Re: 48GB NUMA-Q boots, with major IO-APIC hassles

From: William Lee Irwin III (wli@holomorphy.com)
Date: Fri Mar 28 2003 - 00:08:39 EST


On Wed, Jan 15, 2003 at 02:58:02AM -0800, William Lee Irwin III wrote:
> Minor extrapolation: aside from potential explosions in very unusual
> corner cases, with these hacks/workarounds 64GB NUMA-Q should boot and
> run (slowly) with an approximate LowTotal of 173296 kB. The main obstacle
> is our setup here would require an additional NR_CPUS > BITS_PER_LONG
> patch, and there isn't much local interest in even seeing whether or how
> poorly it would run without working patches (e.g. hugh's MMUPAGE_SIZE that
> I'm fwd. porting) to do something about runaway mem_map lowmem consumption.

I was only 3MB off wrt. mainline's 64GB LowTotal, not bad at all:

HighTotal: 65134592 kB
HighFree: 65116864 kB
LowTotal: 176076 kB
LowFree: 144180 kB

Turns out NR_CPUS > BITS_PER_LONG was avoided. I'll dig up whatever else
I can here. AIM7 with 10000 tasks and various other things were runnable
on 48GB, but I still need to straighten various fragmentation things out
before benching produces meaningful numbers instead of runs vs. doesn't.

-- wli
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/



This archive was generated by hypermail 2b29 : Mon Mar 31 2003 - 22:00:31 EST