Memory Channel Temps are Crazy


Memory channel temps went nuts today (listed as memory channel 0 and 1 rank max). Showing 125F idle, quickly hits 175F at load (Prime 95). Room temp ~60F. Never got this hot before (maybe ~130F max before, I think). All other temps are normal. CPU and GPU ~90F-150F idle-heavy sustained load.

Not sure if this is related to the new release or not. I just downloaded it a couple days ago, so it's seems possible. No other hardware or software changes except installing drivers for a flash drive.

Ran virus scans. Checked event viewer. Ran WD Data Lifeguard. Sfc/scannow. Checked fans. Reseated dimms. Checked DRAM and other voltages. Ensured fans were working. Checked memory temps with infrared thermometer; 80-100F, heatspreader not warm to the touch during 15 min of Prime 95 blend. 3 passes with memtest. Ran chkdsk. Multiple restarts. Nothing abnormal. Problem persists. Deleted and acquired latest hwinfo64 portable again, same. Tried latest x86 portable, temps static for the memory channel. Tried latest beta portable, temps work again, still high. Wish I could find the older version; I'd like to retry it.

Converting video with VLC and putting on newly installed flash drive. VLC would not play created file. The loading bar in the opened folder just kept creeping slowly forward. Upon restarting the file viewer, it would not show anything except refreshing icons (eg; c, a, drives, favorites, showed with no data, with the refreshing symbol next to them). Issue did not reoccur. Restarted PC; that's when I noticed mem. channel temps were nuts.

CPU: i7 2600k 4.5ghz (no issues. OC for 6 months or so. P95 stable)
GPU: GTX 580 3gb (overclocked)
MOBO: ASRock Z68 Extreme4 (not gen.3)
RAM: 2x4gb GSkill Ripjaws X 1600mhz C8 (XMP) 1.5v (memtest stable)
HDD: WD Caviar Black 1tb
PSU: Seasonic 850w
OS Windows 7 Home Premium x64
COOLING: Xigmatek Gaia push/pull. 6 ~40cfm case fans (2 front in, 1 side in, 1 rear in, 1 bottom in, 1 top exhaust). Aerocool XVision fan cont.

I can't help but be paranoid that something is very wrong. Memory channels denote the CPU on die memory, not the RAM dimms, right? Any opinions or suggestions would be greatly appreciated.


Well, the fact is that I don't exactly know what these temperatures mean, since the documentation is not clear about that.
It might be just some sort of virtual temperature, which is used by the BIOS/CPU for memory throttling. That's at least my assumption, because I don't think this is a real temperature of memory modules (which don't seem to contain a thermal sensor - ODTS). Otherwise you'd see DIMM TS module temperatures.
So I wouldn't care much about that.. I'd maybe suggest a full power off-on cycle, or BIOS check/update.
I greatly appreciate the honesty and the quick response.

Memory channel 0 and 1 wouldn't be the on-die CPU memory controller overheating, would it? Wouldn't make much sense I guess, seeing as the rest of the CPU is far cooler, and its not like I'm pushing crazy DRAM voltage/Ram overclock.

I did an off/on cycle. I hesitate to do a BIOS update, as its makes drastic changes for Ivy Bridge support and requires IME driver updates, and etc. etc.; a pain and not always successful. Still, might try it.

Is an older version of the program available? Would be interesting to see if there's any change.
No, this is certainly not the IMC temperature, because as you say, the CPU Package temperature would be higher then as well.
It's a quite common practice, that the system maintains some virtual temperatures for components which don't feature a real temperature sensor. These are calculated (estimated) based on bandwidth and other parameters. Then it's used for throttling of components when they reach a critical threshold.
I haven't changed anything in this area since the last few versions, so that change in reading cannot be caused by the new HWiNFO version.
I think I see what you mean. I'm sure you're right that nothing has changed in the last few versions that would cause this. Odd though that I'm just noticing it, or that it just started to happen.

Apparently, this high reading is/was common among ASRock Z68 boards:

Anyway, thanks again Martin, I appreciate your time, not to mention the killer software. You've eased my mind a great deal.
You're welcome ;)
Maybe I should remove those values from HWiNFO, so that they don't confuse users...
If it might have some value in an extreme situation, it might be worth keeping.

But, because of the lack of documentation (from the mobo mfr. I'm guessing) and that it's very probably useless, IMO I'd lean towards deprecating it.

Update: I ran P95 blend for ~30 min today (added a fan and wanted to check temps), and memory channel temps hit a staggering 230F/111C and rising. System is running perfectly fine. I would think it would be unstable if any component actually got this hot.

This can be considered solved. I would mark the thread as such, but I don't see an option to do so.

I'll put up another thread if those 230F++ temps melt my mobo. ;)