提交 · 977c2071712af36fd8ce1faadc578676e6765ada · MegEngine 天元 / MegEngine

08 10月, 2022 4 次提交
- M
  fix(mgb): fix zero copy error when model end with memory forward opr · 896b0193
  由 Megvii Engine Team 提交于 9月 26, 2022
```
GitOrigin-RevId: 2eba697d85a9970ba8e947cc66cdc8dbbcc32242
```
  896b0193
- M
  feat(mge/opr): add meshgrid opr · d3b2b519
  由 Megvii Engine Team 提交于 9月 09, 2022
```
GitOrigin-RevId: 6f703295be2c33974ba27edd3161744251eea570
```
  d3b2b519
- M
  feat(lite): replace warp when src is discrete input · dc0ab9b6
  由 Megvii Engine Team 提交于 8月 08, 2022
```
GitOrigin-RevId: 2bf7980ac6373b691081ab7be9975ec6fa57f8ae
```
  dc0ab9b6
- M
  feat(mge): implement warp_affine backward · 411253f8
  由 Megvii Engine Team 提交于 8月 11, 2022
```
GitOrigin-RevId: 7d7261cf6404a742c6af6c7a27ea416b9f5883d2
```
  411253f8
02 9月, 2022 1 次提交
- M
  fix(src/tensorrt): trt7 manage all workspace · fb9d73b2
  由 Megvii Engine Team 提交于 9月 01, 2022
```
GitOrigin-RevId: b78c80c8f1b8e6b2080284f945ddb950aeb42475
```
  fb9d73b2
01 9月, 2022 1 次提交
- M
  fix(serialization): when the dump fbsv2 model is used, the middle_tensor becomes optional · dee02289
  由 Megvii Engine Team 提交于 8月 31, 2022
```
GitOrigin-RevId: 3d0bbfd44136c39f50c18cd4a3da5c87fcca50fb
```
  dee02289
18 8月, 2022 1 次提交
- M
  fix(opencl/extern_c_opr): fix cl_mem UAF issue 2/2 · ab34bac4
  由 Megvii Engine Team 提交于 8月 16, 2022
```
GitOrigin-RevId: ef79af3c6a7c1fb776bd351b3dd99db657fb55cb
```
  ab34bac4
16 8月, 2022 1 次提交
- M
  fix(mgb): fix megtee build errors · 116781ba
  由 Megvii Engine Team 提交于 8月 15, 2022
```
GitOrigin-RevId: b351dd39943ef26f25e3e95f89d14f8005c49ab4
```
  116781ba
12 8月, 2022 4 次提交
- M
  fix(mm): fix mm error when use sync · b8c7557b
  由 Megvii Engine Team 提交于 8月 10, 2022
```
GitOrigin-RevId: 63387bda049e51b0a7c65b9a9a0ebd746a713446
```
  b8c7557b
- M
  fix(mgb): fix fbsv2 model format no dump tensor format · 73ad06ba
  由 Megvii Engine Team 提交于 8月 10, 2022
```
GitOrigin-RevId: 29f785b7019bbc5a591d5992020798f6f6a9ae6f
```
  73ad06ba
- M
  perf(serialization): optimized the memory usage when load new format model · 399200b3
  由 Megvii Engine Team 提交于 8月 02, 2022
```
GitOrigin-RevId: 2b7313ebe39a7d4a44a8ab61fa0f3646fd7de566
```
  399200b3
- M
  feat(mgb): warpperspective support multi src input · f31e52d5
  由 Megvii Engine Team 提交于 8月 01, 2022
```
GitOrigin-RevId: 0887656864ea43c7f564882c312d0a8a29a90295
```
  f31e52d5
11 8月, 2022 1 次提交
- M
  fix(lite): fix input invalid bug in lar for fitting mode · fca6c76a
  由 Megvii Engine Team 提交于 8月 05, 2022
```
GitOrigin-RevId: 45d81c9a96dc64c4f41dd87581a5fa2bca76c2b7
```
  fca6c76a
09 8月, 2022 1 次提交
- M
  feat(lite): load_and_run support optimize for inference · cc218550
  由 Megvii Engine Team 提交于 8月 04, 2022
```
GitOrigin-RevId: d9abb8de9eb81c357c3f5110b2467582d80b6c2c
```
  cc218550
08 8月, 2022 1 次提交

fix(opencl/extern_c_opr): fix cl_mem UAF issue when · 9bbe5500

由 Megvii Engine Team 提交于 8月 05, 2022

run model OpenCL + ExternCOprRunner, for example

graph: part_a(OpenCL) --> part_b(ExternCOprRunner) --> part_c(OpenCL)
GitOrigin-RevId: f754b559a28dd1efedb2e6649d01e35f35ee4439

