Sign in with
Sign up | Sign in
Your question

Confusing memory issue

Last response: in Memory
Share
August 18, 2005 12:23:27 PM

Ive started to have the occassional blue screen and lockups upon booting so i went and started memtest with both 2x512 meg sticks in dual channel mode. This came back with quite alot of errors so i thought i'd found my problem. However, when i test the sticks singularly they both show up as fine. Ive tried the memory together in single and dual channel in different dimm slots (my board has 4 Abit-AN8) to attempt to rule out faulty slots. I always get errors when running the sticks together whether single or dual channelled but if i run them seperatley they dont show any errors in memtest86. This is confusing me, what could it be?

More about : confusing memory issue

August 18, 2005 3:06:24 PM

2 sticks put more stress on the memory system than one.

You can try bumping the RAM voltage a notch, that may fix it right up. (what's it at now and what's the rated volts of the RAM?)

Also how good of a PSU do you have? If you have a cheap PSU or its barely powerful enough, the 2nd stick of RAM could cause it to go unstable (I know - a few watts isn't much - but if your PSU can handle 300w - not that its' a 300w psu, but that its only capable of steady power up to 300w - and the extra stick pushes it to 302w...).

Mike.

<font color=blue>Outside of a dog, a book is man's best friend. Inside the dog its too dark to read.
-- Groucho Marx</font color=blue>
Related resources
August 18, 2005 7:48:58 PM

Ok i'll switch out the PSU at some point in the next couple days since i have a spare laying around. Also i'll post up my full spec.
AMD64 3200+ venice
ABIT - AV8
2x512 meg ballistix ram. at 2.8v (as on their website) also at 2-2-2-6 timings.
Geforce 6800 GT
SB audigy 2.
3com LAN card.
480W Thermaltake butterfly

I did just stroll onto the abit website as advised by my brother for a bios update. The following were included since the revision number of my BIOS:

Fixed the issue that with certain VGA card does not work with four 1GB RAM modules.
DRAM Command Rate, Row Refresh Cycle Time, Read to Write Delay now work as intended.
Recalibrate CPU temperature.
System now can resume from S3 without problem
Update code base to 5/31 award code
Cool n' Quiet now works as intended.
Disable Cool n' Quiet function when the multiplier is manually changed in the BIOS.3
BIOS compiled date: 07/04/2005
Update MCT2
Support dualcore cpu
Fix "CMOS Checksum fail message" when select DRAM Timing value.
Delete "workstation" at POST
Full screen logo
Fixed "can not load SATA , LAN ROM issue"

So will go off and do some more testing, perhaps the RAM timing issue might be something to do with it ;) 
August 20, 2005 1:32:07 PM

In an attempt to keep my post alive i'll post further findings.

Someone suggested it may be the power supply. Although i dont have a top of the range PSU i doubt this to be the case as the memory always fails on the same address (at least the first failure it gets).

I upgraded the BIOS on the motherboard and this seems to have taken the amount of memory errors down but then again i havent tested in dual channel mode for hours since i did so im not sure.

The memory still works fine in single channel but with both sticks in (either dual or single channel) it still errors at a specific point.

I noticed that the sticks are hotter when there is more than one in (quite markedly hotter) so i lowered the voltage and timings. This allowed the memory to get through an entire two passes in dual channel mode without any errors in memtest86.

So now i believe it is a heat issue, any thoughts?
August 20, 2005 6:41:07 PM

I cant see what else it is.. Im going to get the memory checked in another machine soon but im just thinking theres a chip on one of the sticks which is on the edge and is overheating in dual channel mode
August 22, 2005 7:03:40 PM

Have you tried to manually relax the memory latency timings in the system's BIOS?

<b>EDIT</b>
If you suspect problems with the memory, its probably not a good idea to increase the voltage. Increased voltage can significantly decrease the lifetime of your memory modules. Better to relax the timings until your machine runs stable. Then, if you find the modules unable to perform as guaranteed, its likely you are entitled to RMA (allthough it may just be your particular setup that can't handle the memory).

