BALISTX Intel E2180 Overclock

[Lithotech]

Originally Posted by donimo

Not Vcore, Vid, the Voltage your cpu asks for, mine says "1.325V", yours says "1.200V", why I wonder?

It is odd, isn't it? Even if there was a core change, that's a pretty big difference. Sometimes, with newer cores and certain chipsets (maybe they are older, or older BIOS version, or maybe the BIOS is new but still doesn't properly recognise the newer CPU core), sometimes the software reports incorrectly.

I've also seen it report the power savings voltage (with it enabled) even though the CPU is kicked into full speed and voltage because it is under load... Everest Ultimate shows full voltage, yet cpuz still reports the lower power saving voltage. Only seen it a couple times, and again might be due to older BIOS or cpu driver etc.

He is still in spec, as are you, the range covers you both as per the chart I posted. But I suspect, and maybe someone can verify, that the low value in the chart refers to the voltage used with power saving modes.

Regardless, I always try to run the CPU volts set manually to the lower value and none of the power saving features on.

First at stock speed, I memtest it a full pass of all tests and 20 or so passes test 5. If it passes with no errors I then let it into windows and run a 32m superpi. If it passes that, I'll run Orthos dual prime for 4+ hours. Nearly all systems will pass this, but occasionally not. I do this with both AMD and Intel systems. For AMD the lower value is definately a "regular-always" setting, and not a power saving setting -- it goes way lower in Cool n Quiet. So instead, I take the lower value and knock a couple more tenths off. If it passes a few hours Orthos, I move on. If it doesn't, I increase the CPU volts one tick at a time, testing each one. I keep increasing it untill it passes Orthos. It better pass by the time I reach the high value in the chart, or I have something amiss. For stock speed Orthos Prime, I want at least 24 hours, and for an overclock setting I want 8 hours passed with no errors. But I don't bother with a 24 hour Orthos at this point because it is not yet the final setting. Once it's stable, I move on.

The thing is, some chips like to be undervolted when overclocking. It's sort of rare, but more common than you'd think and would probably be more common knowledge if more people tested for it. Most people just up the VCore right off the start and push the FSB till it gets unstable. I never do it that way, and always determine how far I can push the FSB at stock volts first and then undervolted next, and sometimes undervolted first and stock second. So...

Second, with the cpu still set to it's lowest voltage rating, I increase the FSB 20-25mhz at a time (80-100mhz quad-pumped) and memtest it one full pass up to and maybe including test 6. I keep increasing the FSB until I get errors in one of the tests. I then know what test will trip first, and from here on I only run that test, saves time (sometimes I am not in a hurry, and let it run more passes or tests). I also now know roughly the limit of the cpu when it is undervolted. Depending on exactly how fast the FSB is running, I'll back off say 5-10% and then memtest it in the test that first tripped, a good 20+ passes. If no errors, I go through the Windows routine (as above) and ensure it is 100% stable up to 8 hours Orthos Prime.

I record these settings in a worksheet I use, right beside the 100% auto settings, maximum overclock at minimum voltage.

Next is to go ahead and raise voltages, starting with the higher value in the chart and using the same sort of technique to determine the cpus maximum ultimate clock speed, and maximum FSB speed. I'll Orthos and record these speeds at max stock voltage and max unlimited voltage.

Of course, I have to already know exactly what my ram can do, and exactly what my mobo is capable of, and adjust as necessary to achieve the optimal CPU settings. I use the same sort of routine to determine ram capabilities, but for determining top FSB speed for the mobo is even easier:

Drop the CPU multiplier to lowest setting - s/b 6 for Intel. Drop the ram to the lowest divider, s/b true 1:1 for Intel (266[533] for 1066FSB and 333[667] for 1333 FSB). Raise the FSB, 20-25mhz at a time is fine but you can start quite a bit higher than stock speed, go for the nutz or be conservative, no matter! Keep raising the FSB untill it doesn't boot. Open the side panel and reset the CMOS, or maybe your mobo recovers gracefully, and comes back after a cold start. Back off 10mhz at a time untill it boots, then do the memtest dance and again back off a few mhz at a time untill it passes. Finally, do the Windows boogie and keep backing off till it passes 4-8 hours prime. Sometimes the CPU can limit the mobo's max FSB speed, you can't test for that untill you try a different CPU in it. You can check overclocking databanks for your mobo make and model to see what others can get, and know roughly where you should be.

