GithubStar/ncnn
1
0
Fork 0
mirror of https://github.com/Tencent/ncnn.git synced 2026-10-29 04:17:40 +08:00
Code Issues Packages Projects Releases Wiki Activity
2,477 Commits 4 Branches 50 Tags
c09d7b359116c02891ef0bb678e596d67cb0d363
Commit Graph
2477 Commits
This Branch
This Branch
All Branches
Author SHA1 Message Date
nihui c09d7b3591 mips msa optimization for convolution int8 (#3675) 
* basic mips msa optimization for convolution int8

* mips msa optimization for convolution int8 gemm

* mips msa optimization for convolution int8 winograd pack8to4/pack8to1

* mention msa maddv/msubv intrinsics bug
 2022-04-05 12:25:25 +08:00
dependabot[bot] 1a0bf1f517 Bump pypa/cibuildwheel from 2.3.1 to 2.4.0 (#3674) 
Bumps [pypa/cibuildwheel](https://github.com/pypa/cibuildwheel) from 2.3.1 to 2.4.0.
- [Release notes](https://github.com/pypa/cibuildwheel/releases)
- [Changelog](https://github.com/pypa/cibuildwheel/blob/main/docs/changelog.md)
- [Commits](https://github.com/pypa/cibuildwheel/compare/v2.3.1...v2.4.0)

---
updated-dependencies:
- dependency-name: pypa/cibuildwheel
  dependency-type: direct:production
  update-type: version-update:semver-minor
...

Signed-off-by: dependabot[bot] <support@github.com>

Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
 2022-04-04 12:10:11 +08:00
nihui 72c467d1d9 mips msa optimization for quantize dequantize requantize (#3672) 2022-04-03 21:37:27 +08:00
WuJinxuanandEdVince e984f9f40d add multiheadattention arm (#3667) 
* add:multiheadattention_arm

* pass the test in local

* add omp

* return naive

Co-authored-by: EdVince <EdVince@users.noreply.github.com>
 2022-04-03 19:05:31 +08:00
nihui 1fcad0e765 loongson mmi optional layer 2022-04-03 09:57:44 +08:00
nihui 0d83ad99f8 link pnnx with pthread, fix minmax issue on windows build 2022-04-02 09:43:13 +08:00
nihui 559e5b23f9 vulkan tensorcore optimization (#3628) 
* query and enable cooperative matrix

* fix build with old vulkan sdk

* implement cooperative matrix optimization

* add nvidia-t4 coverage

* adjust test option for more coverage
 2022-03-30 19:05:17 +08:00
nihui 4302f78f55 less specialization constant for vulkan conv1x1s1d1 shaders (#3657) 2022-03-27 17:47:31 +08:00
Guo Haria 67f52ba73c Update yolov5.py (#3656) 
fix rect position bug.
 2022-03-27 15:19:38 +08:00
Shangxin 94beeaf000 fix word case (#3655) 2022-03-25 22:40:45 +08:00
nihui b934fd53e7 fix vs2019 packaging 2022-03-25 11:52:26 +08:00
tpoisonooo aab476f5b6 fix build warning (#3651) 2022-03-24 13:47:09 +08:00
nihui 9c92df814f better condition for mixing vulkan winograd f23 and f43 2022-03-24 10:35:47 +08:00
nihui 944829838b vulkan conv1x1s1d1 for any packing (#3646) 2022-03-23 14:25:41 +08:00
nihui 62a872bad3 vulkan winograd for any packing (#3645) 2022-03-23 11:23:07 +08:00
dependabot[bot] aac71352fd Bump actions/cache from 2.1.7 to 3 (#3643) 
Bumps [actions/cache](https://github.com/actions/cache) from 2.1.7 to 3.
- [Release notes](https://github.com/actions/cache/releases)
- [Commits](https://github.com/actions/cache/compare/v2.1.7...v3)

---
updated-dependencies:
- dependency-name: actions/cache
  dependency-type: direct:production
  update-type: version-update:semver-major
...