<font color=blue>******
<font color=green>"My Memory is not working today"
<font color=blue>AIM SoCaliCrowley<P ID="edit"><FONT SIZE=-1><EM>Edited by BrentUnitedMem on 08/22/05 12:11 PM.</EM></FONT></P>
August 22, 2005 7:07:20 PM

I tried relaxing the memory timings down the the SPD which detects 2-2-2-8 instead of the 2-2-2-6 timings its meant to run at. I also lowered the voltage to 2.7 with these timings and guess what? No errors in memtest in dual channel mode..
Im confused.
August 22, 2005 9:33:01 PM

2-2-2-8 and 2-2-2-6 are not significantly different. The values are usually written in order of importance, CAS Latency (the first value) is the most significant.

Are you still getting errors when you run 2-2-2-6 under the new BIOS update?
Did you manually adjust to 2-2-2-8 or was the BIOS set to AUTO?

Blue screens upon bootup usually point to the PSU as the culprit. You can use a program called <A HREF="http://mbm.livewiredev.com/download/MBM5370.exe" target="_new">MBM5</A> to monitor your temperatures and voltages and other things.

Also, the VIA chipset might be sensitive to the timings, as wusy pointed out- certainly, two modules will be less stable than one, and more complicated to address.

<font color=blue>******
<font color=green>"My Memory is not working today"
<font color=blue>AIM SoCaliCrowley
August 23, 2005 2:41:32 AM

The machine has only locked up on boot after a BSOD, it seems to do it every time after a BSOD for some reason. I have yet to have a BSOD on boot, i have however had one after loading a game straight up after boot.

I've had some official word from crucial about what timings would be best to relax it to and also what voltage.

The new bios update seemed to make little difference.
Also the board *should* be able to handle these memory timings as i had around 8 weeks of trouble free gaming before this problem arouse. If it was an inherent problem i would have expected it immediatley.
August 23, 2005 2:53:41 AM

YOUR PSU IS PROBABLY THE PROBLEM. I have tested the rails on more than one of the thermaltake 480W PSU's and the 12V rails tend to fluctuate to much. I know your determined that its not, but at least go into BIOS and watch the +12V. 3.3V and 5V rails for at least one minute each. Report back the highs and lows for each of these rails.

ASUS P5WD2 Premium
Intel 3.73 EE @ 5.6Ghz
XMS2 DDR2 @ 1180Mhz

<A HREF="http://valid.x86-secret.com/records.php?PHPSESSID=792e8..." target="_new">#2 CPUZ</A>
SuperPI 25secs
August 23, 2005 4:10:19 PM

You're probably right, i am pretty determined its not the PSU but i'll swap it out soon enough. Ive already checked the rails in the BIOS, there is no amazingly fluctuation. I'll post the high/lows
Also i was unable to check the 12v in the bios and had to use a multimeter which showed no large fluctuation either (crappy abit motherboard)

NB core: high/low 1.24 / 1.24
DDR: high/low 2.83/2.82
CPU: hig/low 1.43/ 1.39
ATX 3.3 high/low 3.42 / 3.40
ATX 5.0 high/low 5.24/ 5.19
ATX12v - No noticable fluctuation from 12v using multimeter

So thats why im doubting its a voltage issue, it could of course be a poweroutput issue but im unsure how to check that.
August 23, 2005 6:07:10 PM

Use MBN5 from an earlier post to check the voltage lines from your PSU.

I agree with mozz seems like a PSU problem. It seems like when you start playing the game the PSU peaks and causes your system to crash.

You may also want to make sure your drivers are up to date, including your VGA driver.

Quote:
i had around 8 weeks of trouble free gaming before this problem arouse

The only difference now is the memory?

<font color=blue>******
<font color=green>"My Memory is not working today"
<font color=blue>AIM SoCaliCrowley
!