[clue-tech] Bleepin' Memory

[Jim Ockers]

Hi Keith,

You could maybe work around this system by changing the timings that
your motherboard or CPU uses for memory access.  This sounds like
a bad timing issue where some components on the motherboard might
be marginal spec, and some components on the RAM might be marginal
spec, and when you put them together you get occasional problems.

Try underclocking your system, if you can change the frequencies in
the BIOS.

Try decreasing the main system bus speed (100MHz => 66MHz, for
example.)

Try telling the BIOS that the memory is slower than it really is.

Try turning off the CPU's external memory cache.

Try increasing the number of wait states or SDRAM access delays.

Look for other BIOS options (since they are all board-specific)
that look like they might have something to do with speeds or
memory access and slow them down.

Obviously if you slow them down too much you will start to notice
that your system is too slow for what you paid for it...

Hope this helps,
Jim

--
Jim Ockers, P.Eng. (ockers at ockers.net)
Contact info: please see http://www.ockers.net/


Keith Hellman wrote:
> 
> 
> Hello all:
> 
> I'm looking for some advice on the next-best-step in a hardware
> diagnosis.
> 
> A little more than a year ago I bought a Gigabyte 865PE motherboard, a
> P4, and a 512MB DDR ram for a new machine.  It has in general performed
> well.  Of particular annoyance, however, was that I was never able to
> successfully backup to my DVDRAM drive.  Kept getting corrupt BZIP files
> that I could not fully decompress.  Oh well, I thought, my drive is
> going on the fritz (it is by now several years old) and I adopted other
> backup solutions.
> 
> A couple weeks back I'm backing up and bzip segfaults.  Amazingly, it
> regurgitates a message saying (paraphrased)
>   This is not a bzip bug.  You probably have subtle memory corruption
>   which other programs may ignore (or not even notice) but bzip is
>   sensitive to minor bit twiddles - so we crash.
> 
>   We repeat - this is not a bzip bug.  Go fix your memory.
> 
> I have to admit I was pretty astounded.  It so happens that about the
> same time I was doing some big GIMP work & I had noticed the GIMP
> crashing occasionally, which certainly is *not* what I'm used to.
> 
> So.  I whip out the ol' memtest86+ program (http://www.memtest86.com)
> and after an hour or so it started to find errors.  I chide myself that
> this is what I get for skimping and buying non-kensington memory, order
> myself a new (Kensington this time) module, and wait.
> 
> The memory arrived the other day, I slap it in, figure I'll run
> memtest86+ again just for giggles and, you guessed it, memmory errors
> again.
> 
> The specific tests that are failing are #3 (about 25% of the time), #4
> (about 75% of the time), and 1 #2 failure out of the 60 errors.  Details
> of these tests are at
>   http://www.memtest86.com/#details
> it seems I have confirmed 'subtle & data sensitive errors'.  I would
> also point out that memtest86+ can go through multiple passes *without*
> any errors and that the memory location reported is inconsistent.  My
> fear is that I have a flaky mobo or p4 cache and this is why I pose
> these questions to the group:
> - has anyone experienced a similar problem with memory?
> - ... and if so how did you determine the culprit?
> 
> I don't like the idea of having to start swapping out either the
> processor or the mobo (particularly since I have neither the spares or
> the free time...)
> - is there a store that could thoroughly test the p4?
> - the mobo?
> - what type of cost am I looking at? =20
> - can anyone recommend a particular place?
> 
> Thanks in advance for everyone's help.
> 
> --=20
> Keith Hellman                             #include <disclaimer.h>
> khellman at mcprogramming.com                from disclaimer import standard
> public key @ www.mcprogramming.com
> 
> "I like long walks, especially when they are taken by people who annoy
> me."
> 
> -- Noel Coward
>