I was just wondering what others thought about this:

Setup:
Case - Antec Solo
CPU - 65W E8400
Old CPU cooler - Arctic Cooling Freezer 7 Pro (blowing air towards back of case)
New CPU cooler - Cooler Master 212 Plus with no fan mounted
Fans - 1x 120mm Slip Stream @ 975 RPM (rear exhaust)
2x 92mm Nexus with larger series resistors (front intake)
Video Card - 8800GT w/ fanless Accelero S1 (slotted pci backplate installed for extra airflow)

I recently changed my CPU HS from the Freezer 7 Pro to the 212 Plus with no fan. The rear exhaust fan is very near the heatsink and has provided ample airflow for CPU. I thought that I had pretty good, if not excessive, airflow throughout the case.

A problem came up shortly after switching heat sinks: a week or so after the upgrade, the computer failed to restart (no post). It was late so I went to bed, figuring I would work on it the next day. Surprise! The computer worked fine the next day, pointing to a problem caused or exacerbated by heat.

A week later, the computer froze. It would not post after a restart. Eventually it did, and I entered memtest86+. Errors galore in module 1! Took out the module with the errors, and the computer started up just fine.

My concern, though, was how warm the heat spreader was. It was hold-able, but barely. And this was after several minutes of trying to get the machine running, so I imagine it was not at peak temperatures.

Is ram supposed to get particularly hot? Do you think I may have an airflow issue? The other module is still running fine, but it probably has a nicer environment now that it is all alone.

Edit: rearranged in an attempt at clarity, per comment by ces

Your facts are confusing. Please discuss each separately. The hot Ram. The working Ram. The CPU temps. The CPU cooler. Perhaps do a paragraph on each without talking about the others. Then explain the known interrelationships between these 4 items and then the suspected interrelationships. Moving from a CPU cooler to the Ram temps is confusing to follow.

Seems like your memory just happened to blow at the same time you changed your cooler. But it is confusing to follow your fact pattern.

I guess it depends on the RAM. See if yours has specification on the vendor's website. Laptop RAM can sure get hot (it often gets no airflow).

Keep in mind some DIMMs do develop a massive number of errors after a while, excessive heat or not. That's much of the rationale for ECC.

Sorry, I thought the paragraphs I had were logically structured. Everything essentially boils down to:
1) No longer have CPU fan.
2) All measurable temperatures are great.
3) RAM died suspiciously soon after removing CPU fan.
4) RAM felt very warm several minutes after turning the machine off.

Just curious if people have to worry about ram cooling assuming everything else looks good. Is it a known phenomenon for RAM to overheat in an otherwise well ventilated case? I've never worried about my RAM in the past.

I should also be more descriptive of the failure: it was catastrophic. 9/10 attempts to reboot resulted in no post. Froze once while memtest started up, and another time at the memory check on boot. It took a huge number of tries to even get into memtest. Memtest reported huge consecutive blocks of errors.

Your system setup sounds perfectly fine for airflow -- both in and out flow fans at relatively high speed (anything more than ~700rpm now seems high to me... ). I expect it is coincidental. Does your other RAM run as hot? If not, perhaps there was an electrical fault or short in the bad one that made it so hot.

Yeah, the 975RPM fan is killing me. I've been meaning to switch it to 5V and see what happens (I'm not a huge fan of series resistors, anyway, although I guess they may help on startup?).

The other RAM was just about the same temperature. I may be making a lot out of nothing. Just seemed suspicious, and wanted a sanity check

Sounds likely to be a coincident. Try buying a replacement ram.

It may be that your model of ram with your configuration just runs hot. If so it needs its own airflow. Nothing good can come from unchecked heat.

Maybe you should consider a Noctua 140mm downdraft cpu cooler.

An important distinction between heat and temperature: the fanless CPU heatsink is not producing more heat but it may be producing higher temperatures.

So, I don't think that the higher CPU temperature is causing the RAM to be hotter, because the CPU produces the same amount of heat; and the heatsink is transferring that heat into the air inside the case.

