[RFC 0/3] nvme uring passthrough diet

[Kanchan Joshi]

On Mon, May 01, 2023 at 08:33:03AM -0700, Keith Busch wrote:
>From: Keith Busch <kbusch at kernel.org>
>
>When you disable all the optional features in your kernel config and
>request queue, it looks like the normal request dispatching is just as
>fast as any attempts to bypass it. 

Did a quick run, and it is not the case.
For that workload [1], this one moves to 3.4M from 3M.
While pure bypass moved to 5M from 3M. Maybe it could have gone higher
than 5M (which was the HW limit); have not checked that part yet.

[1]
# t/io_uring -b512 -d64 -c2 -s2 -p1 -F1 -B1 -O0 -n1 -u1 -r4 -k0 /dev/ng0n1
submitter=0, tid=2935, file=/dev/ng0n1, node=-1
polled=1, fixedbufs=1/0, register_files=1, buffered=1, register_queues=0 QD=64
Engine=io_uring, sq_ring=64, cq_ring=64
IOPS=3.31M, BW=1616MiB/s, IOS/call=2/1
IOPS=3.36M, BW=1642MiB/s, IOS/call=2/2
IOPS=3.40M, BW=1661MiB/s, IOS/call=2/1
Exiting on timeout
Maximum IOPS=3.40M