DeepSeek 发布与 Nvidia 组件对应的 Ascend 工具套件
DeepSeek 发布了与 Nvidia 组件对应的 Ascend 工具套件，包括 TileLang、DeepGEMM、DeepEP、TileKernels、FlashMLA 和 DeepSelect。
DeepSeek released an Ascend toolkit matching its Nvidia components: TileLang, DeepGEMM, DeepEP, TileKernels, FlashMLA and DeepSelect.
It says every TileLang kernel used in its training now has a high-performance Ascend implementation. TileLang powers most kernels in V4-series training.
DeepSeek and Huawei also jointly optimized a 128-card Ascend 950 supernode, with key compute and communication tests reportedly approaching hardware limits.
- TileLang: github.com/tile-ai/tilelang
- DeepGEMM Ascend: github.com/deepseek-ai/DeepG…
- DeepEP Ascend: github.com/deepseek-ai/DeepE…
- TileKernels: github.com/deepseek-ai/TileK…
- FlashMLA: github.com/deepseek-ai/Flash…
- DeepSelect: github.com/deepseek-ai/DeepS…
来源：thexpin · x.com