Make layout transform testcase stable. The profiling result in ci environment will be cached in files.
GitOrigin-RevId: ba2743f35f
3 years ago
Megvii Engine Team
b33ec46e54
fix(mgb/gopt): fix layout gtrans when graph partition has opr with different format
GitOrigin-RevId: 326fdebb0c
3 years ago
Megvii Engine Team
5885b137fa
feat(dnn/arm): support layout like NHWC channel like broadcast on arm
GitOrigin-RevId: fb4300004c
3 years ago
Megvii Engine Team
565466c25f
feat(lite): auto deduce output tensor shape before model forward
GitOrigin-RevId: 78e00dab5d
3 years ago
Megvii Engine Team
a29f1c8c02
fix(cmake/midout): fix cmake midout
GitOrigin-RevId: 0fb93deb19
3 years ago
Megvii Engine Team
369c2ccc5a
style(all): reformat c++ code
GitOrigin-RevId: 3ffd1b211f
3 years ago
Megvii Engine Team
bfb30dcb81
chore(format): fix compile bugs after code format
GitOrigin-RevId: 11a4b06f6f
3 years ago
Megvii Engine Team
2108696588
feat(mge/imperative): add local response normalization
GitOrigin-RevId: 939a4d26dd
3 years ago
Megvii Engine Team
dc42417f33
fix(ci/build): opt macos ci and move bazel to 4.2.1
GitOrigin-RevId: d1494faf0c
3 years ago
Megvii Engine Team
f5cb21ed3a
fix(mgb/opr): add non finite check
GitOrigin-RevId: a9fcd0a350
3 years ago
Megvii Engine Team
8cb201868e
fix(mgb): fix fastrun cache serialization method
GitOrigin-RevId: cd8bed3cc9
3 years ago
Megvii Engine Team
88c1eedbd7
feat(mgb/gopt): enable reduce for nchw44
GitOrigin-RevId: fce59d0762
3 years ago
Megvii Engine Team
fca195351c
feat(gopt): add nhwc fuse conv typecvt optpass
GitOrigin-RevId: adc2301203
3 years ago
Megvii Engine Team
dc28a43409
fix(mgb/bn): fix empty tensor input problem and other minor problems
GitOrigin-RevId: 1708b76cb8
3 years ago
Megvii Engine Team
270f1aa28f
feat(mgb/serialization): add Accessor for OprLoader to fix BN output compatibility
GitOrigin-RevId: 3b95da02c8
3 years ago
Megvii Engine Team
c0ccd0ea7e
feat(mge/bn): add NHWC support for bn
GitOrigin-RevId: 0a5bb6f72d
3 years ago
Megvii Engine Team
b982be567a
feat(mge/imperative): add permutation support for the tensor
GitOrigin-RevId: 7ed0447bfe
3 years ago
Megvii Engine Team
3977b7aa0b
feat(mgb/shuffle): add shuffle opr
GitOrigin-RevId: 80490a6f84
3 years ago
Megvii Engine Team
d7cc4628f6
perf(gopt): opt concat for OpenCL
GitOrigin-RevId: 9bb226d4b1
3 years ago