Merged
feat: vector算子适配ascend960pr(DV122)平台并补齐仓库级960pr支持 #14537
biabuAdd on 22 days ago
feat: vector算子适配ascend960pr(DV122)平台并补齐仓库级960pr支持 #14537
Merged
Pull Request has been successfully merged by CANN-robot.
（Thank you for biabu's contribution.）
22 days ago Added label: cann-cla/yes
CANN-robot
22 days ago Comments：
22 days ago Comments：
Thanks for your pull-request.
The full list of commands accepted by me can be found at here.
You can get sig-info at here.
You can self-configure the PR merge rules for this repository. For more details, please refer to Here.
For more, you also can visit HICANN.
PR Approval Progress
✅ Congratulations! All modules have met the lgtm and approve requirements.
Module Approval Details
| module | lgtm status | approve status |
|---|---|---|
| */*/op_graph/*_proto.h | ✅ 汤磊, 马兵, 刘丹 (3/2) | ✅ 汤磊 (1/1) |
| */*/op_host/*_def.cpp | ✅ xuejinghui, 刘丹, 马兵, 汤磊 (4/2) | ✅ 汤磊 (1/1) |
| attention | ✅ 马兵, 刘丹 (2/2) | ✅ 马兵, 刘丹 (2/1) |
| gmm | ✅ 马兵, 刘丹 (2/2) | ✅ 马兵, 刘丹 (2/1) |
| mhc | ✅ xuejinghui, 刘丹, 马兵 (3/2) | ✅ xuejinghui (1/1) |
| moe | ✅ 刘丹, 马兵 (2/2) | ✅ 马兵, 刘丹 (2/1) |
| posembedding | ✅ 马兵, 刘丹 (2/2) | ✅ 马兵, 刘丹 (2/1) |
| sig-ops-transformer | ✅ 刘丹, 马兵 (2/2) | ✅ 马兵, 刘丹 (2/1) |
| tests | ✅ 马兵, 刘丹 (2/2) | ✅ 马兵, 刘丹 (2/1) |
💡 Tip:
- Committer can comment
/approveor/lgtm- Commenting
/approveimplies both code review (lgtm) and intent to merge (approve)
CLA Signature Pass
biabu111, thanks for your pull request. All authors of the commits have signed the CLA. 👍
CANN-robot22 days ago Set guijianwei,hz36amy_00,yangziqi_007,coder_linx,juyangokok,monologue815,xuanyuandy,chen-linxin4,crystalhu,yu-xinjie62,huangwei791,jiang-lirui,yang-binrong,Allan_Yu,wang-minbo,huang-wei-chen,tang-lei01,xtqh,songkai111,liudan12,shasha_an,wangzhe123456789,Wei_NaChuan,zhangshuoleiokok,luwei_1104,cc-z,kdy18482276080,xuchengyang,shen_weiling,gaoxiang618,songjionghui,lining_210,wenxin_fight,miaofangzheng,mabing1118,tang-hao-hw-gitcode,wangyongguang,haijie_699874,xuejinghui,zzy__,Julius47,yuhao_,L_Euler,jiangjiawei,wangxun21 as approver
22 days ago Set guijianwei,hz36amy_00,yangziqi_007,coder_linx,juyangokok,monologue815,xuanyuandy,chen-linxin4,crystalhu,yu-xinjie62,huangwei791,jiang-lirui,yang-binrong,Allan_Yu,wang-minbo,huang-wei-chen,tang-lei01,xtqh,songkai111,liudan12,shasha_an,wangzhe123456789,Wei_NaChuan,zhangshuoleiokok,luwei_1104,cc-z,kdy18482276080,xuchengyang,shen_weiling,gaoxiang618,songjionghui,lining_210,wenxin_fight,miaofangzheng,mabing1118,tang-hao-hw-gitcode,wangyongguang,haijie_699874,xuejinghui,zzy__,Julius47,yuhao_,L_Euler,jiangjiawei,wangxun21 as approver
CANN-robot22 days ago Set guijianwei,hz36amy_00,yangziqi_007,coder_linx,juyangokok,monologue815,xuanyuandy,chen-linxin4,crystalhu,yu-xinjie62,huangwei791,jiang-lirui,yang-binrong,Allan_Yu,wang-minbo,huang-wei-chen,tang-lei01,xtqh,songkai111,liudan12,shasha_an,wangzhe123456789,Wei_NaChuan,zhangshuoleiokok,luwei_1104,cc-z,kdy18482276080,xuchengyang,shen_weiling,gaoxiang618,songjionghui,lining_210,wenxin_fight,miaofangzheng,mabing1118,tang-hao-hw-gitcode,wangyongguang,haijie_699874,xuejinghui,zzy__,Julius47,yuhao_,L_Euler,jiangjiawei,wangxun21 as approver
22 days ago Set guijianwei,hz36amy_00,yangziqi_007,coder_linx,juyangokok,monologue815,xuanyuandy,chen-linxin4,crystalhu,yu-xinjie62,huangwei791,jiang-lirui,yang-binrong,Allan_Yu,wang-minbo,huang-wei-chen,tang-lei01,xtqh,songkai111,liudan12,shasha_an,wangzhe123456789,Wei_NaChuan,zhangshuoleiokok,luwei_1104,cc-z,kdy18482276080,xuchengyang,shen_weiling,gaoxiang618,songjionghui,lining_210,wenxin_fight,miaofangzheng,mabing1118,tang-hao-hw-gitcode,wangyongguang,haijie_699874,xuejinghui,zzy__,Julius47,yuhao_,L_Euler,jiangjiawei,wangxun21 as approver
218 messages are collapsed here。 See more
5 days ago Added label: approved
5 days ago Added label: lgtm
5 days ago Closed associated issue
5 days ago Merged pull request
描述
DV122（Ascend960PR / dav_9202 / Ascend960PR_969A1）vector 算子适配。适配原则：9201 支持的能力 9202 均支持。
仓库级适配（原仓库完全缺失 960pr）：
注册级适配（15 算子）：
代码级补齐：
注册级适配（66 算子）：
CMakeLists.txt：SUPPORT_COMPUTE_UNIT 追加 "ascend960pr"，SUPPORT_TILING_DIR 按位插入对应 arch35op_host/*_def.cpp：AICore().AddConfig("ascend960pr", <960dt 同款配置>)op_host/config/ascend960pr/：二进制配置（*_binary.json / *_simplified_key.ini，复制自 ascend960dt，与 960dt 内容一致）op_kernel*/CMakeLists.txt：COMPUTE_UNITS 追加 ascend960pr（arch92 kernel 入口同步生效）scripts/kernel/binary_config/ascendc_config.json：compute_units 与 compile_options 补齐代码级补齐（9201 → 9201||9202）：
NpuArch::DAV_9201、__NPU_ARCH__ == 9201、"Ascend960DT"等 960 平台门禁补齐DAV_9202/9202/"Ascend960PR"关联的Issue
https://gitcode.com/cann/ops-transformer/issues/7300
测试
已通过冒烟、UT、ST测试。
--soc=ascend960pr构建 19 个抽样算子（覆盖全部修改类型）：宿主侧（def/tiling/infershape/aclnn）100% 编译通过，0 错误config/ascend960pr/*_binary.json--soc=ascend960dt同算子集构建行为一致，无回归文档更新
不涉及
类型标签
AI/Agent生成声明