nihui
/
ncnn
Not watched
1
0
Fork 0
Code
Releases 46 Wiki evaluate Activity
Issues 0 Pull Requests 0 Datasets Model Cloudbrain HPC
651 Commits
5 Branches
354 MB
2 Download
Tree: 1f4bdd91b5
Branches Tags
${ item.name }
Create branch ${ searchTerm }
from '1f4bdd91b5'
${ noResults }
Commit Graph

405 Commits (1f4bdd91b52dc49fb71cec4b4e35fc62ca51a61c)

Author SHA1 Message Date
  nihuini 1f4bdd91b5 uint32_t typed workgroup size 7 years ago
  nihuini 2e939fab0f fix memleak 7 years ago
  nihuini 532054b453 expose more device info 7 years ago
  nihui dd83284cee prelu shader 7 years ago
  nihui e50b339f04 clip shader 7 years ago
  nihui 69788b0467 reshape shader family 7 years ago
  nihui c41bcd98a3 priorbox shader, fix permute order 1 on image, fix potential staging memory leak 7 years ago
  nihuini 0956d26df1 add absval sigmoid tanh shader 7 years ago
  nihuini b5faa0e519 respect pad param in deconv vulkan 7 years ago
  BUG1989 8e337d440e fix the bug with convdw7x7 op working on int8 mode (#818) 7 years ago
  nihuini 4bc543d85c crop shader 7 years ago
  nihuini 9787625e4b warn users about the old wrong softmax behavior on axis not zero 7 years ago
  nihuini c54e57ed6f Merge branch 'master' of https://github.com/Tencent/ncnn 7 years ago
  nihuini 85a28959e4 fix binaryop shader binding, use shared buffer state, fix blob copy in non-light mode, fix #817 7 years ago
  BUG1989 8ff831f7cd fix the segmentation fault when load int8 model (#811) 7 years ago
  nihuini ff0e8c85c5 bind the same pipeline may cause driver incorrectly optimize into one, use two pipelines to always change the current one 7 years ago
  BUG1989 df3d224484 new int8 implement,better accuracy (#749) 7 years ago
  nihuini 4ac56d3c1c unified memory index is not mandatory, sanity check 7 years ago
  nihuini 5f0ee22a33 treat as unified memory architecture if memory heap is same 7 years ago
  nihui d85775fbcd fix softmax axis order on 3-dim, fix caffe reshape conversion, regenerate ssd param 7 years ago
  nihui 979ed57487 packing param for identity packing when padding disabled, auto packing conversion between cpu and gpu blob 7 years ago
  nihui b49cb56ad9 constify vulkan device handle, use default local vulkan device if not specified 7 years ago
  nihui 5e07749a4a do not emit upload transfer on unified memory 7 years ago
  nihui 9ebac3fe9e dedicated reference counter for staging data 7 years ago
  nihui 68afd1fa17 reset fence 7 years ago
  nihui 81ee56b209 copy buffer has offset alignment limit, re-implement concat as compute pipeline 7 years ago
  nihuini 83efa73cf6 fallback to cpu forward if layer not support vulkan, automatically! 7 years ago
  nihuini bdd305638d command reset 7 years ago
  nihuini 10a088397e concat interleave image row 7 years ago
  nihuini 1ace8068e3 zero detected is not error 7 years ago
  nihuini 14efdd8e00 reorg shader 7 years ago
  nihui b62e9c4b1e shufflechannel shader 7 years ago
  nihuini bb04055e80 permute shader 7 years ago
  nihui 24f423b0c6 fix build on msvc 7 years ago
  nihui cc4376d8e6 do not upload unnecessary pack1 weight, reduce gpu memory usage 7 years ago
  nihui 0ad0c07526 drop duplicated weight data in convolution-fc, use the more light-weight pipelines 7 years ago
  nihuini 43c4b57201 group deconvolution packing family 7 years ago
  nihuini 8547864b6f group convolution packing family 7 years ago
  nihuini 675fcc72a5 interp vulkan 7 years ago
  nihuini 37413ea95c implement depthwise deconvolution vulkan, fix top blob state 7 years ago
  nihuini 468516879f implement deconvolution vulkan family support 7 years ago
  nihuini e213605cd4 reduce memory usage of weight packing 7 years ago
  nihuini 7312887671 transfer command hold data context 7 years ago
  nihuini 4a57f88c3c vkcompute auto begin end, use proper alignment for vktransfer staging buffer offset 7 years ago
  nihuini 39f2c71d5b fix name conflict on ios 7 years ago
  nihui f4e12101c0 fix convolution typed innerproduct pack4 7 years ago
  nihui 0acdbebf3b merge refcount into buffer memory cookie 7 years ago
  nihui 960ffa1a50 optimize workgroup size for convolution depthwise and innerproduct pack4 7 years ago
  nihui a15b389d86 fix innerproduct pack1to4 pack4to1 weight upload 7 years ago
  Emmanuel Benazera a8fd79e1bc fixed cell initialization in LSTM layer 7 years ago