On Mon, Dec 05, 2016 at 10:40:49AM -0600, Karl Rister wrote:
> On 12/05/2016 08:56 AM, Stefan Hajnoczi wrote:
>
>
> > Karl: do you have time to run a bigger suite of benchmarks to identify a
> > reasonable default poll-max-ns value? Both aio=native and aio=threads
> > are important.
> >
> > If
On 12/05/2016 08:56 AM, Stefan Hajnoczi wrote:
> Karl: do you have time to run a bigger suite of benchmarks to identify a
> reasonable default poll-max-ns value? Both aio=native and aio=threads
> are important.
>
> If there is a sweet spot that improves performance without pathological
> cases
On Mon, Dec 05, 2016 at 06:25:49PM +0800, Fam Zheng wrote:
> On Thu, 12/01 19:26, Stefan Hajnoczi wrote:
> > One way to avoid the costly exit is to use polling instead of notification.
>
> Testing with fio on virtio-blk backed by NVMe device, I can see significant
> performance improvement with
On 12/01/2016 08:26 PM, Stefan Hajnoczi wrote:
> v4:
> * Added poll time self-tuning algorithm [Christian and Paolo]
> * Try a single iteration of polling to avoid non-blocking
> ppoll(2)/epoll_wait(2) [Paolo]
> * Reordered patches to make performance analysis easier - see below
This looks
On Thu, 12/01 19:26, Stefan Hajnoczi wrote:
> One way to avoid the costly exit is to use polling instead of notification.
Testing with fio on virtio-blk backed by NVMe device, I can see significant
performance improvement with this series:
poll-max-nsiodepth=1iodepth=32
On Thu, Dec 01, 2016 at 07:26:39PM +, Stefan Hajnoczi wrote:
> v4:
> * Added poll time self-tuning algorithm [Christian and Paolo]
> * Try a single iteration of polling to avoid non-blocking
> ppoll(2)/epoll_wait(2) [Paolo]
> * Reordered patches to make performance analysis easier - see
> v4:
> * Added poll time self-tuning algorithm [Christian and Paolo]
> * Try a single iteration of polling to avoid non-blocking
> ppoll(2)/epoll_wait(2) [Paolo]
> * Reordered patches to make performance analysis easier - see below
Reviewed-by: Paolo Bonzini
> v3:
>
v4:
* Added poll time self-tuning algorithm [Christian and Paolo]
* Try a single iteration of polling to avoid non-blocking
ppoll(2)/epoll_wait(2) [Paolo]
* Reordered patches to make performance analysis easier - see below
v3:
* Avoid ppoll(2)/epoll_wait(2) if polling succeeded [Paolo]
*