So everything is assembled and working. Getting all the power cables sorted was a bit harder partly because of the mATX motherboard and partly because I decided to move the HDD up to nearer the DVD drive.

The benchies are no great shakes at stock speed. About 20% slower than sdbardwick. I think that is because the memory is running 1333/CL9.

[QUOTE]Best time for 768K FFT length: 14.15 ms.
Best time for 896K FFT length: 17.28 ms.
Best time for 1024K FFT length: 19.61 ms.
Best time for 1280K FFT length: 24.53 ms.
Best time for 1536K FFT length: 29.38 ms.
Best time for 1792K FFT length: 35.82 ms.
Best time for 2048K FFT length: 40.49 ms.
Best time for 2560K FFT length: 53.61 ms.
Best time for 3072K FFT length: 64.92 ms.
Best time for 3584K FFT length: 79.34 ms.
Best time for 4096K FFT length: 89.47 ms.[/QUOTE]

But it scales like a beauty. 4 LLs running with no slowdown at all. Aso it is pretty quiet even with 3 120mm fans running and the core temps are in the 40-46C range. Plan to get overclocking in a day or two.

Some help getting a stable AND correct overclock
 

I built my i5-750 over the weekend. I went with the Gigabyte P55M-UD2, the Noctua U12P and G.Skill 1600/CL9 memory.

After running stable for a day at 3320MHz, I upped the clock yesterday to 3660MHz. The system is stable and no crashes or freezes. The memory is at 2:8 so running at 1464MHz and it chose an SPD setting of 10-10-10-27. Thanks to the terrific cooler the temps for the four cores at full load - 4 doublechecks - are 49-55C. However, I got six [B]sumout [/B]errors overnight on one of the four cores (the third). So I am thinking that there are four possible courses of action I can take depending on what is causing the errors.
[LIST=1][*]The CPU voltage is currently 1.28125 V from the standard of 1.25625V. Maybe I need to up it a bit more for correctness. I have plenty of room temp-wise.[*]The memory needs more voltage? Currently at stock 1.5V. Or any of the other components such as MCH, PCI-e etc.[*]Memory is too fast and I need to dial it back a bit - seems unlikely to me.[*]The CPU has reached its limit for correct operation and I need to dial it back. I will not get a stable and correct overclock at this speed.[*]While I am at it, what it the XMP mode for memory?[/LIST]So any ideas folks on what might be causing the problem and how I can push the chip further?

As you are not stressing the memory too much, I would recommend to start by upping the CPU VCore voltage: something like 1.30 - 1.35 V.
Check the LLC value (level). At 3.66 GHz, it could be necessary to set it to level 2 (I think UD2 supports it).
VTT voltage may also need to be set up to something around 1.3 - 1.35V.
Upping the memory voltage may also be worth a try, but I don´t think it is a real issue at 1464 MHz.
Your CPU, unless you´ve stumbled upon a really crappy sample, should be able to operate stable at 3,66 GHz.

I've read conflicting information about setting LLC. I have pulled back the overclock just a teeny bit to 3600 and it seems to be holding with voltage at 1.28. I'm going to try 3800 and 1.35 next. Thanks for your suggestions. You are right in that memory is likely not the problem here.

I think you should stick with the voltage specified by the packager(G.Skill) for your memory and that is 1,6 V.

Jacob

I have the memory at 1.5V and it should run at 1600 on that voltage. BTW, there are no errors at 3.6GHz so I'm guessing my CPU voltage was just a tad low.

Just thought I'd provide an update to those interested. I have the CPU stable at 3800MHz now with a couple of verified double-checks already turned in.

BCLK = 190, CPU = 3800MHz, Memory = 1520/9-9-9-24 1T

DRAM voltage = 1.56V, QPI = 1.25V CPU voltage = 1.3625V (I got one round off error at 1.35V so decided to up the voltage for safety. Full load effective voltage is 1.264V)

Core temperatures in the 54-64 range depending on ambient temperature and the core number - 1&3 are running hotter for some reason and 4 is running the coolest (it is running P-1).

Actual iteration time for 1280K FFT while running four tests: 17.8-18.2 msec.

[QUOTE=garo;195759]I had a small heart attack for a second when I mistakenly thought that the Noctua is not compatible with the UD2. [/QUOTE]

Ditto here....the Noctua manual says the MB must have a back plate (I assume it is so it doesn't break from the weight of the cooler). It appears this UD2 has a back place but NOT big enough that the cooler plates can bolt right into it. Rather the cooler plates are sized to bolt into the same holes that the Intel cooler does that are purely in the plastic with NO metal on either side. This concerns me again that the weight will break the plastic.

Am I misreading something?

Thanks

[QUOTE=petrw1;196939]Ditto here....the Noctua manual says the MB must have a back plate (I assume it is so it doesn't break from the weight of the cooler). It appears this UD2 has a back place but NOT big enough that the cooler plates can bolt right into it. Rather the cooler plates are sized to bolt into the same holes that the Intel cooler does that are purely in the plastic with NO metal on either side. This concerns me again that the weight will break the plastic.

Am I misreading something?

Thanks[/QUOTE]

Think I'm ok now....
thanks anyway

Stupid(?) question: Do I remove the black cover from the RipJaw RAM or is that there to protect/cool it?

thanks

It appears to be held on with some sticky tape which leads me to believe leave it on???

[QUOTE=petrw1;196944]Think I'm ok now....
thanks anyway[/QUOTE]

Except....airflow may not be an issue because I CANNOT fit the side cover with the Noctua cooler installed .. it's too high :confused: