On Mon, 2006-05-15 at 20:37 +0200, Andi Kleen wrote:
> My arguments for remove:
> - The code is very hackish - it was written before the proper ACPI
> infrastructure is in place - and NUMA on 32bit in general needs a lot
> of hacks because of the limited ZONE_NORMAL.
> - NUMA on 32bit is kind of broken by design.
> - It isn't used much.
> - It breaks often
> - It tends to not work on Opterons and hits the users who try it
> there.
I think you are right: there are very few end users running with
CONFIG_NUMA on normal x86. But, there is a disproportionately large
number of developers who do it. There are quite a few IBM (and maybe
more via OSDL) developers who's only access to real NUMA hardware is x86
NUMAQs and Summit machines. When somebody says "foo is broken on NUMA",
I go right to an x86 box.
Anyway, I'd like to think that we've contributed enough to the generic
NUMA code to have earned our keep and allow our little x86 NUMA "hacks"
to remain. x86 is a legacy architecture now anyway, right? ;)
-- Dave
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
[Index of Archives]
[Kernel Newbies]
[Netfilter]
[Bugtraq]
[Photo]
[Stuff]
[Gimp]
[Yosemite News]
[MIPS Linux]
[ARM Linux]
[Linux Security]
[Linux RAID]
[Video 4 Linux]
[Linux for the blind]
[Linux Resources]
