Need help with overclocking Corsair Dominator 1600 MHz Triple Channel

[jareklajkosz]

Hey all,

Yeah, long story short, my Dominator set doesn't work with the 8-8-8-24 timings @ 1704 MHz, I know this because Prime95 blend fails pretty fast. I changed to 9-9-9-24 and so far it's working pretty well. I'm curious though, would it be better to downclock to what would be about 1360 MHz and using lower timings with that setting, or am I better off using 1704 MHz with the 9-9-9-24 timings? Also wondering if I can get any better timings than 9-9-9-24.. I suspect no, since Prime failed at the rated timings for 1600 MHz very quickly on two threads (I didn't let it run any longer than that, there was no point).

Some background:

I originally increased my processor's base clock to 170 to get a 3.6 GHz overclock. I set the vcore to 1.2V and the uncore to 1.275V, and the dram to 1.66V. I ran Prime for 22.5 hours at small FTTs without a single error, so I was very happy with my CPU's performance. However, like I said, I tried doing the blend test and it failed within minutes. Initially I got a blue screen that blamed the tcpip.sys driver (a problem beyond the scope of this), and when I did something to hopefully fix that, I managed to get Prime running and it failed quickly. I ran 3 passes of memtest86 without any errors, so I figured the RAM was fine (curious since memtest was displaying that it was testing the modules at the 8-8-8-24 timings @ 1700 MHz and I got no errors..). Ran Prime again, it failed again, so I switched to the new timings and as I am typing this it is running without issues.

So yeah, quick summary, am I ok to keep these timings, maybe try to go for something like 8-8-9-24 and keep this overclock, or downclock and use stricter timings? I'm especially interested to hear from people that have practical experience with this particular set of RAM, but otherwise, any and all help is greatly appreciated.

 

[jareklajkosz]

This is almost more trouble than it's worth.

memtest doesn't return any errors for the RAM modules, I tried them one by one, however, when I was doing this, I was also getting an "Overclocking failed" message. I didn't reset my overclocked settings when I was testing the RAM, so I thought maybe that was the issue, but I'm not sure. Also, now, when I try to overclock higher than 3.6 GHz (I went up to 3.8 GHz), the computer will no longer boot into Windows no matter how the BIOS is set up. I tried 1.3V vcore, 1.35V QPI, nothing worked. I noticed that when this was happening, the computer was only recognizing two of the three RAM modules. Is it possible this is a RAM problem?

I should say that while I was playing around with the RAM settings, at 1.3V vcore and 1.35V QPI at the 3.6 GHz overclock with the 8-8-8-24 RAM timings, Prime95 fails almost immediately (I get a rounding error). When I start loosening the timings, Prime95 starts running longer. I read a review that with the appropriate QPI voltage setting, these RAM modules can get near 1900 MHz with the 8-8-8-24 timings, so this is very strange to me. I haven't ever pushed my voltages to extremely high numbers (highest I've done with the vcore is 1.3V, QPI 1.35, and dram 1.66). Is it possible to damage components with too low a voltage? Currently I'm running at 1.2V vcore and 1.3V QPI, and it seems to be running fine. I'm getting some very strange problems here. Any help would be greatly appreciated!

 

[jareklajkosz]

Alright, I might have found out the problem, but I'd like some feedback. Before when I overclocked to 3.8 GHz, I had never actually manually set the timings for the RAM and the voltages because I was just testing my heatsink at the time. This is the first time I've overclocked for the sake of overclocking. I'm thinking I'd either need to increase the vdimm to get to 3.8 GHz stable at the stock timings, or increase the stock timings further at 1.66V to keep it stable as well. Unfortunately I'm not at home right now to check, but if anyone could provide some input, that would be great. Thanks!