I just upgraded my dead motherboard [Topic Here] with a Gigabyte GA-EP35-DS3L and happily overclocked to 3.31GHz. It passed an hour of Orthos 2004 and an hour and half of Prime95.
Question is, do you guys think there is still some overhead? I don't want to run this chip down to the last megahertz, I just want some sustainable speed and make the chip last more than a few years.
The RAM is a combination of PC2-5300 2X1GB sticks and a single PC-6400 2GB stick. Don't ask why, the guy gave me free RAM so I took it. Its running at 418MHz X 2 at the moment, I believe.
Vcore is same as default, 1.09V. PCIe frequency is locked at 100MHz.
Using Gigabyte G-Power cooler, just applied the thermal paste little over 24 hours ago, so the pastes probably didn't set in yet. Though I did have to lift the cooler up more than once to get the pins in, could this have introduced air pockets to the contact?
When I set the system to Multiplier 8X and 400MHz FSB it wouldn't boot. I didn't change PCIe and RAM multiplier setting, so maybe that's why.
Wow, damn nice voltage on that chip. I have no experience with that chip or 45nm's all other than an E8400 I put in my wife's PC i built a few months ago, but I didnt OC it.
About all I can susgest personally is to work on the RAM timings now, which may prove difficult with the 2 different types of RAM you got in there. I see you tried it at 8x400 which would have done the job with RAM @ 800mhz, but im sure the 2 different RAM speeds/types are throwing that off. Have you tried 8x400 with just the 2gb set?
I would take out the ddr2-5300 memory; it is holding you back. With that voltage, you probably have a lot more overclocking space. The slower RAM you have will not run at 400mhz fsb. Just get rid of it, and buy a 4gb dual channel kit when you can afford it. The p35-ds3l will probably require a little extra voltage to the NB as well to run 400mhz stably. Keep fiddling until you can get the fsb to run @400mhz, then start raising your multiplier. You'll have to raise your vcore settings, but probably not by a ton. I'll bet you can run 400x9.5 for 3.8ghz without much issue at all.
Also, be aware that that mobo has quite alot of vdrop. Check your voltages in windows, and use that to decide how high you can safely raise the voltage in bios. 1.36v is the "limit" for the 45nm chips in general, but you're probably safe with anything under 1.4v. With that said, I seriously doubt it will take that much for your chip to clock that high.
Right now I'm freaking out about heat, which I know isn't much, but others have higher clockspeeds and lower temperatures - and before I changed the heatsink it was running at 60C idle, so I don't want to push it too far. That's why I didn't want to raise the Vcore too much.
The funny thing is, before I slowed the RAM down to 675MHz (I think) I ran them at 2.5X which gave me nearly 820MHz I think, and they passed Memtest fine at 600%.
Suppose another notch of voltage won't hurt, I will give that a shot. Maybe it will get me to 3.8GHz and 1600FSB!
I know for sure that this chip can handle up to 74.1C as stated on Intel's website, and people usually agree that anything less than 1.4V is fine - so I'm going to try to keep it lower than 1.35V. Besides, 3.8GHz will suffice for me, if I do get that =)
that's good and all, but all of that mismatched RAM is doing you no good. take out the slower sticks, and you will have more sucess.
also, if it's idling @60c, something is wrong. The hsf is not seated correctly, TIM is applied wrong, or it's just a terrible hsf. you should absolutely correct your heat problems before continuing to OC. Running a 45nm cpu@74c is not a good thing. Spend a few dollars on a real hsf, not a gigabyte POS. I aslo advise you search the forum and try and learn a little more about overclocking.