I have a new machine I am building. I have installed two 512mb sticks of DDR. Everything runs fine with no errors if I run only 512mb of ram. When I install the both of them I get random blue screens on Win2k about memory issues, physical ram dump ect… reguardless of the app running. I have tested each stick and they are working fine as singles in the machine. This problem only surfaces when running both sticks together. They are a matched pair of Mitsubishi sticks. It would seem there is something in Win2k that is not allowing any more than 512mb. Any ideas for resolution?
Rhetorical question: Why is it that everyone instantly blames the OS when the hardware/firmware acts up?
Question for you to actually answer: Would you please post your hardware and firmware stats such as: motherboard, memory's CAS latency, special memory information (registered and/or ecc, etc.), BIOS settings regarding your memory configuration, number of case fans running and their placement, temperature readings if you have them, ambient room temperature, etc., etc., etc.
Its not that I blame the OS just I have no other idea what may be causing the problem. I have been reading some of the tech articles on MS and for some items that reference manual settings that must be made in the OS; Pagefile, Vcache ect…
For your specs here they are: Asus A7V-266E, 1700XP, PC2100 DDR, mobo temp 27’c, CPU 42’C, All 9 case fans in a LianLi PC70 filled with Swiftechs. Ram is currently configured with the auto settings (SPD) of the mobo. Just have the base OS installed with mobo drivers only. No PCI cards and cheap AGP card just to get started running on the OS generic video drivers.
Check in the BIOS and see if you have memory interleaving (something like that) turned on. If it is on, try turning it off and see if that helps things... From what I understand, that setting controls whether or not the system tries to really take advantage of the second stick of ram by accessing both at once (in laymens terms). With it off, it treats it more like it is all 1 stick of ram... accessing only one at a time... (I could be completely off on that explanation... but I think the general gist of it is right).
Therefore, if you turn off interleaving and things become stable... it is the more agressive setting allowing memory interleaving that is screwin ya over.
Thanks for the suggestion. The Interleaving setting is there and was set to auto. I set it to none but it did not work completely. It just take longer to fail. I know this is just a configuration issue with the BIOS or Windows.
Bum_JCRules, Win NT and 2K are 'notorious' (though in actuality, <i>most</i> people blue-screen <i>less</i> with 2K than with 98SE or ME) for memory-related blue screens because the API is designed to re-use recently released memory first before assigning memory that hasn't been touched yet. This in theory improves performance as theoretically the recently released memory is physically closer to the memory that a program is already using. The reason for 99.99% of those blue screens is because whoever wrote the program that you are running freed up the memory and then tried to access it again after it has been freed. In Win95, 98, and ME this wasn't as much of a problem because Windows would instead use a new section of memory before using a released section of memory, thus allowing this freed section of memory to go untouched for long periods of time.
So most of the time the resulting memory-related blue screens that you see are in fact <i>not</i> a bug in the OS, but instead very bad programming on the application side which is covered up by the 'home' use OSes with their inefficient memory handling but revealed by the 'professional' use OSes and their enhanced memory handling.
Michael_N2QJI, do you know if your RAM is CAS 2, 2.5, or 3? Somtimes automatic detecting fails. And sometimes memory doesn't meet it's specifications perfectly and you need to slow it down a little.
Also, how many watts is your power supply, and do you have the Asus probe software installed and/or can you otherwise monitor for voltage fluxuations? And you said it's a cheap AGP video card. What video card and do you know if it tries to reserve system memory for it's own use?
OldBear, you have a good point too about using the latest patches. Michael_N2QJI, have you installed the most recent (but non-beta) drivers for, well, anything and everything?
Let me just say that you are a wealth of knowledge.
I have never had stack dumps on 98, 98SE, and I don't use ME. However I still blame the OS. It is more fun too even though it is some glitch in the application's programming relating with the OS and the hardware in turn.
It is kind like arguing with you TV even though it is not the TV that is speaking but the idiot that you're watching saying something stupid or when a referee makes a bad call.
It happened to me last night, not the proverbial memory stack error, but conflicts with Quake II.(Anyone remember compatibility-mode vs. performance-mode?) Yes, I was bored and decided to reinstall Q2. Boy are the graphics terrible. I think it is too funny that we used to think that that was the coolest thing on the planet. Well, until Half-Life came on the scene.
Thanks for the replies. To answer some of the questions: I am running the 1.06 bios, I have NOT installed the service pack 1 or 2 yet. I have been getting the stack dumps on base win2k no other progs are installed. No cards are installed yet. The video card is a savage3 I think with 8mb. This will be replaced with a Radeon AIW 64mb card sitting in the box on the floor. I like to get things running and the base OS installed before I add the cards. I will make a base image before getting crazy. Yea, I BACK UP MY HDs, PowerQuest Drive Image Pro. The power supply is a Enermax 431 watt (EG465-VE). I have not installed Probe yet but I have two Digital Docs installed for voltage monitoring and fan monitoring & control. I do not use the temp probes or temp monitoring with digital docs. All voltages are very stable as are the fans when I turn the on. As for the CAS, This is Mits ram that the BIOS is reading a 2.5. I have done some searching on web about the chips on the stick and it appears that it is indeed CAS 2.0/2.5. I have also found a direct Mits rep that this RAM is OEM ONLY for IBM servers and the chips are on IBM laptop mobos. I have the complete spec sheet on the sticks and the chips. Any other specs to assist in my memory dump issue?
I am still having the dump issue when I install more than 512mb of ram. I have configured the ram to the spec sheet and still have the dump. I reloaded win2k last night after a low level format of the drive. Same results, stack dumps. I want to run more than 512 of ram, so any advice wouyld be great.
Well, everything sounds reasonably in order. You have the latest BIOS. The power supply should be ample. The voltages are steady. The exteme number of case fans should be keeping anything from overheating. It leaves me with only a few remaining thoughts:
If you haven't done so yet, try moving the DIMMs around to different slots.
Try upping the CAS latency to 3 to see if that improves stability.
If your RAM supports ECC, try enabling that.
One last thing that caught my eyes as unusual. You said:
I have also found a direct Mits rep that this RAM is OEM ONLY for IBM servers and the chips are on IBM laptop mobos.
This seemed an odd statement to me since what you are describing sounds more like a SO-DIMM than a DIMM. If that were so though, then the pin count should be incompatible. I haven't worked with SO-DIMMs, but I would hope that the chip wouldn't fit into a regular DIMM slot. Do you know the pin count and voltage of your RAM?