9bbe5500

05 8月, 2022 1 次提交
- M
  fix(opr): fix the compatilibity of elemwise multitype new mode · 1f7bf1ad
  由 Megvii Engine Team 提交于 7月 29, 2022
```
GitOrigin-RevId: ee58271276ee4a31e11aa26c53c466f5c07dd019
```
  1f7bf1ad
04 8月, 2022 2 次提交
- M
  feat(mgb/trt): update tensorRT toolchain to 8 · 198ee068
  由 Megvii Engine Team 提交于 7月 06, 2022
```
GitOrigin-RevId: d7cbb722b84e44ea6f82b03346375b1c5bfa8ea9
```
  198ee068
- M
  fix(test): fix test for brainpp docker env · 626222c6
  由 Megvii Engine Team 提交于 8月 03, 2022
```
GitOrigin-RevId: c4c2cc73d2b9c97f0e86f1a4333ec35b2a89d1f9
```
  626222c6
03 8月, 2022 1 次提交
- M
  feat(gopt): support nchw44 global pooling with fuse_grain · fac67e7c
  由 Megvii Engine Team 提交于 7月 22, 2022
```
GitOrigin-RevId: 4c43a149f8214aae7f48ae72bbd6928c1e18f0f5
```
  fac67e7c
01 8月, 2022 6 次提交
- M
  fix(mgb): change the check method of cuda sm code · 32b31fd5
  由 Megvii Engine Team 提交于 7月 24, 2022
```
GitOrigin-RevId: 23dbc9b5747128813efdb6f7d4f557004a5cbf95
```
  32b31fd5
- M
  Revert "feat(dnn): add elemwise modes" · 5f863682
  由 Megvii Engine Team 提交于 7月 28, 2022
```
This reverts commit cb713ddb24e09b9426eb12c5a9767c308d65fd60.

GitOrigin-RevId: 02adf025e602415be9ea0367e6fd1cb8feca6492
```
  5f863682
- M
  Revert "feat(mgb): add cumprod opr" · d2a1905a
  由 Megvii Engine Team 提交于 7月 28, 2022
```
This reverts commit 3436c3bdaa4c11de35e3e37849a11202318cc99f.

GitOrigin-RevId: 95ab3d1aa7dd316f4cf52f7b45a0206bcf60af54
```
  d2a1905a
- M
  feat(mgb): add cumprod opr · 49e14f87
  由 Megvii Engine Team 提交于 4月 25, 2022
```
GitOrigin-RevId: 3436c3bdaa4c11de35e3e37849a11202318cc99f
```
  49e14f87
- M
  feat(dnn): add elemwise modes · 87aedc29
  由 Megvii Engine Team 提交于 6月 14, 2022
```
GitOrigin-RevId: cb713ddb24e09b9426eb12c5a9767c308d65fd60
```
  87aedc29
- M
  refactor(imperative/ops): extends DnnOprCaller with template · c49d3070
  由 Megvii Engine Team 提交于 7月 01, 2022
```
GitOrigin-RevId: 402cba209a3d190285e09c54b7b658988f3bef3d
```
  c49d3070
14 7月, 2022 1 次提交
- M
  fix(mgb/version): fix nvinfer.h not found · 3fbceb3a
  由 Megvii Engine Team 提交于 7月 13, 2022
```
GitOrigin-RevId: 981dd9a7d2759b4d07a33ac451aacea0f1220070
```
  3fbceb3a
12 7月, 2022 1 次提交
- M
  feat(mge/device): enable to get cuda/cudnn/tensorrt version · d60d028a
  由 Megvii Engine Team 提交于 6月 27, 2022
```
GitOrigin-RevId: 5864c61d10972aa30923cf87e457e9819e8ea52d
```
  d60d028a
08 7月, 2022 1 次提交
- M
  fix(win7): workaround for hang when progress exit on win7+32bit · 6dfd5a4c
