关于GCU、沐曦GPGPU、MLU、0卡V100资源4月7日恢复上架的公告>>> 关于共建具身智能开源数据集的倡议>>> 关于云脑任务中统一路径访问方式的公告>>> 关于将启智集群GPU资源迁移至智算集群的公告>>>
OpenI
/
MegEngine
mirror of https://github.com/MegEngine/MegEngine.git
Not watched
23
13
Fork 8
Code
Releases 26 Wiki Activity
Issues 0
You can not select more than 25 topics Topics must start with a chinese character,a letter or number, can include dashes ('-') and can be up to 35 characters long.
3244 Commits
40 Branches
26 MiB
1054 Download
 
 
 
 
 
 
Branch: master
Branches Tags
${ item.name }
Create branch ${ searchTerm }
from 'master'
${ noResults }
MegEngine/dnn/src/cuda
Megvii Engine Team dc24f9c7b5 fix(dnn): fix little cuda compile error 2 months ago
..
adaptive_pooling chore(license): remove all license in file header 2 years ago
add_update chore(license): remove all license in file header 2 years ago
argmxx chore(license): remove all license in file header 2 years ago
argsort chore(license): remove all license in file header 2 years ago
batch_conv_bias chore(license): remove all license in file header 2 years ago
batch_normalization chore(license): remove all license in file header 2 years ago
batched_matrix_mul feat(dnn/cuda): add naive bmm 1 year ago
check_non_finite refactor(imperative/ops): extends DnnOprCaller with template 1 year ago
checksum chore(license): remove all license in file header 2 years ago
concat chore(license): remove all license in file header 2 years ago
cond_take refactor(imperative/ops): extends DnnOprCaller with template 1 year ago
conv_bias fix(dnn/cuda): fix internal error in channel_wise conv with cuda118 5 months ago
convolution chore(license): remove all license in file header 2 years ago
convolution3d chore(license): remove all license in file header 2 years ago
convolution_helper chore(license): remove all license in file header 2 years ago
convpooling chore(license): remove all license in file header 2 years ago
correlation chore(license): remove all license in file header 2 years ago
cross feat(dnn,megbrain,imperative): add cross opr 9 months ago
cub ci(check): add clang-format in check stage 2 years ago
cumsum chore(license): remove all license in file header 2 years ago
cutlass feat(dnn): add rrconv wgrad, support int32 and uint8 region mask 1 year ago
cv chore(license): remove all license in file header 2 years ago
cvt_color chore(license): remove all license in file header 2 years ago
dct chore(license): remove all license in file header 2 years ago
deformable_conv chore(license): remove all license in file header 2 years ago
deformable_ps_roi_pooling chore(license): remove all license in file header 2 years ago
diag chore(license): remove all license in file header 2 years ago
dot chore(license): remove all license in file header 2 years ago
dropout feat(opr): add multiattention cuda backend 1 year ago
elemwise feat(dnn/elemwise): support elemwise uint16 compute 8 months ago
elemwise_multi_type feat(dnn): add elemwise modes 1 year ago
eye chore(license): remove all license in file header 2 years ago
fake_quant chore(license): remove all license in file header 2 years ago
fill feat(imperative): speed up fill 1 year ago
flip chore(license): remove all license in file header 2 years ago
gaussian_blur chore(license): remove all license in file header 2 years ago
general_norm feat(opr): optimize general_norm cuda forward and backward implementation 1 year ago
group_local chore(license): remove all license in file header 2 years ago
group_norm feat(dnn,src,imperative): add instancenorm 1 year ago
images2neibs chore(license): remove all license in file header 2 years ago
indexing_multi_axis_vec fix(mgb): make error infomation of advanced indexing out of bound more readable 1 year ago
indexing_one_hot chore(license): remove all license in file header 2 years ago
kernel_common feat(cuda): upgrade x86_64-cuda to 11.8 1 year ago
lamb feat(mge/dnn): add lamb optimizer 1 year ago
layer_norm chore(license): remove all license in file header 2 years ago
linspace chore(license): remove all license in file header 2 years ago
local chore(license): remove all license in file header 2 years ago
local_share chore(license): remove all license in file header 2 years ago
lrn chore(license): remove all license in file header 2 years ago
lsq chore(license): remove all license in file header 2 years ago
mask_conv chore(license): remove all license in file header 2 years ago
masked_fill feat(opr): add masked_fill op 1 year ago
matrix_inverse chore(license): remove all license in file header 2 years ago
matrix_mul fix(dnn/cuda): fix too many resources requested error of matmul with cuda118 5 months ago
max_tensor_diff chore(license): remove all license in file header 2 years ago
megcore chore(license): remove all license in file header 2 years ago
mesh_indexing chore(license): remove all license in file header 2 years ago
multi_head_attn feat(opr): add multiattention naive proxy backend 11 months ago
non_zero fix(ci): fix the missing of override keyword in non_zero kernel 8 months ago
norm feat(dnn/naive/norm,-dnn/cuda/norm,-dnn/test/norm): add norm dnn opr, 1 year ago
padding perf(dnn): speed up pad kernel 1 year ago
param_pack refactor(imperative/ops): extends DnnOprCaller with template 1 year ago
pooling chore(license): remove all license in file header 2 years ago
powc chore(license): remove all license in file header 2 years ago
ptx/uint4_int4 fix(dnn/cuda): fix ptx mma algo compute bugs 1 year ago
reduce chore(license): remove all license in file header 2 years ago
