On Tue, Nov 27, 2018 at 21:38:22 -0800, Richard Henderson wrote:
> The intent here is to remove several move insns putting the
> function arguments into the proper place. I'm hoping that
> this will solve the skylake regression with spec2006, as
> seen with the ool softmmu patch set.
On Fri, Nov 30, 2018 at 08:15:56 +0100, Laurent Desnogues wrote:
> On Fri, Nov 30, 2018 at 4:00 AM Emilio G. Cota wrote:
> >
> > On Thu, Nov 29, 2018 at 19:39:15 -0500, Emilio G. Cota wrote:
> > > A64 and POWER9 host numbers:
> > >
> > > https://imgur.com/a/m6Pss99
> > >
> > > There's quite a
On Fri, Nov 30, 2018 at 4:00 AM Emilio G. Cota wrote:
>
> On Thu, Nov 29, 2018 at 19:39:15 -0500, Emilio G. Cota wrote:
> > A64 and POWER9 host numbers:
> >
> > https://imgur.com/a/m6Pss99
> >
> > There's quite a bit of noise in the P9 measurements, but it's
> > a shared machine so I can't do
On Thu, Nov 29, 2018 at 19:39:15 -0500, Emilio G. Cota wrote:
> A64 and POWER9 host numbers:
>
> https://imgur.com/a/m6Pss99
>
> There's quite a bit of noise in the P9 measurements, but it's
> a shared machine so I can't do much about that.
>
> I'll update the A64 results with error bars
On Thu, Nov 29, 2018 at 11:23:09 -0800, Richard Henderson wrote:
> On 11/28/18 2:15 PM, Emilio G. Cota wrote:
> > Unfortunately, it doesn't seem to help, performance-wise.
>
> That is really disappointing, considering the size gains are huge -- even more
> dramatically for non-x86 hosts. I will
On 11/28/18 2:15 PM, Emilio G. Cota wrote:
> Unfortunately, it doesn't seem to help, performance-wise.
That is really disappointing, considering the size gains are huge -- even more
dramatically for non-x86 hosts. I will see about some more benchmarking on
this for other host/guest combinations.
On Tue, Nov 27, 2018 at 21:38:22 -0800, Richard Henderson wrote:
> The intent here is to remove several move insns putting the
> function arguments into the proper place. I'm hoping that
> this will solve the skylake regression with spec2006, as
> seen with the ool softmmu patch set.
>
> Emilio,
The intent here is to remove several move insns putting the
function arguments into the proper place. I'm hoping that
this will solve the skylake regression with spec2006, as
seen with the ool softmmu patch set.
Emilio, all of this is present on my tcg-next-for-4.0 branch.
r~
Richard
