It's probably not a problem, but run the Large FFTs test on your friend's machine before hand. If there's anything causing his to be unstable for whatever reason you won't want to mistake that for being your RAM. It's probably not going to be a problem, but if it is you will save yourself a lot of time and money on not buying new RAM.
I have no enthusiast friends and certainly no friends with ddr3 so i can't test out the ram on another machine but the memory controller overheating sounds correct. Very interesting problem and it explains all my problems especially the main problem with failing large fft at 1600MHz but not at 1280MHz. The options i see are buy new 1.35v or 1.5v ram or run my ram at 1280MHz. Unfortunately i don't have money to be buying new ram right now so i'll just run it at a lower speed. Also about my temps i only get to 60c average on my cores with my current settings which is stock voltage. I got to 70c average when i applied 1.2v on the cpu to test if it was lack of voltage to the cpu. Are you wanting me to rotate my fenrir so that the front is facing my ram or the front is facing my top case fan? Also it is my cores that fail prime. My ram passed 6 hours of memtest so i'm sure my ram is alright.
60*C at stock settings with a Fenrir is still too high... I would expect the stock cooler to put out temps similar. As an indicator, my i7 with HT on (jacks temps up) doesn't exceed 67*C under full load small FFT's at 3.8GHz, 1.23750v. I'm saying to rotate the Fenrir so that it's situated how it's supposed to be, then take the fan off, and put it on the opposite side so it pulls air through it and out the back. If that doesn't drop your temps by at least 10*C, then it may be time to look into either a new cooler, or lapping yours along with your processor. Out of curiosity, what are your ambient temps? Running Memtest isn't 100% guaranteed to show every error... Are you running the DOS version or the Windows version? Either way, neither of those programs put much stress on the CPU, so there is a very good chance it's the memory controller, but you can never be 100% positive.
I agree with JaredC01 temps are way too high for your default settings. I had a similar problem with my Fenrir when I installed it the first time, it turned out I had not put enough TIM on to fill all the gaps where the pipes mate with the block. I would suggest you put a little more TIM on than you usually would... Stocks about 32-34 As for orientation I also have mine exhausting to the roof fan and am also getting 70c....... but @ 100% load @3.8GHz, 1.240v - 24/7. This may come down a little as my running time increases from installation. Hope this helps and you get your memory problem sorted.
I'm using the DOS version which is installed automatically with ubuntu and added to the bootloader. I realise that my temps are very high for my system but I'm running my fenrir with the 6v power cable that it comes with so that it runs quiet and I'm using arctic silver 5 which I know has a 200 hour burn in period before my temps will be as low as they can go with arctic silver 5. I've not used my computer anywhere near that length of time yet so they could still go down. I will try changing the orientation of my cooler but I'm a little unsure how to have the fan attached on the back instead of the front? My ambient temps are most likely in the low 20's, I like a cold room and always have my radiators off and my window open with only my computer and xbox creating heat. I can also guarentee that I am not suffering from lack of paste, if anything I used far too much, I put a line on my cpu, put the cooler on, took it off to see the spread then put lines on each of the copper heat pipes on the cooler surface and placed it back on, there's no way I could've not put enough on. BTW by default settings I mean 3.2GHz @ 1.107v with 1.651v RAM voltage.
your temps for a fenrir are supremly high 70c is worse than the stock cooler could do check the air flow in your case
Hi storm, I have the same problem where my pc would freeze doing anything but the small fft test. I later found out that my corsair mem kit was manufactured around oct 2009 and was known to be from a batch which would not run anywhere near 1600mhz (normally not above 1333mhz) even though its rated at 1600mhz. I managed to get it close to 1500mhz by loosening the mem timings in the bios to 9 instead of 8. Maybe try the same and see if you have any luck, you may be able to get a little more than 1280mhz if its the same problem. If you were to ramp up the qpi volts and mem volts past recommended I think it would eventually be stable but in my opinion too risky past 1.35v qpi and 1.66v dram which is where I have mine. Oh I forgot to mention I had much more luck with a command rate of 2T instead of the defaulted 1T set in the bios.
Is there any real benefit to running the system at 6v? I realize it drops the noise of the fan, but is the fan on the heatsink really any louder than any other part in your system? If you want to go for speed, set it at 12v, if you want to go for noise, you'll have to live with the lower clock speed. Hard to have both (unless you go full blown water setup). Is your graphics card not louder than your CPU already anyway? I know it's the loudest thing I have in my system, and that's saying a lot considering I'm running 5x 120mm Scythe S-Flex E fans at 100% 12v with an Antec 200mm Big Boy exhaust in the top running at 100% 12v as well. For changing the orientation, mount the cooler 180 degrees to 'standard' (with the fan at the back of the case), then flip the fan around so it's pulling air through it. From what I can tell, the reason why the Fenrir isn't performing very well, is because there's very little direct heatpipe contact to the cores. The cores in the i5 run top to bottom (as shown on the AS5 application instructions HERE), and with the Fenrir mounted how you have it, the heatpipes are perpendicular to the cores, meaning they aren't making much direct contact. With it the 'standard' way, or 180 degrees as I said, the two middle heatpipes are making direct contact with the cores all the way across the chip. It should result in better temps... In theory anyway. As for the AS5, the most you'll see for a drop is around 5*C. I used AS5 on my previous build, and after curing didn't notice a difference... Maybe ~1-2*C, which could also be due to change in ambient temp. On that note, using too much paste is about as bad as not using enough paste. Personally I would put a bit on the cooler base and use a plastic card to fill in any gaps between the aluminum and heatpipes, and then remove the remaining paste so you're left with a flat base with just the gaps filled with thermal paste. From there, apply a thin line as the AS5 instructions indicate above. Once you have the cooler seated, twist it a few times to make sure it's making good contact. 3.2GHz and 1.107v is still on the low side for an overclock, and 70*C is still too high for those settings. Anyway, try out those changes, and if it drops temps and you can pass Large FFT's, then you know what the problem was and it's fixed. If it drops temps and it still fails Large FFT's, then you know it's a RAM issue.
