zizo: 4202MHz @ 1.32 volts is excellent. Maybe you could add a picture to the Wikipedia and show the world what a grade 'A' E8400 looks like!What's your date and batch code (Q7--A---) printed on the top of your processor or from your box? I've run my E8400 at 4200 MHz but to be Prime stable instead of needing 1.32 volts I'm probably going to need about 1.42 volts. I haven't decided whether I'd push it over 1.40 volts for 24/7 use. Probably not.
I've noticed during testing that whenever Orthos fails it always fails on Core0. That made me realize that Core0 is definitely the weakest link but I was curious to know just how weak is it?
I dropped the core voltage to 1.28 volts and loosened the timings to CL5 and booted up at 9x401MHz. I decided to run SuperPI on each core individually by doing a SetAffinity... in TaskManager and forcing SuperPI to run first on Core0 for some testing and then forcing it to run on just Core1.
Using SetFSB, I kept bumping up the MHz until SuperPI would fail on each core. Because Core0 seems so weak I thought this test would show a large difference in each core's abilities but it didn't.
Core0 could do a SuperPI 2M run repeatedly at a FSB of 449 MHz. As soon as I bumped it to 451 MHz, it failed.
I tried the same thing on Core1. It could run at 451 MHz and 453 MHz but at 455 MHz it failed and locked up the computer.
This test confirmed that Core0 is the weakest link. The FSB where each core could run SuperPI 2M reliably at 1.28 volts was only a difference of 4 MHz (449 vs. 453). Not really a good enough reason to EBay it off. You might want to try this same sort of test using the single threaded version of Prime95 and then swap it back and forth between cores until you find your weakest core. SuperPI benches like this would also give you a good baseline before priming at high volts and high MHz for 82 hours. It would be a great way to test for a degrading CPU, sometime in the future.
To finish off testing I reset everything to my 24/7 OC and did a couple of SuperPI benches. I'm hoping some other users can do me a favor and post their 1M and 8M times at 9x450 using similar memory timings or even with DDR3. I think my P5B Dlx is still competitive and I'm interested in finding out how it compares to some of the newer Intel boards when the clocks are more or less equal.
1M = 11.515s
8M=2m 28.937s
Reply With Quote
Bookmarks