GPU UUID Explained
Precise and Elegant: HAMi's Scheduling Feature by GPU Type and UUID
In this article, we analyze HAMi's scheduling feature: fine-grained scheduling based on GPU type and even UUID.
Insights, tutorials, and news from our team
In this article, we analyze HAMi's scheduling feature: fine-grained scheduling based on GPU type and even UUID.
HAMi v2.5.0 is released! It adds dynamic MIG support and brings significant improvements in stability and usability. Come and see what's new.
This is the fifth article in the HAMI principle analysis series, providing a simple analysis of the working principles of HAMI-Core, including how it takes effect, how CUDA APIs are intercepted, and how it implements resource limits for GPU core and memory.
This is the fourth article in the HAMI principle analysis series, analyzing how hami-scheduler selects nodes during scheduling, i.e., how advanced scheduling strategies like Spread and Binpack are implemented.
This is the third article in the HAMI principle analysis series, analyzing the hami-scheduler workflow.
In the last article, we analyzed hami-device-plugin-nvidia and understood how HAMI's NVIDIA device plugin works. This is the second article in the HAMI principle analysis series, analyzing the implementation of hami-scheduler.
This is the first article in a series analyzing the implementation principles of the open-source vGPU solution HAMI, focusing on the hami-device-plugin-nvidia.
In the previous post, I introduced the open-source vGPU solution HAMI. This article focuses on testing its GPU Core & Memory isolation features.
This article introduces HAMi, an open-source GPU virtualization solution, including installation, configuration, and usage.
Get the latest technical articles, tutorials, and HAMi community updates.