  由 Megvii Engine Team 提交于 7月 04, 2022
```
GitOrigin-RevId: b49b0b230e6367eea1bc5b0c061915352409fcb3
```
  6dfd5a4c
05 7月, 2022 2 次提交
- M
  feat(mgb): supports value infer and empty input tensor in ElemwiseMultiType · 51397bfc
  由 Megvii Engine Team 提交于 6月 24, 2022
```
GitOrigin-RevId: 05577a8bc8e214dcd7d7fc138ef952fc881c7a88
```
  51397bfc
- M
  feat(mgb/dnn): add modes that the output type is bool in elemwise · 247e2f59
  由 Megvii Engine Team 提交于 5月 09, 2022
```
GitOrigin-RevId: fd0134fca201e64a66402a65d2748b89f64613ef
```
  247e2f59
30 6月, 2022 2 次提交
- M
  feat(persistentcache): change file persistent cache with append model · cfed86f9
  由 Megvii Engine Team 提交于 6月 27, 2022
```
GitOrigin-RevId: 7a427bdab4198a48190f0ecd7a45f4c0c889be8c
```
  cfed86f9
- M
  fix(imperative): add alloc TensorPtr in imperative · d19fc2c1
  由 Megvii Engine Team 提交于 6月 06, 2022
```
GitOrigin-RevId: 1b438fc436cc752f5437b2bd047006f0e0b1b2e5
```
  d19fc2c1
29 6月, 2022 1 次提交
- M
  feat(mge/distributed): add support for batch send recv op · ac51f780
  由 Megvii Engine Team 提交于 6月 01, 2022
```
GitOrigin-RevId: eb3d712704f7a1d0abc6c611cec7c93ad3f5e8bf
```
  ac51f780
28 6月, 2022 1 次提交
- M
  fix(ci): relax async timeout · 35cf0422
  由 Megvii Engine Team 提交于 6月 24, 2022
```
GitOrigin-RevId: 551de481f0719249ab5826fc9dbb5f4eae76d759
```
  35cf0422
24 6月, 2022 3 次提交
- M
  fix(mgb): fix profile skip condition · 5a355138
  由 Megvii Engine Team 提交于 6月 14, 2022
```
GitOrigin-RevId: f196eabc9810b5ac32744a10a1473a6b445f7f4a
```
  5a355138
- M
  fix(mgb/fastrun): fix megbrain fastrun memory overflow bug · 5bdc430e
  由 Megvii Engine Team 提交于 5月 24, 2022
```
GitOrigin-RevId: f56aa5a5059ef6c5ea9d51fe31d89a0d778aa702
```
  5bdc430e
- M
  fix(mgb/atlas): fix issue with profiling on Atlas · a4f019e7
  由 Megvii Engine Team 提交于 6月 15, 2022
```
GitOrigin-RevId: fbae81940de3a33b329ddb7f2bc9514e9af2fd66
```
  a4f019e7
22 6月, 2022 2 次提交
- M
  fix(mge): fix grad of maximum(x, x) · cfc41648
  由 Megvii Engine Team 提交于 6月 20, 2022
```
GitOrigin-RevId: e0e2efb71bbe507bd5b4dab539b5b9cfe79d1187
```
  cfc41648
- M
  fix(cmake): fix compilation options for benchmark in cmake · ed92b9c1
  由 Megvii Engine Team 提交于 6月 10, 2022
```
GitOrigin-RevId: d763fd32622b0ee9267ae04c637ff83639d90bb1
```
  ed92b9c1
20 6月, 2022 1 次提交
- M
  feat(mgb): upgrade flatbuffer · f0088335
  由 Megvii Engine Team 提交于 6月 15, 2022
```
GitOrigin-RevId: 7b1a04934ea3a7936661bb0558e4c58b258cdff2
```
  f0088335

MegEngine 天元 / MegEngine 1 年多 前同步成功

MegEngine 天元 / MegEngine
1 年多前同步成功