Sorry forgot to add, CPU cooler like this one: http://www.svc.com/fan-xtrem2.html or better probably (the thermal compound that comes with that one is fine).
I have recently experienced a similar issue on my PC, and the ultimate source of the problem was unexpected.
Like the OP, the PC was freezing up in Windows - mouse and keyboard would not respond. I could not go a full day without at least one freeze-up. At first, the problem was assumed to be in bad RAM or overheating or possibly a virus, but that turned out not to be the case. Ultimately the source of the problem was determined to be a faulty hard drive. The faulty drive was a SATA drive connected to the motherboard, but was not the drive the OS was installed on. With that drive no longer connected, the freeze-ups disappeared.
This may not necessarily be the problem the OP has, but something to consider as it was not intuitively obvious to me that a hard drive could be the source.
No.
S’up.
Just to be clear, 68C is a perfectly acceptable load temperature. The people wanting you to do heat stuff are higher than a Georgia pine.
This was common with IDE drives (don’t know why). Doesn’t seem to be as much so with SATA but obviously still a potential problem. Thanks for the info.
I’ve been doing some OC’ing lately after a hiatus of several years. This is how you find out if a system is stable - especially when you have intermittent faults that follow no observable pattern. While OC’ers are clearly lunatics, moi included, these tools do have other diagnostic uses.
I agree with Palooka, 70c is on the high side but shouldn’t cause the symptoms you see - . One memory error is one too many, start with that.
ETA: after a bit of Googling I see the max temp spec for the Q9400 is just under 72c. So in theory we’re right and that temp is barely OK, but perhaps a marginal CPU could be misbehaving at that temp, so do try to cool it better (as others have suggested). Simply having the case off and a desk fan pointed at the CPUs would be a good test to see if you get lower temps and still get the hangs with those games.
I’m posting from work at the moment.
It turns out that the RAM modules (two of them) are fried. I was at the vendor’s site doing an exchange, when the staff tell me they couldn’t even boot up their test computer with the RAM. That was surprising, as my PC did boot up fine last night before I removed them. Either I accidentally fried them or some weird forces of the universe kept them working till I pull them out.
This is why I hate working with hardware.
They probably didn’t set them for the right voltage.
I didn’t want to mention this since I didn’t think that futzing with the bios would be your idea of a good time. I had some Geil black dragons that simply would not work consistently at their highest JEDEC rating until I bumped the voltage from 1.65 to 1.7
I think there is something funky going on between the new IMC’s and the northbridge but I have no proof.
I’ve seen RAM with intermittent errors die completely once fully powered down. That isn’t out of line with the previously reported memory errors.
Why are you over-thinking this so badly?
Moi? Mainly because they simply didn’t work properly at the highest of 3 JEDEC levels on the spec’ed voltage. And I’d heard of others having the same problem.
But honestly, I think that may have been solely the result of overclocking the 3.2ghz cpu to 4ghz. I can’t say for sure that it would have been a problem at all stock (except for the dram - which HAD to be set manually - at least as to clock speed - 1600mhz).
Normally that shouldn’t matter. The bus is still running at 200mhz which is all the dram should care about. Well, that, the mem multiplier and the voltage. It shouldn’t give a squat about the CPU speed since the CPU multi runs off of the same 200mhz Northbridge bus. So for the memory - 10x, 15x, 9999x - doesn’t care - or so we all believed.
But there is definitely something going on with the integrated memory controllers (IMC) on the Phenom II X6’s. There’s some kind of feedback coming back across the NB to the IMC from the memory. I don’t know if the IMC is doing some kind of stabilization, fine tuning or whatever, but it shouldn’t make a diff and it definitely does.
Anyway, since the cpu multi isn’t supposed to influence the dram, I would guess that it could be a problem even at stock. Whether it’s likely or not is something else I suppose. But when the only idea you have left is a hammer, everything looks like a nail - right? :smack:
Not sure if this is directed at me, but if yes, well, think from my point of view. I am unaccustomed to and wary of dealing with hardware. So I got a little worried that I have zapped something or done something wrong. It took me five or six tries just to get the new RAM in.
