RAM Problem with Gigabyte GA-BX2000+ Motherboard

[mlv]

My homebuilt computer has a Gigabyte GA-BX2000+ motherboard, Intel 440BX
AGP chipset and Intel Pentium III processor 600MHz. It currently has 2 x
256MB (total 512MB) of PC133 SDRAM fitted and is running Windows XP SP2.

IIFC, the settings are:

Main Clock 133MHZ
PCI Run 33.3MHZ
Bus Speed Multiplier x 4.5

I decided to increase the RAM up to the motherboard's maximum of 4 x 256MB
(total 1GB). I seem to remember from when I built the computer that the
Gigabyte GA-BX2000+ motherboard is a bit fussy about RAM and that it's best
to use identical, batch-matched memory.

After some research, I purchased four sticks of Crucial SDRAM PC133 CL2
non-ECC 256MB memory. The manufacturer's part number is:
CT32M64S4D7E.16LTG. I checked with Crucial, and this is the recommended
memory for my motherboard. Unfortunately, the motherboard won't have it!

After fitting the new memory, the first boot-up reported 131,072K of Ram
(128MB), which was a bit disappointing as I'd just fitted 1GB.

After a re-boot, 262,144K of RAM (256MB) was reported.
After a second re-boot, 524288K of RAM (512MB) was reported. I figured a
couple more re-boots and the system would find the whole 1GB!

The next re-boot also reported 512MB, whilst the one after found 786432K
(768MB).
The next re-boot found 917504K of RAM and then the computer locked up solid.

I have reverted to my old 2 x 256MB SDRAM and the computer runs fine.

Is there anyone familiar with the Gigabyte GA-BX2000+ motherboard who can
suggest what the problem might be? Is there something I need to set up
before the MB will run 1GB RAM?

Thanks for any help.
--
Mike

[flambe]

It is possible that some or all of your new ram is bad, the memory
controller on your motherboard can not really handle all four slots (a
common problem) or you don't have enough stable power to run all four
memory slots.
I would tediously check each stick one by one in each socket to see if the
computer boots.
If the ram sticks and all the slots are ok then I would suspect a power
supply problem on your now ancient machine.

[Paul]

You can run four sticks at a PC100 rate.

You can run three sticks at a PC133 rate.

Those are approximate rules (someone in the audience will undoubtedly
report some miracle they performed with PC150 RAM).

I did something pretty close to what you're attempting. Only
in my case, I have several different FSB100 processors, and the
RAM runs at PC100 rates. So I figured it would be a slam dunk
to use the above mentioned Crucial 256MB sticks. I bought four
of them.

Now, according to the approximate rules, I should be able to use
four. The RAM is rated PC133, but because of the clocking for
my processor, it is only being asked to run at PC100. I should be
able to run four sticks at PC100.

The computer posts and boots fine with four sticks. If I don't start
any applications, and just let the computer sit there, it freezes.

The freezing stops, if I use just 2x256MB sticks. It doesn't matter
which pair of sticks I select from the set of four. In fact, with
just 2x256MB, I can run Prime95 torture test for 16 hours with no
problem at all.

Initially, I was blaming this on the use of Win98SE on the computer.
(It is an old computer after all, and not worth changing the OS.)
It wasn't until I reproduced the freezing symptoms in Linux, that
it occurred to me that this was a hardware level problem. i thought
I was running into the 512MB limit on Win98SE.

I don't think this is the "Photoshop bug". That existed on motherboards
which didn't have adequate bypass capacitance on some termination voltage.
After all, I'm sitting idle in the desktop, so there is hardly anything
going on, and not the long string of consecutive of 0's or 1's on the bus,
hypothesized to trigger the Photoshop bug.

So what would I try ?

1) As Flambe mentioned, test the sticks one by one. If you can boot
with just one stick, you could try Prime95 torture test from mersenne.org.
Memtest86+ from memtest.org is another test tool you can use.

2) For multi-stick configs, and seeing as you're at FSB133, I'd try a max
of three sticks. Test with memtest86+ first (boot the machine with it),
to verify the RAM is at least stable enough not to show any errors in
a couple full passes of the test. Then boot into Windows and use
Prime95. As a final test, use something like 3DMark2001SE, if you have
an AGP video card, as sometimes bashing the AGP port will tip a machine
over. 3DMArk2001SE has a "demo loop" you can leave running over night,
as a test.

