On 8/5/2019 6:09 PM, Ming Lei wrote:
I am wondering why you use 2 * num_possible_nodes() as the limit instead of num_possible_nodes(), could you explain it a bit?
The number comes from most systems being dual socket systems, thus a numa node count of 2. Some of these dual socket systems can be high cpu counts per socket. We did see a difference, on different architectures and where cpu counts were high per socket, that more hwqs per socket did help. So if there can be more than 1 hwq per socket then I think that is goodness. Additionally, we saw that 4 was a fairly good number memory size wise - it was still big (several hundred MBs with can_queue counts of 4k or 8k), but doubling it to 8 started to make it approach the high 100's of MB. So, unless we had higher numa node counts, I didn't want to raise it any more than 2x the node count. And as 8 looked so big, even with a high numa node count, that seemed a reasonable cap.
-- james
