C2D multipliers, EIST/C1E, and sleep/suspend

[nullpointerus]

I've been attempting to find the best overclock for my C2D rig (see sig). The chip itself seems to peak out around 2.66 GHz (266x10) at stock voltage (3.12v load in CPU-Z), which is probably where I will end up running it for daily usage. However, with this configuration, my RAM divider must be changed so that the RAM is 266x2.5=667 MHz.

To see whether I could get a stable 1:1 memory setup, I tried setting the CPU to 333x8=2.66 GHz and memory to 333x2=667 MHz. This configuration runs fine (stable in prime95 and for hours in my games) except that Vista refuses to properly resume from its sleep state (suspend to HDD?) when the power-saving mode kicks in.

But here's the interesting part: if I force the system to power down and then boot up through the normal POST screen to the Vista Resume screen, the system seems to resume properly. So My current theory is that Vista's resume process raises the processor multiplier to 10x (the chip's maximum), which I know prevents this rig from posting. Is this theory correct?

Or does this board have trouble hitting 333 MHz FSB at stock voltage?

Gigabyte GA-P35-DS3R Manual, see page 51 for names of voltage settings

I'm not familiar with exactly what the voltage settings do since I'm coming from an Athlon64 system.

I DID disable C1E in the BIOS options, but the option to disable EIST disappeared with the latest BIOS update (i.e. F3 -> F4). I will set Vista's power saving mode to maximum performance, which should disable CPU throttling. Maybe this will keep the CPU multiplier at the desired 8x value.

 

[SerpentRoyal]

Run the CPU at 10x multi and use 1:1.25 memory divider if you want to run the memory at 333MHz.

The CPU should be able to hit 3 to 3.2GHz with 1.35-1.45Vcore. To try this setting, use 10x multi, 300MHz FSB, 1.40Vcore, and 1:1 memory divider (synchronous with FSB speed). You may need to bump Vnorthbridge to 1.4.

 

[nullpointerus]

Thanks! I gave that a try, but no luck.

The CPU barely boots into Windows with 10x300 @ 1.45v, and the computer hard-locks soon after that. The chip won't post at 10x333 even with high voltage. I did not bump the northbridge voltage (and the BIOS says nothing about what it is!), but I will give it the +0.2v setting and see what happens.

I just checked: the board is supposed to support 1333 MHz bus speed processors, so 333 MHz speed "should" be fine.

BTW, Vista's "maximum performance" mode did not prevent the computer from locking during sleep mode.

 

[nullpointerus]

I got the CPU to run 1-hour Prime95-stable 10x300 @ 1.44v (as seen in CPU-Z) and +0.2v on the NB, but the sleep power-saving issue persists. I think I'll just stick with 10x266 @ 1.31v as that's essentially stock voltage and quite compatible with the sleep function. (My PC is in my room, yet it records TV overnight, so I need the sleep function to work to avoid heating the room up).

 

[SerpentRoyal]

Run Memtest86+ to check your RAM at 10 x 300. Keep the memory divider at 1:1 for now. Abit IP35 series don't need a bump in chipset until +460MHz, but you may need to raise VTT, northbridge, and southbridge with the DS3.

 

[VirtualLarry]

How do you know that the power-saving issue isn't just a Vista thing?

 

[nullpointerus]

The power saving works at stock and at 10x266, but not at 10x300 or 8x333 even when the CPU tests stable.