High-Performance Embedding and Reranking on Ascend Hardware with mis-tei

In the Ascend hardware ecosystem, enference engines like vLLM have demonstrated excellent performance for generation tasks and are widely adopted. However, for retrieval-related scenarios such as Embedding and Reranker tasks, specialized inference components like mis-tei offer efficient capabilities that are better suited to these specific use ...

Posted on Thu, 20 Aug 2026 16:24:39 +0000 by mescal

Querying CANN Version and NPU Device Details Using npu-smi and System Files

To determine the installed CANN software version and inspect NPU device status on Atlas platforms, administrators can use both file-based metadata and the npu-smi command-line utility. Checking CANN Toolkit Version via Installation Metadata The CANN toolkit and related packages store version information in plain-text configuration files within ...

Posted on Sat, 11 Jul 2026 17:13:02 +0000 by ublapach

Accelerated Multi-node Inference with Ascend: Simplified Deployment of Large-scale Models Using GPUStack

Deploying large-scale models on Ascend NPUs often presents a significant challenge due to the complexity of configuring distributed inference using the standard MindIE engine. Although its performance is acceptable, the setup process involves intricate steps such as environment preparation, initialization, and fine-tuning of parameters. Even mi ...

Posted on Fri, 08 May 2026 09:15:03 +0000 by spasme