reduce_helper feat(opr): optimize general_norm cuda forward and backward implementation 1 year ago
region_restricted_convolution feat(dnn,imperative): remove the restriction of tensor shape when using uint8 region mask 1 year ago
relayout chore(license): remove all license in file header 2 years ago
relayout_format chore(license): remove all license in file header 2 years ago
remap chore(license): remove all license in file header 2 years ago
repeat chore(license): remove all license in file header 2 years ago
resize feat(opr): add interpolate trilinear 10 months ago
rng feat(dnn,megbrain,imperative): add multinomial opr 6 months ago
roi_align chore(license): remove all license in file header 2 years ago
roi_copy chore(license): remove all license in file header 2 years ago
roi_pooling chore(license): remove all license in file header 2 years ago
rotate chore(license): remove all license in file header 2 years ago
separable_conv chore(license): remove all license in file header 2 years ago
separable_filter chore(license): remove all license in file header 2 years ago
sleep chore(license): remove all license in file header 2 years ago
sliding_window_transpose chore(license): remove all license in file header 2 years ago
softmax chore(license): remove all license in file header 2 years ago
split chore(license): remove all license in file header 2 years ago
svd chore(license): remove all license in file header 2 years ago
tensor_remap chore(license): remove all license in file header 2 years ago
tile chore(license): remove all license in file header 2 years ago
topk fix(dnn/cuda): fix accuracy error of topk fp16 with cuda118 5 months ago
tqt chore(license): remove all license in file header 2 years ago
transpose chore(license): remove all license in file header 2 years ago
type_cvt chore(license): remove all license in file header 2 years ago
warp_affine chore(license): remove all license in file header 2 years ago
warp_perspective fix(dnn): correctly using MEGDNN_DISABLE_FLOAT16 directives 1 year ago
where feat(mge): add kernel for where operator 8 months ago
atomic_add.cuh chore(license): remove all license in file header 2 years ago
cuda_shfl_compat.cuh chore(license): remove all license in file header 2 years ago
cudnn_with_check.h chore(license): remove all license in file header 2 years ago
cudnn_wrapper.cpp fix(dnn/cuda): fix conv_bwd_data/conv_fwd_implicit_precomp_gemm which is shake with cudnn860 5 months ago
cudnn_wrapper.h feat(opr): add multiattention cuda backend 1 year ago
cudnn_wrapper_v8.cpp fix(dnn/cuda): close algos from cudnn-v8 temporarily, need fix 6 months ago
cudnn_wrapper_v8.h fix(dnn/cuda): close algos from cudnn-v8 temporarily, need fix 6 months ago
elemwise_helper.cpp chore(license): remove all license in file header 2 years ago
elemwise_helper.cuh fix(dnn/cuda): fix misaligned address after index bool tensor 3 months ago
elemwise_helper_q4.cuh chore(license): remove all license in file header 2 years ago
error_info.cuh chore(license): remove all license in file header 2 years ago
fp16_help.cuh chore(license): remove all license in file header 2 years ago
handle.cpp fix(dnn): fix little cuda compile error 1 month ago
handle.h chore(license): remove all license in file header 2 years ago
handle_create.cpp feat(dnn,megbrain,imperative): add multinomial opr 6 months ago
int_fastdiv.cpp chore(license): remove all license in file header 2 years ago
int_fastdiv.cuh chore(license): remove all license in file header 2 years ago
integer_subbyte_utils.cuh chore(license): remove all license in file header 2 years ago
memory_utils.cuh chore(license): remove all license in file header 2 years ago
ptx_loader.cpp feat(cuda): add convbias ptx algo 1 year ago
ptx_loader.h feat(cuda): add convbias ptx algo 1 year ago
query_blocksize.cpp chore(license): remove all license in file header 2 years ago
query_blocksize.cuh chore(license): remove all license in file header 2 years ago
query_blocksize_impl.cu chore(license): remove all license in file header 2 years ago
reduce_helper.cuh chore(license): remove all license in file header 2 years ago
reduce_helper.cuinl chore(license): remove all license in file header 2 years ago
transpose_utils.cuh chore(license): remove all license in file header 2 years ago
utils.cpp chore(license): remove all license in file header 2 years ago
utils.cuh chore(license): remove all license in file header 2 years ago
utils.h chore(license): remove all license in file header 2 years ago
Delete task

Are you sure you want to delete this task? Once this task is deleted, it cannot be recovered.

cancel
confirm

天元（MegEngine）是旷视自主研发的开源深度学习框架，于2020年3月正式在 OpenI 启智社区开源，天元能够帮助开发者高效的完成深度学习算法的设计、训练、部署，有效提升AI研发工作效率。

https://megengine.org.cn/

C++ Cuda Python C Starlark other

Please read the following content carefully:

Dear OpenI User

Thank you for your continuous support to the Openl Qizhi Community AI Collaboration Platform. In order to protect your usage rights and ensure network security, we updated the Openl Qizhi Community AI Collaboration Platform Usage Agreement in January 2024. The updated agreement specifies that users are prohibited from using intranet penetration tools. After you click "Agree and continue", you can continue to use our services. Thank you for your cooperation and understanding.

For more agreement content, please refer to the《Openl Qizhi Community AI Collaboration Platform Usage Agreement》

Agree and continue
Disagree, exit