变更摘要

本 PR 主要为 distribute_barrier 与 moe_update_expert 两个 MC2 算子补充新硬件平台 ascend960dt（DAV_9201 架构）的支持。核心是在算子的原型配置、tiling 与 ACLNN 接口分支中，将 DAV_9201 与既有的 DAV_3510 同等处理，使新平台能够走扩展（extend）通路并采用对应的 worldSize 上限配置；其余改动为指针书写风格（Type * 调整为 Type*）的统一整理，无业务行为变化。

主要改动

• distribute_barrier 接口新增 DAV_9201 分支: 在 aclnnDistributeBarrierGetWorkspaceSizeBase 与 aclnnDistributeBarrierBase 中，将 Ops::Base::DAV_9201 与 DAV_3510 一并纳入判定，使其同样调用 aclnnInnerDistributeBarrierExtendGetWorkspaceSize 与 aclnnInnerDistributeBarrierExtend 扩展通路。
• distribute_barrier tiling 的世界规模上限调整: CheckAndSetAttrs 中 MAX_WORLD_SIZE 的判定新增 npuArch == Ops::Base::DAV_9201，使其与 DAV_3510 一样使用 MAX_WORLD_SIZE_A5。
• distribute_barrier 注册新平台配置: distribute_barrier_def.cpp 中新增 this->AICore().AddConfig("ascend960dt", aicore_config_apt);。
• moe_update_expert 注册新平台配置: moe_update_expert_def.cpp 中新增 this->AICore().AddConfig("ascend960dt", aicore_config);。

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

💡 Tip:

• Committer can comment /approve or /lgtm
• Commenting /approve implies both code review (lgtm) and intent to merge (approve)

CLA Signature Pass

yezhuoxi, thanks for your pull request. All authors of the commits have signed the CLA. 👍

/approve