Hi, having some trouble with my overclocking, basically I overclocked to 3.2GHz on stock voltages by only base clock overclocking and I stress tested using Small FFT's since it is more stressful to the CPU. I ran it for 10 hours with no errors but last night I tried large FFT for the hell of it and 2 of my cores failed within 30 minutes. No matter how much I try a core will always fail within 30 minutes. Since small FFT stresses the CPU more I got thinking that it is either my motherboard or my ram causing the failure but I just stress tested my ram with memtest for 6 and a half hours with no errors. Which leaves the motherboard, should I need to increase the CPU VTT and PLL for the 160MHz Base Clock I'm using or do I just need extra voltage for my CPU? Everything is at stock apart from: 160MHz BCLK DRAM Timings (copied what my X.M.P information was and set them manually) CPU voltage 1.107v (auto is 1.106) VTT voltage 1.122v (auto was 1.097v) PLL voltage 1.830v (auto was 1.800v) DRAM voltage 1.651v Thanks for the help in advance. EDIT: Just saw this on another thread Could that be my problem? Or is that just limited to LGA1366?
more voltage on VTT, 1.20v sounds good. im running 181 with 1.25v. i always put PLL to 1.9v, it magically made my 4GHz overclock rock solid. so now i'm always using 1.9v. it's worth a try.
I raised PLL to 1.9, set VTT to 1.2 and it failed within 30 minutes, raised VTT to 1.25 and it failed, raised CPU voltage to 1.2 and it is STILL failing within 30 minutes, my CPU temps even raised by 10C minimum by going to 1.2v but it still keeps failing >.< What else can I do to try and fix this?
I think you need to wind down the clock and get it stable, then start working up again a bit more slowly. If you've already tried that you might have hit the limit of your hardware if voltage tweaking isn't proving useful.
3.2Ghz and 160 Bclk is by no means high. have you tryed the most common setting? eg, 3.8GHz 181x21 at something like 1.3v. or try to find out each component's max first. first do your motherboard's Bclk, what's maximum for that and what voltage is required. then work your way around your system components.
That's the thing I went up in increments of 5MHz and ran 1 hour of small FFT before increasing again. I thought my system was completely stable untill I decided I'd give Large FFT a whirl. It must be my motherboard that's limiting me because it isn't my ram and my CPU is perfectly fine with the more stressful test. I'm thinking either my motherboard or northbridge so I'll try large FFTing at stock to see if it could be a hardware problem and I'll try setting the minimum multiplier on my CPU to find my maximum BCLK. It's gonna be a long night. I'm extremely worried at going to 1.3v, my CPU already hits 70c at 1.2v which I'm guessing is down to my thermal paste still being in it's 200 hour burn in period although it could be the restricted airflow from having my fan pointed down thanks to my tall ram I'll try finding my max BCLK then post back.
Right I have just done a 12 hour large FFT stress test at 160MHz BCLK with 1.2v VTT and 1.9v PLL with my CPU multiplier at 15x and RAM at 1280MHz and got absolutely no errors, so I'm going to try setting my CPU back to 20x and keep my ram below 1333MHz and see what I get.
Interesting turn of events, 3.2GHz didn't get any errors even after 2 hours while memory was at 1280MHz but when I increased the memory back to 1600MHz and stress tested 2 cores failed in 20 minutes and 1 core failed at 40 minutes which means that I have a problem with either my RAM or I have a problem with my memory controller on my CPU? Any advice anyone?
and that was at 1600mhz? tbh id be tempted to nudge the VTT up till it was appraoching the same as the vcore
You don't mension cooler but 70c on air is high Espicially as your overclock is very little above stock
Sorry, haven't edited my signature, it's a titan fenrir with the fan pointing at my graphics card because the cooler protrudes over 2 ram slots on my motherboard and the second slot is DIMM1 so I mounted in facing the floor.
i know its offtopic, but did you think about having it setup front to back but with the fan in pull rather than rush?
pull air from top of graphics card and then blow out of the top? do you get crashes in every-day stuff? if not, just forget about getting it that stable.
My titan fenrir is pointing downwards with the fan pulling air from the bottom of the case and blowing out of the top, the GPU is in my second PCI-E slot as my first is a PCI-E x1 slot so there is a gap between the fan and the GPU. I get absolutely no crashes doing anything, I've played games for hours at a time and my sister has been playing the sims 3 on my computer a lot lately too completely disregarding me saying "It's not stable, don't use it" and hasn't complained of any crashes. Only thing I've had problems with is when I was playing Street Fighter IV with my cousins, every now and then it would freeze for a few seconds untill I got an error saying the display driver stopped working and then it would unfreeze and still be playable, I'm putting that down to driver errors more than anything though.
Two things... 1. Put that Fenrir back in the normal orientation with the fan in a pull setup! 2. The memory controller is overheating, or the memory is bad. There's no thermal sensor or regulator for the memory controller in the chip, so it's possible to fry it without knowing until it's too late. If you're hitting 70*C on your chip with a Fenrir then there's a good chance your memory controller is getting too hot and failing. If you've got a friend with a newer machine, test your RAM in his computer running at proper speeds in Prime Large FFT's. If the cores don't fail, you'll know it's your memory controller. If it DOES fail, you'll know it's your RAM. If changing the orientation of the Fenrir doesn't drop your temps to a 'good' running temperature, it may be time to look into lapping.
