QX9xxx overclocking results

[Little Man]

1.56v wth

shouldnt it atleast run at 1.8???

[karlcloudy]

Originally Posted by Little Man

1.56v wth

shouldnt it atleast run at 1.8???

Oh, sorry, didn't clarify, it is DDR3 and it is rated at 1.5V, not the 1.8V of DDR2 RAM

[Spork!]

hmm, I have some Corsair xms DDR3 rated to 2.1 max, although mine are happy (@1333) with 1.8...
They need the 2.1 for 1600 apparently, but are only compatible with this MB to a little over 1333...

[karlcloudy]

I've been running my Q9550 at 3825Mhz for the last week and have had absolutely no toruble with it or anything. When I did that, I ran LinX and prime95 to test for instabilities in my system. I only selected small FFTs for the test and thought nothing of it, and it passed without any issue. I decided to run prime95 again last night, to make sure was OK and to check the max temps again, etc, but this time I chose large FFTs. Not long later and core 0's thread dies pointing to a rounding error. The other cores keep processing as if nothing ever hapened and didn't crash. I briefly tried increasing the voltage ever so slightly, but all that did was shift the problem to core 1. So to really make sure nothing occured during a small FFTs test I left it going for hours today and no errors occured.

Since Prime95's description of large FFTs mentions RAM, I assume that maybe the issue isn't with the CPU voltage, although I really don't know much about that. Here are my BIOS settings for reference:


Click to view full size!

and...


Click to view full size!

EDIT: I forgot to mention my specs, but it's the system in my sig, so just look below my post for them

[karlcloudy]

Hmm, this thread is dead silent

[snappy1]

Could be lots of things, NB voltage is a bit low, cpu may a touch more volts and you could try 1.6 volts on your ram, trial and error

[karlcloudy]

Originally Posted by snappy1

Could be lots of things, NB voltage is a bit low, cpu may a touch more volts and you could try 1.6 volts on your ram, trial and error

OK then, I had tried more NB volts before, although it didn't work, but I'll up it, the RAM and the voltage a bit, hopefully all goes well

[snappy1]

With one core failing that's all you can do, keep trying.

I don't want to jinx you but an overclocked cpu is only as good as the weakest core, I hope it is just a setting issue and not a wall.

If you have no luck I'd drop it to 3.8Ghz, you won't notice the drop from 3.825

[karlcloudy]

Originally Posted by snappy1

With one core failing that's all you can do, keep trying.

I don't want to jinx you but an overclocked cpu is only as good as the weakest core, I hope it is just a setting issue and not a wall.

If you have no luck I'd drop it to 3.8Ghz, you won't notice the drop from 3.825

Yeah, well theres not much difference in that is there, but I'm trying it now, the only real reason it was that funny number to start with is that 450 FSB was a neat number. I'm also debating about whether to "not bother" worrying about it as I've had it running for quite a while without any detrimental effect on actual system stability. It hasn't actually crashed or done anything funny.

[snappy1]

I do like my systems to be prime and linx stable but I must admit I have run systems for months and months that wouldn't pass prime and with no issues.

[karlcloudy]

Originally Posted by snappy1

I do like my systems to be prime and linx stable but I must admit I have run systems for months and months that wouldn't pass prime and with no issues.

Yeah, well this system is LinX and Prime95 small FFTs stable. It can also execute 3 threads so far (by setting affinity to 3 cores in Task manager, and selecting 3 threads, it executes on 3 cores only) even when core 0 and 1 are involved, so it's a mystery to me

EDIT: it stalled on one core again, but only when I stopped playing games :/. I think I've hit the wall, because before I was trying to get 4.0GHz and even with a huge increase in voltage (to 1.35V) it didn't make the system much more stable

[cokeeffekt]

oops i already posted :P

[toXin]

Originally Posted by cokeeffekt

Try setting the strapping to 2.00b (333mhz??) instead or D (400mhz) also bump up the MCH +0.1 my board flakes out beore my Q9550 does, it sits at 4ghz on stock vcore, but MCH is 1.28 (1.1 stock)

Thanks dude tryin it now.

[karlcloudy]

I think I may have stumbled across the actual reason for this issue (hopefully). Yesterday, the computer finally did a dummy spit (starting to play up), so I started up Prime95 and did a small FFTs test, which was enough to tip windows off the edge of the cliff and it BSODed. I increased the CPU voltage by a fair amount (from 1.3 to 1.33V) and did the large FFTs test (as usual) and it still bit the dust (interestingly, more voltage makes core 1 fail and less voltage makes core 0 fail). I decided to try something out in the BIOS. Without changing anything else (apart from the memory ratio setting to get DDR3 1333) I set the CPU voltage back to stock and then set the CPU multiplier to 6.0 to make a speed less than the stock speed. Ran large FFTs and it cooked itself again. Seems like the NB might be the problem? I also did a control test with all stock settings and I didn't run it for that long, but long enough that the problem would have occured before. Because it would run on smal FFTs, I assumed something else had to be the issue and now think I have found the NB/board to be holding me back. At the moment, my settings in the top post still stand apart from:

DDR: 1.56V (the speed is set to DDR3 1353Mhz)
NB: 1.2V
CPU 1.33V (I had it on 1.3V when it hissy fitted yesterday and tried more to see whether that solved it and it did, but it was a once off, so I'm not sure what would have happened if I hadn't done it)

[cokeeffekt]

bump the MCH to 1.3 just to test