[linux-elitists] Dual bore - a question on diagnosis

Phil Mayers p.mayers@imperial.ac.uk
Thu Dec 21 06:19:50 PST 2006

On Wed, Dec 20, 2006 at 02:08:01PM -0500, Dave Jones wrote:
>On Wed, Dec 20, 2006 at 01:49:39PM -0500, Greg Folkert wrote:
> > > or some lower value (keep it less than 1 gig) to the kernel at the GRUB
> > > or LILO command line.  I have seen cases (with first generation Centrino
> > > laptops) where this would eliminate extreme sluggishness in computers
> > > with >= 1 gig of RAM and a CONFIG_HIGHMEM=y kernel.
> > 
> > Okay, but let us say I have 4GB of RAM installed.
> > 
> > Effectively what would that buy me? Are you saying that 3GB would be
> > wasted and not be worth putting in?
> > 
> > Or are you saying that 1GB would be used initially, then at some point
> > later the machine would magically (or non-magically) recognize the
> > additional memory giving full value.
>CONFIG_HIGHMEM ? wtf is this ? 1999 ?
>In the unlikely event that the computer has the one generation
>of core duo that didn't have 64bit support, send it back and ask
>for a real computer.

Hi Dave - Will says hello.

It's not that (I think). The 64bit FC6 actually runs better than the 
32bit one, but still suffers from the slowdowns.

What's really annoying is, it's hard to pin down what is slow. The 
entire machine feels (extremely) slow, but that could be any one of 
SATA, RAM, irqbalance going crazy, bus, power saving down-clocking the 

I've been told I can't return it, because HP have said:

RHEL5 will support the hardware and is out in January

...so I've been told to wait a couple of weeks. Quite what HP are doing 
making such announcements to our purchasing manager I don't know, but 
what I do know is that neither a fully-patched FC6 nor Rawhide, nor a 
2.6.20rc1 kernel will run at normal speed on it, meaning it's 
phenomenally unlikely RHEL5 will be any different. If so, I'm looking 
forward to kicking some arse in January when HP turn out to have been 

HP, you just made *the list*.


More information about the linux-elitists mailing list