Hi!

I've always had that (a little) old RAM brick in my drawer. I think it was the one my old computer (with now melted CPU ) had. Following are its specifications: Crucial, 128MB, DDR, 266MHz, CL2.5. The thing is that after getting a new computer (after that embarrassing melting accident) I had in mind putting the old RAM brick in the new computer as an additional RAM brick. That was about one year ago and I then was running 2.4-something kernel. I can't remember exactly, but I think my Slackware back then complained something about being unable to create ram disk. And I pretty much left it there, thinking that the old RAM brick was not compatible with the mobo (ASUS A7V600).

But yesterday, while I was cleaning a year supply of dirt and dust inside my computers cabinet, I got this crazy idea of trying to install that old brick again. So I put it in, and voila! It worked! I was rather amazed by that fact, but nevertheless I continued working. I assumed that since I now was running a 2.6 kernel, something must have changed there. So the system was more than fine yesterday.

Today I began my day with playing NWN. And, after about an hour of play (unsaved mind you ), the computer just hung. Nothing helped and I had to restart. I hadn't done that in several years. Anyway I rebooted, fired up nwn. Played a bit (saving regularly this time) and bam! Hung again. I was getting really annoyed and I restarted once again. Though this time I got kernel panic and a "not syncing" message. I thought that that was a bit odd. And from then on I was unable to boot at all. I first thought that something was wrong with one of the two HDDs I've got installed because on the last lines of system output I got a lot of messages about ReiserFS being corrupted (wrong length or something...).

So, after a while of pointless fiddling with the cables I got it to work. Phew, I thought, that's great. The old RAM brick was STILL INSERTED. But everything went fine, I was able to boot and all. Then I turned the computer off, went to the local store to buy some food. When I came back I was again unable to boot. Again the same not syncing and ReiserFS error messages. What the heck?

Then I removed the old RAM brick and voila! It worked at once. The thing is that the present RAM brick was of following specifications: Crucial, 512MB, DDR, 400, CL3.

It seems that the CL-value and frequency do not match. So was that really the problem? But how the heck could I still boot and run my Debian on a system with conflicting RAM bricks????

PS: I think that the crashes during playing NWN occurred because cache had filled the whole first (native and working)RAM brick. But when memory manager tried to write to the other brick, it failed because of the wrong freq.

PS2: sorry for writing a damn long essay on the subject

So, I guess the moral of the story is that using the wrong memory or defective memory is bad? Good to know.

Hehe... Well, certainly Such knowledge is so rare to come by that I chose to post it here for everyone to see and learn

No, seriously... I actually thought it more like a question... The point is that though the two bricks were not of the same type (and therefore unusable together, in theory at least), this hardware configuration actually worked for some periods just fine (and the amount of RAM was recognised by BIOS and all).

I'm just wondering how that is possible since there clearly is a hardware conflict?

Shouldn't the system be impossible to boot at any time as long as both bricks are present on the mobo?

And why the heck did I get errors about ReiserFS when RAM brick incompatibility was to blame? (and no, there were NO errors concerning RAM neither before nor after that)

If you are using >1 stick of RAM, they ideally should be identical. It is possible to run with mis-matched sticks, but that can lead to instability (as you've seen) due to the fact that the slower stick is a gating factor, and the faster stick more or less has to keep stopping and waiting for the slow stick to catch up.

Overall, it's best just to use similar RAM rather than to mix and match

Oh, thanks, J.W. That really clarifies a lot. But instability is one thing, being unable to boot is another. As I understand from what you are saying, the instability should occur while running the system. Why then do I get kernel panic during boot? No distro, neither live nor installed, were able to boot (SLAX, Debian, Kororaa, even Debian install cd). But then I replugged RAM sticks and the harddisks and everything worked fine. After a little while everything crashed again.

By definition, instability is unpredictable, and a crash could just as easily happen during the first couple of seconds of bootup as it could after using the system for a few hours. My point is that RAM is in use during bootup, and so a failure could occur then just as easily as later.

The critical fact here seems to be that prior to installing the old, slower RAM stick, your system was running smoothly, and that these booting and crashing problems only started appearing after the addition of that slower RAM stick. To me, that's a pretty clear cause and effect, and therefore if I were you I'd just put that old stick back in the desk drawer where you found it. As I described previously, pairing a fast stick with a slow stick results in a situation where the fast stick constantly has to stop and wait for the slow stick to catch up, and that can make things wobbly. From what you've outlined, and knowing there's one stick of DDR266 and another of DDR400, I'd say that the poor behavior you've seen is the result of a simple hardware (ie, RAM) incompatibility. I doubt this has anything to do with Linux or the particular distro you are running