I know this is a looooonng post, but PLEASE read it and help me if you can
I've been having issues with my rig, some of you have already offered some assistance - thanks for that - and it is still baffling me.
Please, if there are any masters at o/c'ing that can help me out, I would greatly appreciate it. I must be missing some important piece, which is entirely possible.
Here's the deal:
Everything, except the hard disks, is brand new, bought in the last 4 weeks. I have been building systems for myself, friends and family over the last 20 years, so I'm comfortable with all of this (pretty much I know what I don't know).
Ok, here's where the problems come in. I can't get a decent stable O/C to save my life. I think it might be driver related.
I set everything to stock, and stayed up all night Wednesday to try and put this thing to bed once and for all.
I ran the memory through it's clocking and tested in memtest86+
on each try and let it run one pass. I figured that should be fine.
To me stable means I can use it as I want and not worry about bsod's or re-booting etc... So, I figure if it passes IntelBurn
on high for 20 runs and memtest
then I should be fine for everyday use. No video rendering or gaming for me, CS4 Master Coll. and Lightroom are the most intensive programs I use.
The ram clocks in at 1680Mhz, 7-7-7-16-27-1t @ 1.75v (it might have been fine with a little less volts, but that's what I tested it at), and it held like a champ through one full pass of memtest
(25 mins. ~).
Moving on to the NB.
Here I run into a surprise. I assumed I would be able to run the NB to around 2800 or more. I have an aftermarket cooler on the NB (don't even ask how I managed it with the HT/SB connected via the pipe to the NB), and I didn't break or cut anything, it never tops about 32c, usually idling between 24c-27c and loading at about 27c-32c. I was able to get the NB to hold an IntelBurn
on high for 10 passes @ 2604Mhz by running up the speed and volts as necessary (1.325v), but it wouldn't even post past that, no matter what. Everything else was stock settings or under-clocked a touch, including the ram (ran that at 1066mhz with the same settings above except 1.65v) as I clocked each piece. However, the bios is a bit buggy at this speed. I assumed it needed a slight voltage tweaking (it did pass 10 runs of IntelBurn
on high), and I moved on to the CPU.
I have decent water cooling, it never topped 51c no matter how high I tried it (and I've tried 1.6v
). It passed IintelBurn
@ 4.04Ghz at 1.56v. Temps range, under full load, from 38c-52c, depending on how much voltage I'm giving it.
I use the CPU-VID (1st option in the bios) as high as I need for CPU voltage increases, and then add in the CPU v (3rd option in the bios, the 'offset') if I want more than 1.550.
I use the CPU-NB VID (2nd option in the bios) for the NB clocks, and have the NB core v (down closer to the bottom in the bios) up to 1.24v with the higher clocks, but I have even tried bumping it all the way to 1.4v.
Ram is anywhere between 1.65v and 1.75v depending on the speed, usually 1.65-1.7v, as I don't run it much over 1640Mhz, usually even lower, but I would like to be able to run it at 1680Mhz if I could get everything else stable enough to allow me that.
Ok, here's where the problem comes in. I have all the clocks written down with the v's needed to sustain each incremental speed, all the way up to each components max.
I set the bios to 202mhz, CPU @ 20x, NB @ 13x, HT @ 9x and the ram at 1616Mhz (1:4), 7-7-7-16-27-1t, 1.75v (probably more than it needed, but I didn't think it would effect stability, as it needed it once I got past 1640Mhz during testing (1.7v failed memtest
, and I went right to 1.75v).
I'm a bit bummed that this is all I'm getting out of the NB and the CPU, but hey, I am running a 64-bit O/S, so I am at least happy to have broken the 4.0Ghz barrier, stable, for the money I've spent building this. That was the main reason for this build, to break 4.0Ghz.
No game, it's buggy as hell in the post and crashes within 10 minutes in Windows, and I'm not doing anything more than surfing the net. I do have the video card clocked to 1005 and 1100, but it has passed Vantage
@ 1015 and 1145, so I'm assuming this shouldn't be an issue (It never goes above 57c, ever, usually staying well under 54c). And, I have taken it back to stock speeds, same issues. So, it's not the card, drivers maybe? I'm using CCC v9.9 for 7 x64.
Ok, I clock it down, and figure I'll start pushing components up one at a time and see where it's having issues. The long story short is that I can't even get 3.94Ghz, and 2.44Ghz NB stable, even at higher voltages than they passed the stress tests at!
Any time I try the NB at about 2585+ it suffers post issues, won't post, needs a few reboots to post, or if I wait, it tells me the overclock failed. Seriously, ****??? Didn't I just
test every piece individually and get higher speeds stable?
I ran IntelBurn
on 3.434 CPU (auto v), 2.02 HT (auto v), 2.424 NB (v core-down bottom-auto, and CPU-NB VID @ 1.2875 - 2.46 had previously passed at 1.25v BTW) and ram @1616, 7-7-7-16-27-1t, 1.7v. It passed fine at these settings . I then went right to 3.94 at 1.4875v (it had preiously passed 3.98 @ 1.500v). It failed at about 6-8 runs in on two attempts.
Also, just for the record, my vdrop/vdroop is about .024v, that seems like a lot to me, but not huge. Is there any way to minimize the vdrop/vdroop on this board?
As soon as I get a decent clock to pass stress tests, I'm getting random reboots in Windows (surfing the net, nothing stressing), lines across the screen reboots, no bsod's at all, and I don't know what that means.
For the Record, I have all the most updated driver installed, but none of the Windows requested drivers. I can provide versions if needed.
It passed IntelBurn
, 20 runs on high, @ 3.94 - 1.525v, 2.424 - 1.2875v, 2.02 auto, 1616 7-7-7-16-27-1t 1.7v while I was writing this. But, just now failed (twice) at 3.866 - 1.4875v, NB 2508 - 1.325v, HT 1880 - auto v, and ram 1672 - 1.75v, 7-7-7-16-27-1t.
, someone give me the magic pill for this ailing rig!
I just don't understand why it passes stress tests, and then gets all buggy in Windows, nor can I understand why I can't ever get a clock to hold in Windows over about 3.94Ghz~ or over 2.585Ghz on the NB.
Thanks everyone for actually reading this way too loooonngg post!!