Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Ascend #9

Open
wants to merge 241 commits into
base: main
Choose a base branch
from
Open

Ascend #9

wants to merge 241 commits into from

Conversation

zhangyue207
Copy link

No description provided.

Chamberlain0w0 and others added 30 commits June 12, 2024 06:57
@Chamberlain0w0
add cnnl support and rmsnorm
1c54dfe
@Chamberlain0w0
add swiglu
d7f5c99
@PanZezhong1725
fix: 修改python寻找算子库方式，更新文档
5eb752c
@Chamberlain0w0
Merge branch 'dev' into cnnl_support
4600218
@PanZezhong1725
feat (operator): 添加cpu和cuda的causal softmax算子
e13b03b
@PanZezhong1725
Merge pull request #14 from PanZezhong1725/causal_softmax
66706fa 
feat (operator): 添加cpu和cuda的causal softmax算子
@kilinchange
feat(ops): 添加 matmul cuda 算子和 blas 矩阵抽象
5fb0d54
@kilinchange
feat(ops): 添加 matmul cpu 算子
828173b
@kilinchange
feat(ops)：添加 matmul 测试；重构 matmul cpu & cuda 实现
4659579
@kilinchange
feat(ops)：将 cublas handle 的创建移动到算子外部
3e39243
@kilinchange
feat(devices): 添加 handle 池的无锁栈实现
5c94a53
@kilinchange
refact(matmul): 适配框架变动
d09c3ee
@kilinchange
feat(devices): 改成每张卡创建一个 handle 池
6494217
@kilinchange
fix(devices): 按照 reviewer 意见修改代码。
a4dc3f9
@kilinchange
fix(devices): 按照 reviewer 意见修改。
78424f7
@kilinchange
fix(devices)：修正 pool 移动构造函数。
2643516
@PanZezhong1725
fix (kernel): 支持cuda rms norm动态维度
645670a
@PanZezhong1725
Merge pull request #11 from PanZezhong1725/add_matmul
c349f50 
添加 matmul 算子
@Chamberlain0w0
fix
f40e54f
@Chamberlain0w0
Merge branch 'dev' into cnnl_support
b32d978
@Chamberlain0w0
fix rope on mlu
2b2929e
@Chamberlain0w0
Merge branch 'cnnl_support' of github.com:PanZezhong1725/operators in…
e1a25eb 
…to cnnl_support
@Chamberlain0w0
fix format
f8b55da
@Chamberlain0w0
fix
0e30765
@bitzyz
feat: 添加reform的cpu/cuda实现代码
12ddb2b
@bitzyz
feat: cpu的reform算子使用openmp加速
a71ae02
@Chamberlain0w0
fix swiglu
3b1d8ef
@Chamberlain0w0
rename
56851a5
@PanZezhong1725
reformat：reform
3cfd795
@PanZezhong1725
Merge pull request #13 from PanZezhong1725/dev-reform
eb48fd1 
添加 reform 算子
zhangyue and others added 30 commits September 23, 2024 14:25
fix swiglu test
9dcd9e7
fix format
af0145d
@PanZezhong1725
Merge pull request #59 from PanZezhong1725/refactor_rms_norm
84325ac 
重构rms_norm算子的cpu、cuda接口
Merge branch 'dev' of github.com:PanZezhong1725/operators into ascend
b9bb601
@Ziminli
Merge branch 'dev' of github.com:PanZezhong1725/operators into add_Re…
c73f852 
…sNet_operators
meet ci
e38b22a
meet ci
5f6a32a
@Ziminli
Remove duplicate xmake -fPIC flag
564ab1a
@Ziminli
Changed the implementation of operator== of DataLayout
362fb94
@PanZezhong1725
Merge pull request #41 from PanZezhong1725/add_ResNet_operators
248559d 
refactored Add operator and fixed f16 and f32 conversion functions
@Ziminli
Fixed add to correctly handle data size that is less than 4 or not a …
8237cf2 
…multiple of 4
@Ziminli
Add empty test case
19802cd
@PanZezhong1725
Merge pull request #65 from PanZezhong1725/add_ResNet_operators
d25e228 
Fixed Add to correctly handle any data size
@PanZezhong1725
fix: 在rope测试中使用int64 posid以支持旧版pytorch
1d414ab
change name from DEVICE_NPU to DEVICE_ASCEND
e05d7a2
合并 use_aclnn
9d33b60
delete mallocworkspace func
97ea157
add shared_ptr for AscendHandle
dec43af
fix softmax test
d7c92fc
fix handle
ac5a6fa
matmul 支持不连续tensor
6a35053
fix storageshape
22868e9
@PanZezhong1725
Merge pull request #69 from PanZezhong1725/rope_pytorch
2075ba9 
修复rope测试旧版pytorch不支持uint64的问题
fix rms_norm datatype
e12c484
swiglu aclnn
0dfc4b4
Merge branch 'dev' of github.com:PanZezhong1725/operators into ascend
8b0b089
回滚 test
d870fb9
回滚cpu编译参数
55eb726
fix format
ab4550a
fix format
867868a
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers
No reviews
Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
9 participants
@zhangyue207 @Chamberlain0w0 @PanZezhong1725 @kilinchange @bitzyz @crapromer @xgqdut2016 @JYMiracle305 @Ziminli