On 25/08/2017 12:26 PM, Robert Hoo wrote:
(Sorry for yesterday's wrong sending, I finally fixed my MTA and git
send-email settings.)
It's hard to benchmark 40G+ network bandwidth using ordinary
tools like iperf, netperf (see reference 1).
Pktgen, packet generator from Kernel sapce, shall be a candidate.
I then tried with pktgen multiqueue sample scripts, but still
cannot reach line rate.
Try samples 03 and 04.
I then derived this NUMA awared irq affinity sample script from
multi-queue sample one, successfully benchmarked 40G link. I think this can
also be useful for 100G reference, though I haven't got device to test yet.
This script simply does:
Detect $DEV's NUMA node belonging.
Bind each thread (processor from that NUMA node) with each $DEV queue's
irq affinity, 1:1 mapping.
How many '-t' threads input determines how many queues will be
utilized.
I agree this is an essential capability.
This was the main reason I added support for the -f argument.
Using it, I could choose cores of local NUMA, especially for single
thread, or when cores of the NUMA are sequential.
Tested with Intel XL710 NIC with Cisco 3172 switch.
It would be even slightly better if the irqbalance service is turned
off outside.
Referrences:
https://people.netfilter.org/hawk/presentations/LCA2015/net_stack_challenges_100G_LCA2015.pdf
http://www.intel.cn/content/dam/www/public/us/en/documents/reference-guides/xl710-x710-performance-tuning-linux-guide.pdf
Signed-off-by: Robert Hoo <robert...@linux.intel.com>
---
Regards,
Tariq Toukan