3) If three sticks won't work, drop down to two. You might be running into
whatever problem I was seeing. With two sticks of 256MB low density, my
machine was rock solid, and will pass any test you can think of.

I feel pretty silly, buying four sticks and only getting to use two.
But that is how it goes some times. In my case, I don't think it is the
sticks, because I can use any two from the four of them, and get solid
results. YMMV, as your board is made by a different company. I haven't
seen other references to my particular set of test results.

HTH,
Paul

[mlv]

Paul wrote:
>
> You can run four sticks at a PC100 rate.
>
> You can run three sticks at a PC133 rate.
>
> Those are approximate rules (someone in the audience will
> undoubtedly report some miracle they performed with
> PC150 RAM).
>

> <snip useful info>

Thanks for your comprehensive reply. I'll work through the various
recommendations you and Flambé have made and see where it gets me.

Maybe I will be able to run three 256MB sticks at 133MHz.

The computer has always performed faultlessly with 2 x 256MB PC133 Hynix
SDRAM, which I might as well stay with if I am to be restricted to two
sticks. That will mean I have just bought four superfluous sticks of
Crucial CT32M64S4D7E PC133 SDRAM :-(
--
Mike
-Please remove 'safetycatch' from email address before firing off your
reply-

[Baron]

mlv wrote:

> Paul wrote:
>>
>> You can run four sticks at a PC100 rate.
>>
>> You can run three sticks at a PC133 rate.
>>
>> Those are approximate rules (someone in the audience will
>> undoubtedly report some miracle they performed with
>> PC150 RAM).
>>
>> <snip useful info>
>
> Thanks for your comprehensive reply. I'll work through the various
> recommendations you and Flambé have made and see where it gets me.
>
> Maybe I will be able to run three 256MB sticks at 133MHz.
>
> The computer has always performed faultlessly with 2 x 256MB PC133
> Hynix SDRAM, which I might as well stay with if I am to be restricted
> to two
> sticks. That will mean I have just bought four superfluous sticks of
> Crucial CT32M64S4D7E PC133 SDRAM :-(

What has been said previously is basically correct. However make sure
that you have the latest bios updates installed. If I recall there
were some fixes for ram recognition issues.

--
Best Regards:
Baron.

[mlv]

Baron wrote:
>
> What has been said previously is basically correct. However
> make sure that you have the latest bios updates installed. If I
> recall there were some fixes for ram recognition issues.

Thanks for the info.

I've done some searches, but cannot find any reference to RAM recognition
issues in any of the BIOS updates for the GA-BX2000+. Doesn't mean they're
not there, of course.

My Bios is currently version FC and it seems version FD was released 30
November 2001, and nothing else since.

Version FD apparently fixes "system hang on POST after enabled mouse power
on function."

[Baron]

mlv wrote:

> Baron wrote:
>>
>> What has been said previously is basically correct. However
>> make sure that you have the latest bios updates installed. If I
>> recall there were some fixes for ram recognition issues.
>
> Thanks for the info.
>
> I've done some searches, but cannot find any reference to RAM
> recognition
> issues in any of the BIOS updates for the GA-BX2000+. Doesn't mean
> they're not there, of course.
>
> My Bios is currently version FC and it seems version FD was released
> 30 November 2001, and nothing else since.
>
> Version FD apparently fixes "system hang on POST after enabled mouse
> power on function."

I'm sorry that you did not find an explanation or a bios fix for your
problem. The truth is that a number of motherboards do have problems
associated with ram. Many manufacturers stopped putting more than two
slots on their boards because of these. If I recall Intel and Rambus
were one of the early casualties of the third slot problem. I still
have a couple of terminator boards that were put into the third slot to
stop people from using it and to terminate the memory bus correctly.

My own machine Gigabyte KN400MX only supports a maximum of 512 Mb per
slot and the memory speed drops by about a third with all three slots
occupied.

I also found that I couldn't mix speeds either, usually the fastest
stick got ignored whilst the other two didn't. I ended up putting in
three DDR400 CL2.5 to get all slots working. I'm not entirely
convinced that I am getting the best performance at that!

HTH

--
Best Regards:
Baron.