The loudest noise in my case are my 23cm case fans but the only noise they make is air moving at a comfortable noise level, 6v is important and I am not after speed, I only wanted a low overclock for better performance in dual core scenario's like my PS2 emulation where every MHz helps. My graphics card is definately the loudest component in my case but when it's idle I can't hear it at all and when I play games the noise from my speakers or headphones outweigh the noise of my computer, I wouldn't mind the noise as much but I have to sleep with thing at the bottom of my bed so noise is a big factor I need to take into account. I'll change my cooler posistion today or tomorrow and for the 3rd or 4th time I only got 70c with 3.2GHz @ 1.2v not 1.1v, I get 60 max on smallFFT with 1.1v
After reading a couple of reviews on the Fenrir, there's a delta T of about 6*C between running the fan at low speed and high speed. If you want the performance difference between low and high, you could hook the fan directly up to the motherboard and control the fan speed via the BIOS. You should be able to set a temperature threshold to increase the fan speed from a set percentage... Might be worth looking into if you want the best of both worlds. On that note, after looking at some reviews it would seem the Fenrir can keep a 3.8GHz 1.4v i7 920 (probably not a D0 due to the volts) at the same temps you're getting stock on your i5. Granted I'm sure that's with the fan on high, but even with it on low I can't imagine it hitting 70*C. The last line in your last post: "BTW by default settings I mean 3.2GHz @ 1.107v with 1.651v RAM voltage." I was getting the impression that THAT was what was hitting 70*C. Just re-read your other post and noticed you said 70*C with 1.2v (which is still too high).
I've reorientated my cooler and changed how much thermal paste I used, I now have my cooler 180 degree's of the posistion it's supposed to be and the fan is sucking air through the cooler, I applied thermal paste in a cross shape and made sure that all the screws were tightly fastened untill they were firm but without enough to stress the motherboard, my temps are 55-60c on my cores with my first core being the hottest (it always was) but this time my third core is my coolest core instead of being the second hottest like before. I'm currently running largeFFT with 1600MHz ram to see if the memory controller is cooler than before. I also took some pictures of how my system looks: Aswell as a screenshot of my temps: I'll update to let you know if prime fails or not.
is there any reason you are running 3 sticks in a dual channel system? does it fail with 2 sticks? nice PSU btw.
Yeah at the time it was going to cost me £90 for 2x2GB but aria had the patriot's on offer for £105 at the same speed/timings so I went for them and thought I'd see if they run in dual channel like some motherboards let you with 3 sticks and my motherboard runs them in dual channel according to my motherboard when it posts and CPU-Z. I don't know if it fails with 2 sticks as I've never ran anything but 3 sticks but if it fails prime (which it hasn't done yet even after 40 minutes) then I'll take one out and try. And thanks about my PSU , the fan doesn't even spin for my system Here's a validation of CPU-Z: http://valid.canardpc.com/show_oc.php?id=1131549
Temps improved for all except my first core, my first core at the moment is 4C higher than my 4th and 5C higher than my second and third cores, I imagine the only way of improving that is to lap my CPU which I'm really not ready to do, I don't think I have enough experience under my belt to start lapping nevermind the money to replace anything I may break. Prime's still going well with 3 DIMMs at 1600MHz after 3 hours so I'm guessing I have better contact with my memory controller so I may have fixed the problem thanks to everyone's help
Glad to hear its still going strong... seems the memory controller may indeed have been overheating. As for lapping your processor, its really not difficult. The main thing is to take your time and not to apply any pressure to the processor, letting it use its own weight. I made a thread a while back ill see if i can dig up that showed how i lapped my E8400. Edit: HERE ya go.
Well it's been 4 hours and still no errors so I'm happy to stop the stress testing, I reckon I'll see some further temperature improvement once my AS5 breaks in again, but untill then I'm happy with my clock speed and my temperatures aswell as the all important noise (although the idea of watercooling sounds nice). Thanks a lot for all the help everyone and I'm very surprised that it was something like a memory controller overheating but oh well, I'll add it to my memory bank
Glad everything's running smoothly again! Also glad to be of help... If I hadn't read about someone losing their memory controller in a 920 due to overclock and stress of 12GB of RAM, I probably wouldn't have thought it could be the memory controller myself.