If the RAM is getting hotter when there is no fan on the CPU heatsink, then it was getting some benefit from the air flow from the fan. But not because the CPU temperature is higher. I hope this makes sense?

And if the RAM is overheating, then I think that is the RAM's fault -- on different motherboards or with different CPU heatsinks, the air flow could well be quite different, so they should not be depending on cooling from the fan.

There's also the matter of the board getting hotter, Neil. It heats the RAM (or prevents it from cooling as well as it would otherwise if you want to look at it that way).

I too believe it's coincidental if the other modules are working normally.

Something to try if you want to replace the module is to go to your motherboard's product page, it should have a list of compatible RAM sorted by capacity, speed, and manufacturer.

If the RAM kit isn't on that list then it's possible that it won't work properly; one of my customers' systems got BSODs minutes to hours after login because the RAM kit from a big name manufacturer wasn't compatible with their model of motherboard.

@HFAT: do you mean the ground plane on the motherboard heating up? I never really thought about the RAM using the ground plane as a heat sink. It does not seem like it would have a great path? But then, I don't know much about the RAM interface.

Components exchange temperature over time. There are many ways this happens starting with infrared radiation. I don't know what effects are the most important in any particular case but a hotter CPU will cause higher temperatures for everything that's not actively cooled.

The only way it cant be a coincidence is if you disturbed the RAM (a lot) when you were swapping heatsinks, or possibly disturbed the heatsinks on the RAM.

Something for you to check quickly, stick the good stick of RAM into the slot that the bad stick of RAM came out of and re-run memtest, I have seen a DIMM-slot become faulty before - yet the rest of the machine continued to work perfectly

As far as RAM running hot is concerned, memtest puts a massive strain on RAM exactly as you would expect, this obviouslty makes the RAM temp skyrocket. However you might actually have 800MHz RAM cunningly disguised as 1066MHz RAM - essentially it might be factory overclocked - hence the heatsink. A bit of research might give you an answer there - this is not of importance if the RAM works though, it would just go a long way to explain the skin-melting RAM temp. Just to clarify what others have said, RAM does not need airflow to function so long as it is not overclocked and overvolted and sold a bin-grade above its natural performance level.


Andy

I think this deserves investigation.

I think this also deserves investigation[/quote]

it'd be nice to know what ram you're running, and at what voltage and frequency.

If you were able to hold the ram without feeling pain, chances are the temperatures were not that high. The threshold of pain is at about 55 degrees, that`s certainly not hot enough to damage ram chips in such a short amount of time, on the contrary, it should be well within their limits. So, I`m with those thinking that the failure was a coincidence or perhaps the result of physical damage during the installation of the heatsink.

I would still recommend to mount a fan on the cpu heatsink if only to be able to lower the speed of your exhaust fan. In my experience, a couple of fans running at 600-700rpm should be quieter than the single ~1000rpm fan you have now.

RAM was G Skill 2x2GB 800MHz P/N F2-6400CL5D-4GBPQ running at factory defaults. So no fear of factory overclock unless it's actually 667MHz ram
http://www.newegg.com/Product/Product.a ... 6820231122

I was waiting until the weekend to play around with the other slot, etc. I actually removed the ram prior to installing the heat sink, and nothing particularly violent happened. Re-seating did not help.

I actually intend to slow the exhaust fan down to 600-700RPM if I can, but currently it has a static series resistor and 12V source.

As for the temperature of touch, I assume if I am touching it and it is not currently in use, that a significant amount of heat has already been lost. That combined with the low thermal mass of the aluminum heat spreader means 'barely touchable' seems likely a fair bit higher than 55C during operation. But I could be wrong! But I can grab a thin aluminum tray 30s-1m after it was in a 200C oven.

Hmm...my earlier post failed to...post...

Most likely: Your stick of RAM barely met timing specs with the earlier cooling setup. CMOS transistors slow down with temp rise. The change in cooling raised the temps enough so the stick no longer met specs and starting throwing errors with Memtest.

If you can underclock it, say 20%, and the errors go away...or if you artificially cool it by placing a fan directly at the RAM and the problem goes away...it's a thermal/weak spec'd stick.