I'll still play a bit with the other voltages, just to be certain that any errors I get from above are indeed caused by CPU voltage and not something else. For example, if I get to fairly high FSB speeds before the undervolting trips and error, I'm probably going to up the NB, MCH or FSB voltage before I touch the CPU volts. I quick test to see if the errors go away, especially if they don't go away when I raise the CPU volts. It'll depend on what test errored too, some tests are more NB volts related, and some are ram related -- generally, if it is a NB related test it is also a CPU related test. On AMD systems the mem controller is in the CPU, so CPU related errors can show up in memory related tests. If you search around, you should be able to find a couple of tutorials explaining the tests and which components they relate to.

So... after ALL THAT... definately try setting yours manually to 1.2 or 1.15, and see where she goes, and particularly, how many degrees temp you can save. I've never picked up more than a degree or two at stock idle, but have picked up considerably more under load. Depends a lot on the heatsink fan too, if it is a really good one then you don't see anything on the lower end of the scale since it is already overkill and probably getting the maximum possible low temps -- in other words, you could put a fan with 3x the RPM on it and the temps won't budge, they are already room temp or just below. But on cheaper HSFs it is more noticeable. You can also slow the CPU fan down to see more difference, and less noise. I can't say anything about Intel's stock cooler, because I have yet to actually use one, lol.

Originally Posted by donimo

fair enough, maybe they will trade me for a 9650...

Hell, I'll trade you for a 9650! Will you take an E2160 for it? I'll even throw in 6 or 7 stock Intel heatsink fans!

.

[Lithotech]

In case anyone is interested, here is a PDF of one of the worksheets I use. It is generic, and should work well for any nForce chipset and not bad for P965 or P35 chipsets.

I have a number of custom ones, specifically for certain motherboards, where the items are identical to the BIOS settings... same order and names etc.

I'll be starting a thread somewhere here and put links up for all of them, hopefully in the not too distant future. When I do, I'll cross post it here.

Link to PDF:

Generic nForce Memory & BIOS Worksheet

Low res screenshot:



Enjoy!

.

[3.0charlie]

Originally Posted by donimo

Not Vcore, Vid, the Voltage your cpu asks for, mine says "1.325V", yours says "1.200V", why I wonder?

Here's your answer: Intel® Pentium® Dual-Core Desktop Processor E2180 - SLA8Y

VID Voltage Range: 0.85V – 1.5V

Quite a range.

[donimo]

holy crap stick, I wonder if anyone HAS a 0.85 volter...

I may try that upstairs after I get this one going...

[donimo]

Hell, I'll trade you for a 9650! Will you take an E2160 for it? I'll even throw in 6 or 7 stock Intel heatsink fans!

done, of course I think you got me backwards from my meaning....

[Lithotech]

Originally Posted by donimo

done, of course I think you got me backwards from my meaning....

Doh! Was hoping you wouldn't notice that...

Huge range! The .85 must be with speedstep and c1e. It's still lower than what I found a while back, wonder if it is a newer core that'll do it. My 2180 was produced 09/07/2007, ancient by todays terms.

.

[Mr. Cipher]

Originally Posted by 3.0charlie

Here's your answer: Intel® Pentium® Dual-Core Desktop Processor E2180 - SLA8Y



Quite a range.

That's exactly what I had found on the intel processor spec finder website. Certainly conflicts with Wiki, and the reason why I was wondering about everyone saying 1.35v is the max.

[Sennett]

I set the vid for my E2140 to over 1.4V. I've tried beyond 1.6V in attempts to squeeze more hertz out of it. No luck. But my point is, 1.35V is not the max. :)

[IntelClocker]

Don't forget to add 15C to the temp sense your core temp says Tj. Max 85C and not 100C