allocate CPU and GPU on the same numa node for guaranteed pod #1286
Description
What would you like to be added:
Allocating CPU and GPU on the same numa node
What type of PR is this?
/kind feature
What this PR does / why we need it:
In scenarios where frequent communication occurs between the CPU and GPU, accessing across NUMA (Non-Uniform Memory Access) nodes can lead to increased latency, limited bandwidth, and other issues, thereby compromising the overall system performance.