Dear Stephen,

The problem has not been solved, but I found a workaround. According to the documentation (https://doc.dpdk.org/guides/prog_guide/gpudev.html, sec 11.3), rte_extmem_register should be invoked with GPU_PAGE_SIZE as an argument. If GPU_PAGE_SIZE is set to 2 MB instead of 64 kB, registration of 72 GB of GPU memory (on a Grace Hopper) is done in about ten seconds, not hours.

  rte_extmem_register(ext_mem.buf_ptr, ext_mem.buf_len, NULL, ext_mem.buf_iova, GPU_PAGE_SIZE);

Thanks, John Romein

On 05-10-2024 00:16, Stephen Hemminger wrote:

Dear Slava, Thank you for looking at the patch. With the original code, I saw that the application spent literally hours in this function during program start up, if tens of gigabytes of GPU memory are registered. This was due to qsort being invoked for every new added item (to keep the list sorted). So I tried to write equivalent code that sorts the list only once, after all items were added. At least for our application, this works well and is /much/ faster, as the complexity decreased from n^2 log(n) to n log(n). But I must admit that I have no idea /what/ is being sorted, or why; I only understand this isolated piece of code (or at least I think so). So if you think there are better ways to initialize the list, then I am sure you will be absolutely right. But I will not be able to implement this, as I do not understand the full context of the code. Kind Regards, John