Anyway the new RAM are in, and Mass Effect 2 gets past its spots where it usually crashes. Of course, this is not enough for me to draw a definite conclusion that all is well. I guess I’ll take some time off for gaming and see what happens. Another option is for me to run stress tests, but with my CPU temperature almost nearing the max, and me in a tropical region, I guess I better not.
As for the PSU, here’s what it is: GE-P450P-C2, so yeah yoyodyne, it’s that one.
The next thing I may want to tackle is the temperature; Is there any fan that I can add to the motherboard without changing the casing? How often do I have to apply new thermal paste for the CPU and the heatsink?
It was directed at dzero’s bragging-disguised-as-advice.
You don’t have a heat issue at all. 72 C isn’t the fail temperature of your CPU, 100 C is. You’re way under that.
Not that you’ll believe me, but I normally go out of my way to be as helpful as possible and do it with as little ego as possible. If I pull something mildly esoteric out of my ass like the NB/IMC/voltage business, it’s because I really believe it’s a possibility.
However, I don’t like people to think I’m talking down to them either. So I won’t always try to squeeze in a course on microelectronics in along with my posts. Being pedantic is also rude. I realize that in general I will be talking down to one part of the audience and over the heads of the rest. There is no middle ground that I have been able to divine. In cases like this, I try to gage my comments to a level that I think is good for the OP. But again, I miss more than not. Knowing that, I would rather having them ask for clarification than think I’m a dick for talking to them like a 3 year old.
so you see, it’s something I’ve devoted a lot of thought to. It might be nice to see something similar from you the next time you chose to criticize me for no good fucking reason.
Oh, just occurred to me. Do you think I was bragging about the 4ghz oc? That’s ka-ka dude. It’s not something it would even occur to me to brag about because virtually ANYBODY can do it with the right mobo - for real.
I would like to thank everyone for their help. I don’t have a finger on how things are but I do appreciate all advice given. Yes this is me trying to ‘cool’ the thread.
So far system works fine; games that were previously unplayable are doing good.
It’s all good. I just assume people think I’m a douche so I can never really be insulted, just pleasantly surprised. 
I still think you should open the case and have a desktop fan pointed at the innards, and see how much the temp drops. Indeed having new RAM I’d like to see how much it drops just from that.
shrug competing cites: CPU World says 71.4 . I know which one I’d trust. IMO 100c is absolutely insanely high for a laptop (built to accommodate higher temps) let alone a desktop.
Make sure you wear an anti-static wrist strap before touching any bare electronic boards or RAM sticks. You can get a cheapie wrist strap at Radio Shack. Make sure the computer is unplugged (not just turned off) before opening the case and attach your wrist strap to the case. Handle everything carefully and you should be fine. You can google “ESD precautions” for more info about how to avoid zapping things.
They make fans that fit into disk drive bays. You’d have to look at the design of your case to tell if that would be of any benefit to you.
You have to apply new paste any time you remove the heatsink from the CPU. Thermal paste isn’t a standard maintenance item though. You don’t have to replace it periodically, if that is what you are asking. If you don’t touch the CPU or heatsink then you can just leave it alone and forget about it.
100 C is where the CPU will fail pretty much instantly. That doesn’t mean that you can run the CPU fine at 72.
As a rough measure, every 10 deg C above room temperature cuts the expected life of a chip in half. This rough rule of thumb is a bit inaccurate on the low and high ends of the scale, but otherwise it’s not bad. Your reliability numbers really don’t start to take a nose dive until you get above 45 deg C. So, ideally, you want your CPU to have a max temp around 45 to 50 deg C.
Most computers these days run hotter than that. Manufacturers don’t mind shaving a few years off of the expected life of the CPU if they can save some money on the heat sink and fan, and let’s face it, modern CPUs generate a LOT of heat. It gets downright difficult to remove that much heat from a modern CPU. So these days, temperatures in the 50 to 60 deg C range aren’t exactly uncommon.
72 is too high though. Sure, it will run, but I’d be surprised if it lives longer than 2 or 3 years.
Before we get into dueling cites (since two conflicting ones have already been given), my cite is that I am an electrical engineer with 22 years of experience, the last 12 or 13 or so of those as being one of the lead hardware and software designers for a company that makes industrial computers and controllers. I was trained on how to do reliability analysis using U.S. Department of Defense training manuals, and I currently design rugged industrial computers for a living. That’s my cite.