Signed-off-by: dependabot[bot] <support@github.com>

Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
 2022-03-22 11:24:08 +08:00
nihui 9e3cc1c5df two stage vulkan innerproduct (#3642) 2022-03-21 20:59:57 +08:00
nihui 677d54d496 fuse vulkan winograd pad and crop (#3640) 2022-03-20 13:00:29 +08:00
nihui 002c07d4ec mix vulkan winograd f23 and f43 (#3639) 
* mix vulkan winograd f23 and f43

* larget epsilon for winograd optimization test
 2022-03-19 23:31:46 +08:00
nihui d42e048b56 pnnx convert torch.addmm (#3634) 2022-03-18 12:24:05 +08:00
nihui 3ddd65e18c massive vulkan optimization part3 (#3632) 
* implicit gemm

* unroll direct conv by 2x2x2
 2022-03-17 18:39:28 +08:00
nihui cfcb1cffa9 massive vulkan optimization part2 (#3621) 
* vulkan local memory optimization for conv1x1 pack4 and winograd on dgpu

* unified innerproduct pipeline creation

* reorder deconvolution weight layout

* flexible local memory data type

* more local memory optimization for conv/deconv gemm
 2022-03-16 17:14:45 +08:00
tpoisonoooMegEnginetpoisonooo
 		edd3a78ffe style(src/layer): remove unused para and var (#3623) 
Co-authored-by: MegEngine <megengine@megvii.com>
Co-authored-by: tpoisonooo <tpoisonooo@users.noreply.github.com>
 2022-03-16 09:33:35 +08:00
tpoisonoooandMegEngine 6e12647985 docs(how-to-build.md): update jetson description (#3622) 
Co-authored-by: MegEngine <megengine@megvii.com>
 2022-03-15 12:27:59 +08:00
tpoisonooo b8f36c258e Update how-to-build.md (#3619) 2022-03-14 19:16:28 +08:00
nihui f9663d7726 pnnx support torch 1.11.0 (#3617) 
* adapt torch 1.11.0 api changes

* find python library for torchvision linking
 2022-03-13 11:49:59 +08:00
nihui 6e19ab26ba massive vulkan optimization (#3602) 
* vulkan deconvolution sgemm col2im

* vulkan convolution winograd43

* improve fp16s numeric stablity

* vulkan convolution im2col sgemm

* check squeezenet top2, as top3 vs top4 score too close..
 2022-03-12 10:16:58 +08:00
nihui 8f25ba0cab enable fp16a on mali-g31 2022-03-09 23:11:02 +08:00
Kenji Mouri 8e29c42080 Improve SSE2 implementations in x86 targets. (#3605) 
* Make some typos for SSE2 floor.

* Improve the implementation of SSE2 abs.

* Improve the implementation of SSE2 ceil.
 2022-03-09 15:01:03 +08:00
Kenji MouriandMouriNaruto 2b4a2125e6 Add SSE2 implementation of floor and ceil in x86 targets. (#3595) 
* Add SSE2 implementation of floor and ceil in x86 targets.

* apply code-format changes

* Update the SSE2 floor implementation.

Co-authored-by: MouriNaruto <MouriNaruto@users.noreply.github.com>
 2022-03-09 10:59:42 +08:00
MouriNaruto f4f4cfd784 apply code-format changes 2022-03-05 14:37:58 +00:00
nihui 8af8e52cb0 add nanodetplus pnnx example 2022-03-05 22:31:43 +08:00
Kenji Mouri 3ba5d9765f Add arm and arm64 targets support for MSVC. (#3592) 2022-03-04 14:17:43 +08:00
Kenji Mouri e4f6b118a2 Fix comment typo because it should be itm[6][6]. (#3591) 2022-03-03 22:36:37 +08:00
nihuini b053a8c6d5 fix unlocked pool allocator destroyed too early issue in gpu convdw and deconvdw inference 2022-03-03 18:48:40 +08:00
nihui 30e106b185 add another mali g52 device id 2022-03-02 22:40:43 +08:00
nihui 41cd40ac1f convert pnnx torch.unbind and torch.ones/ones_like family (#3583) 
* convert torch.unbind

