After I filled them all up, the PC reseted
right after RAM counting screen.

Is this a power issue? Or something else?

--
@~@ Might, Courage, Vision, SINCERITY.
/ v \ Simplicity is Beauty! May the Force and Farce be with you!
/( _ )\ (Xubuntu 8.04) Linux 2.6.24-16-generic
^ ^ 20:30:01 up 8:34 2 users load average: 1.00 1.06 1.14
? ? (CSSA):
http://www.swd.gov.hk/tc/index/site_...ub_addressesa/

Man-wai Chang ToDie (33.6k) wrote:
>
> After I filled them all up, the PC reseted
> right after RAM counting screen.
>
> Is this a power issue? Or something else?
>

I can tell you what happened on my system. I purchased
4x256MB of Crucial RAM (low density, 16 chips per DIMM).
Was detected as 1GB total, so no problem there.
But had problems once I got into Windows OR Linux.

The only config that was stable for me, was 2x256MB, running
at PC100 (because my processor runs at 100MHz as well). If
I added one more stick of RAM, I could boot into Windows, but the
computer would crash due to an AGP problem. The symptoms
were reproducible in Linux (a display problem too), so it
was not limited to some issue with Win98SE. That is when I
realized I had an actual hardware problem of some sort.

My motherboard was P2B-S, with 440BX chipset.

So you could be seeing symptoms of the same problem I
noticed.

A second alternative, is you may be trying to run
four sticks at PC133. I would estimate that three
sticks would be stable at that bus speed. It could
be that running four sticks at PC133, won't work
with good stability.

Since it happens after RAM count, maybe it is a
resource issue, and an overlap of resources as
set up by the BIOS.

To test for a power problem, you could try disconnecting
the hard drives and CDROM, and unplug their power
plugs. That will save a few watts. Rerun the test
and see if it gets further, and allows you to enter
the BIOS and stay there, without crashing. The
old processors might use 30W or so, so they don't
really use that much power.

Paul

> The only config that was stable for me, was 2x256MB, running
> at PC100 (because my processor runs at 100MHz as well). If
> My motherboard was P2B-S, with 440BX chipset.

Exactly like my situation. Now using 2 bars of PC133 SDRAM.
Adding one more and gcc under linux would crash compiling
a big program....

Guess I have to settle for that... is this a limitation
of i440BX?

--
@~@ Might, Courage, Vision, SINCERITY.
/ v \ Simplicity is Beauty! May the Force and Farce be with you!
/( _ )\ (Xubuntu 8.04) Linux 2.6.24-16-generic
^ ^ 12:37:01 up 46 min 0 users load average: 1.00 1.08 1.18
? ? (CSSA):
http://www.swd.gov.hk/tc/index/site_...ub_addressesa/

Man-wai Chang ToDie (33.6k) wrote:
>> The only config that was stable for me, was 2x256MB, running
>> at PC100 (because my processor runs at 100MHz as well). If
>> My motherboard was P2B-S, with 440BX chipset.
>
> Exactly like my situation. Now using 2 bars of PC133 SDRAM.
> Adding one more and gcc under linux would crash compiling
> a big program....
>
> Guess I have to settle for that... is this a limitation
> of i440BX?
>

I don't know. I couldn't find any other mention of similar
symptoms. There are other issues, with some 440BX motherboards,
such as the "Photoshop bug", but I don't think my symptoms
match. The Photoshop bug is triggered by large data bursts,
and shouldn't really happen while sitting idle in the desktop.
But my problem happens in under a couple minutes, doing next
to nothing. If I use just two sticks, I can run Prime95 for
16 hours, with not one single error reported. So it is quite
a contrast in stability.

The rework for CUBX for the Photoshop bug, is shown here.

http://web.archive.org/web/200302201...photoshop.html

The CUBX is an S370 board, and the processor terminator voltage
plane, on the back of the motherboard, is pictured in that article.
When the rework is done to the P2B-LS, the fix is applied to the
slot1 processor module (since a bypass capacitor should be used
nearer to where the voltage is being used - the higher the frequency
of interest, the closer the cap should be to where the circuit is
using the current).

http://web.archive.org/web/200209021...rework-01.html

So that gives you the basic idea, of the fix for the Photoshop bug.
It is just a question, of whether what you are seeing, is the
Photoshop bug or not. Mine didn't seem to be, as mine seemed to
be more related to AGP/video card, and I was sitting idle
at the time.

Paul

"Man-wai Chang ToDie (33.6k)" <(E-Mail Removed)> wrote in message
news:483ce1ea@127.0.0.1...
>> The only config that was stable for me, was 2x256MB, running
>> at PC100 (because my processor runs at 100MHz as well). If
>> My motherboard was P2B-S, with 440BX chipset.
>
> Exactly like my situation. Now using 2 bars of PC133 SDRAM.
> Adding one more and gcc under linux would crash compiling
> a big program....
>
> Guess I have to settle for that... is this a limitation
> of i440BX?
>
I remember having a CUSL2....? Think that is right...:-). It was the same
thing. Anything over 2 sticks of ram would crash it. I think it was the last
440 board I ever had. I also think it was the first board I ever hit 1ghz
with............:-). IIRC it was a 100/133MHz board and if you had a good
100MHz P3 overclocker you could run them well over 133MHz FSB.

Ed

> ever hit 1ghz with............:-). IIRC it was a 100/133MHz board and if
> you had a good 100MHz P3 overclocker you could run them well over 133MHz
> FSB.

Running a P3-700 at 933MHz now....

--
@~@ Might, Courage, Vision, SINCERITY.
/ v \ Simplicity is Beauty! May the Force and Farce be with you!
/( _ )\ (Xubuntu 8.04) Linux 2.6.24-16-generic
^ ^ 20:57:01 up 9:06 0 users load average: 1.29 1.10 1.03
? ? (CSSA):
http://www.swd.gov.hk/tc/index/site_...ub_addressesa/

> The rework for CUBX for the Photoshop bug, is shown here.
>
> http://web.archive.org/web/200302201...photoshop.html

Why modify the CPU voltages? It's a RAM problem
isn't it?

--
@~@ Might, Courage, Vision, SINCERITY.
/ v \ Simplicity is Beauty! May the Force and Farce be with you!
/( _ )\ (Xubuntu 8.04) Linux 2.6.24-16-generic
^ ^ 21:00:02 up 9:09 0 users load average: 1.02 1.06 1.02
? ? (CSSA):
http://www.swd.gov.hk/tc/index/site_...ub_addressesa/

Man-wai Chang ToDie (33.6k) wrote:
>> The rework for CUBX for the Photoshop bug, is shown here.
>>
>> http://web.archive.org/web/200302201...photoshop.html
>
>
> Why modify the CPU voltages? It's a RAM problem
> isn't it?
>

The Photoshop bug involves corruption of data on the FSB,
the bus between the processor and Northbridge. It isn't a
RAM problem in that case.

http://homepage.hispeed.ch/rscheideg...ination_issues

Paul

In article <483d56be@127.0.0.1>, Man-wai Chang ToDie (33.6k)
<(E-Mail Removed)> writes

>Running a P3-700 at 933MHz now....

So you're overclocking the processor and the i440BX chipset, as well as
loading up the memory bus with four (double-sided?) DIMMs.

Not a good combination for stability.

--
(\__/) Bunny says NO to Windows Vista!
(='.'=) http://www.cs.auckland.ac.nz/~pgut00...ista_cost.html
(")_(") http://www.cypherpunks.to/~peter/vista.pdf