[Techtalk] Trouble with upgrading RAM

[Riccarda Cassini]

Hi,

thankyou very much to all who replied!  (to keep it short, I'm replying
to just one message here - but I equally appreciate all other input...)


Tracey Clark wrote:
> RAM chips that look identical can still have some major things different
> about them. An important one is ECC - Error checking. If one is ECC
> memory and the other is not, that will cause problems. Also if they are
> different clock speeds, that would definitely cause problems, but
> probably make the sytem unable to boot at all. Check the manufacturer
> model numbers on the stickers on the RAM. Look those up on the
> Fujitsu-Siemens website, and make sure these specs are the same. That's a
> good starting point.

Okay, did my homework: after quite some time of unsuccessful poking
around at the company's website, I finally remembered google and
entered HYS72V32300GU. The model numbers of both sticks start with the
common substring HYS72V32300GU-7-D, followed by some other differing
numbers (I figured those could be a serial number or something...)

Thanks google I'm now quite sure that I have

  PC133, SDRAM 256MB, 32M x 72-Bit, ECC, 133MHz, CL=2, 3.3V, 168-pin,
  unbuffered DIMM module  (manufactured by Infineon)

and that both sticks are apparently the same.  After having read (or
tried to read) a couple of other documents on that subject, I think I
even have a rough understanding of what that means :-)

The guy who I got the memory stick from, also had taken a look at both
modules, before he told me that there should be no problem using both
sticks together (as far as I can tell, he's at least somewhat
hardware-savvy). But I 'forgot' to ask him what made him so sure - that
might have saved me a bit of googling...

BTW, I was more than naive as to the gazillion of different DRAM
memory types there exist: SDRAM, DDR, DDR2, RLDRAM (all in either
unbuffered, registered, ECC, 1.8/2.5/3.3V, etc. versions)...

I downloaded the datasheet PDF for my stick, but then concluded that
its 16-pages full of technical details is definitely more than I ever
wanted to know...  Strangely, though, I didn't find that single one
parameter which I would've found interesting (WRT the power supply
theory), i.e. the total power consumption of the stick (that's what my
coworker told me to look up).  The only similar parameter I found is
under "Maximum ratings" and reads "Power dissipation per SDRAM
component: 1W".  Assuming that "component" refers to a single chip (of
which there are nine), then we'd have 9W.  Doesn't seem too much in
relation to the 300W that the power supply claims to be able to handle.

Unfortunately, that still leaves the issue unresolved for the moment.
Someone else suggested (off-list) that the second socket might be the
problem.  I guess I'll do some more testing. This will take its time,
though, because the crashes are only occurring sporadically.

> 
> > Someone at work suggested it might be the power supply.
> 
> What is their reasoning?

if I'm repeating it correctly, his reasoning went along the lines of
the power supply voltage dropping by some (tiny) amount due to the
additional power consumption. Presuming it was already near the lower
limit before (due to ageing, tolerances, whatever), this might then
just suffice to cause sporadic failures, especially if some transient
spikes happen to come in via the power line. Something like that.

On the other hand, I don't have too many components in that computer -
it's pretty standard: 1 harddisk, CD-ROM, CD-burner, floppy, graphics
card, soundcard, NIC.  However, I was told that this particular version
of the Athlon processor is consuming a substantial amount of energy... 
Also, I never tried to plug in some other additional component - maybe
that would've caused the same kind of problem(?)


> > [*] BTW, any recommendations on which filesystem to use?
> 
> I've had good experiences with xfs. It's recovery time after crashes is
> nearly nill. It used to have stability problems, but those are generally
> rumored to be cleared out. A presenter at my local Linux group brought in
> a 5 blade server running xfs doing streaming video, shut it down while it
> was streaming, brought it back up, and the boot process took maybe a
> whole second longer. Since he worked for Industrial Light & Magic
> (Lucasfilms) they were very worried about reliability and stability of
> their servers. They had only good experiences with xfs.

Sounds good.  Thanks also to Devdas Bhagat for the additional comments
on the other filesystems.
Well, I guess I'll give xfs a go, sometime soon.

Riccarda



-- 
"Sie haben neue Mails!" - Die GMX Toolbar informiert Sie beim Surfen!
Jetzt aktivieren unter http://www.gmx.net/info