Re: [Qemu-devel] MTTCG status updates, benchmark results and KVM forum plans

[Alex Bennée]

Peter Maydell <address@hidden> writes:

> On 15 August 2016 at 11:46, Alex Bennée <address@hidden> wrote:
>> I only ran up to -smp 8 as that is as
>> much as the -m virt model will actually accept.
>
> FWIW, -machine gic-version=3 should allow you more than 8 cores.

Good to know. Thanks.

>> I have noticed some instability in the test though for high -smp values
>> which caused the test runners timeout protection to kick in. These look
>> like guest hangs and maybe barrier related (store-after-load re-ordering
>> can happen). I plan to apply the barrier patches and see if this
>> improves the stability of the tests.
>
>> From my point of view I think we are looking pretty good for merging but
>> I would like to get input from the TCG maintainers who are the ones that
>> will need to accept the work into their tree.
>
> Your note above about instability and hangs is the main thing that
> makes me nervous about merging...

Don't worry I won't be proposing any merge while I can still provoke
hangs in the guest!

My point is you actually have to work quite hard to trigger these and
they are subtle emulation failures that trip up the guest rather than
crashes that take down QEMU itself.

I wanted to post the numbers I'd collected so far because I feel we are
shaping up quite well. Just one more hill... ;-)

--
Alex Bennée