* torch.ones torch.ones_like

* torch.full torch.full_like

* torch.randn_like

* torch.empty torch.empty_like
 2022-03-02 14:41:31 +08:00
dependabot[bot] 8b68ec050d Bump actions/checkout from 2 to 3 (#3588) 
Bumps [actions/checkout](https://github.com/actions/checkout) from 2 to 3.
- [Release notes](https://github.com/actions/checkout/releases)
- [Changelog](https://github.com/actions/checkout/blob/main/CHANGELOG.md)
- [Commits](https://github.com/actions/checkout/compare/v2...v3)

---
updated-dependencies:
- dependency-name: actions/checkout
  dependency-type: direct:production
  update-type: version-update:semver-major
...

Signed-off-by: dependabot[bot] <support@github.com>

Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
 2022-03-02 14:22:26 +08:00
nihui 2880eff264 deconv1d deconv3d (#3584) 
* fix sigmoid returns nan with very large input
 2022-03-01 21:50:17 +08:00
dependabot[bot] 556aeb675b Bump actions/setup-python from 2 to 3 (#3586) 
Bumps [actions/setup-python](https://github.com/actions/setup-python) from 2 to 3.
- [Release notes](https://github.com/actions/setup-python/releases)
- [Commits](https://github.com/actions/setup-python/compare/v2...v3)

---
updated-dependencies:
- dependency-name: actions/setup-python
  dependency-type: direct:production
  update-type: version-update:semver-major
...

Signed-off-by: dependabot[bot] <support@github.com>

Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
 2022-03-01 12:24:12 +08:00
YohYohYoh-Z
 		f5b90a5333 Add unaryop x86 (#3579) 
Co-authored-by: Yoh <wangpeizhou@bilibili.com>
Co-authored-by: Yoh-Z <Yoh-Z@users.noreply.github.com>
 2022-02-25 14:11:08 +08:00
nihuini bc188ece58 update modelwriter for new operators 2022-02-25 11:27:01 +08:00
nihui 457e066eb5 x86 f16c infrastructure (#3577) 2022-02-24 18:29:18 +08:00
nihui 38ae671391 add yolov5 pnnx example 2022-02-23 16:00:20 +08:00
YohandYoh-Z 4b68e3f9c1 Opt avxmath (#3563) 
* optimize x86 avx exp log sincos fma

* optimize avx&sse_mathfun exp,log,cos,sin,sincos and add fnmadd to x86_usability

Co-authored-by: Yoh-Z <Yoh-Z@users.noreply.github.com>
 2022-02-16 22:50:34 +08:00
nihui 920aa79f04 drop x86 avx2 fp16 (#3568) 2022-02-16 22:45:57 +08:00
nihui 6b2495cc24 add reshape before and after pooling 123d with no batch dimension (#3566) 20220216 2022-02-15 21:40:35 +08:00
nihui 76e32e9ee6 fix interp nearest by scale factor, fix issue #3555 (#3565) 
* lets accept the float div vs recp error
 2022-02-15 16:58:12 +08:00
nihuini 6ee5ab72f6 handle reshape 5d with batch index 0 2022-02-15 11:07:42 +08:00
tpoisonoooMegEnginetpoisonooo
 		cfacba273f improvement(binaryop): use MAKE_FUNCTION macro (#3559) 
Co-authored-by: MegEngine <megengine@megvii.com>
Co-authored-by: tpoisonooo <tpoisonooo@users.noreply.github.com>
 2022-02-15 10:23:45 +08:00