[openib-general] Re: [PATCH 18 of 20] ipath - kbuild infrastructure

2006-03-13 Thread Bryan O'Sullivan
On Mon, 2006-03-13 at 20:36 +0100, Sam Ravnborg wrote: > Use the kernel settings. We cannot have this modified by each and every > driver. Fair enough. http://openib.org/mailman/listinfo/openib-general To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general

[openib-general] Re: [PATCH 18 of 20] ipath - kbuild infrastructure

2006-03-13 Thread Sam Ravnborg
On Mon, Mar 13, 2006 at 11:24:28AM -0800, Bryan O'Sullivan wrote: > On Mon, 2006-03-13 at 19:10 +0100, Adrian Bunk wrote: > > > I'm still a bit surprised, since in the rest of the kernel we are even > > going from -O2 to -Os for getting better performance. > > > > Robert said he wanted to post s

[openib-general] Re: [PATCH 18 of 20] ipath - kbuild infrastructure

2006-03-13 Thread Bryan O'Sullivan
On Mon, 2006-03-13 at 19:10 +0100, Adrian Bunk wrote: > I'm still a bit surprised, since in the rest of the kernel we are even > going from -O2 to -Os for getting better performance. > > Robert said he wanted to post some numbers showing that -O3 is > measurably better for you [1], but I haven'

[openib-general] Re: [PATCH 18 of 20] ipath - kbuild infrastructure

2006-03-13 Thread Robert Walsh
> I'm still a bit surprised, since in the rest of the kernel we are even > going from -O2 to -Os for getting better performance. > > Robert said he wanted to post some numbers showing that -O3 is > measurably better for you [1], but I haven't seen them. As you might have guessed, I kind of forg

[openib-general] Re: [PATCH 18 of 20] ipath - kbuild infrastructure

2006-03-13 Thread Adrian Bunk
On Thu, Mar 09, 2006 at 04:35:48PM -0800, Bryan O'Sullivan wrote: >... > --- /dev/null Thu Jan 1 00:00:00 1970 + > +++ b/drivers/infiniband/hw/ipath/MakefileThu Mar 9 16:17:00 2006 -0800 > @@ -0,0 +1,42 @@ > +EXTRA_CFLAGS += -O3 I'm still a bit surprised, since in the rest of the kernel

[openib-general] Re: [PATCH 18 of 20] ipath - kbuild infrastructure

2006-03-09 Thread Sam Ravnborg
On Thu, Mar 09, 2006 at 11:00:07AM -0800, Roland Dreier wrote: > Sam> Eventually - yes. But not just now. Kbuild was introduced > Sam> because it was needed in the top-level directory and it made > Sam> good sense to do so. But for now keeping Makefile is a good > Sam> choice. Thi

[openib-general] Re: [PATCH 18 of 20] ipath - kbuild infrastructure

2006-03-09 Thread Sam Ravnborg
On Thu, Mar 09, 2006 at 11:00:07AM -0800, Roland Dreier wrote: > Sam> Eventually - yes. But not just now. Kbuild was introduced > Sam> because it was needed in the top-level directory and it made > Sam> good sense to do so. But for now keeping Makefile is a good > Sam> choice. Thi

[openib-general] Re: [PATCH 18 of 20] ipath - kbuild infrastructure

2006-03-09 Thread Roland Dreier
Sam> Eventually - yes. But not just now. Kbuild was introduced Sam> because it was needed in the top-level directory and it made Sam> good sense to do so. But for now keeping Makefile is a good Sam> choice. This is anyway what people are used to. OK, disregard my suggestion then.

[openib-general] Re: [PATCH 18 of 20] ipath - kbuild infrastructure

2006-03-09 Thread Sam Ravnborg
On Thu, Mar 09, 2006 at 09:53:04AM -0800, Roland Dreier wrote: > > + depends on 64BIT && (PCIEPORTBUS || X86_HT) > > > --- /dev/null Thu Jan 1 00:00:00 1970 + > > +++ b/drivers/infiniband/hw/ipath/Makefile Thu Mar 9 08:46:47 2006 -0800 > > I've been suggesting that new files be c

[openib-general] Re: [PATCH 18 of 20] ipath - kbuild infrastructure

2006-03-09 Thread Roland Dreier
> +depends on 64BIT && (PCIEPORTBUS || X86_HT) Why do you depend on X86_HT? I think you got confused: the HT stands for hyperthreading, not hypertransport. In fact if you compile a kernel optimized for K8, X86_HT is disabled. And why do you depend on PCIEPORTBUS? I don't see you using any