View Full Version : Why disable cXe or c1e?
homefry
09-05-2009, 07:59 PM
Why not just stress test with it enabled?
I've read many guides that say to disable it, stress test, find the stable settings, then enable and see if you're stable normally.
But that defeats the whole point of the first stress tests because once you enable cXe, you are not sure of 100% stability.
zanzabar
09-05-2009, 08:36 PM
it wont let the cpu go to 100% load with it enabled and you shouldent use speedstep or c1e or the like unless u have an i7 and need to force turbo if u are ocing
CxE works fine on most boards and does allow 100% loading of the CPU. The benefit of first testing with it off is it removes it as an issue of stability while testing CPU and memory.
homefry
09-05-2009, 08:58 PM
I have an i7 and am forcing turbo on. I'm stress testing with it disabled.
If enabled, I'm unstable. 3 hours in with it disabled, I'm so far stable.
But if I end up being stable with it disabled, I can't just enable it and call it a day (like all the guides out there say).
homefry
09-05-2009, 09:00 PM
CxE works fine on most boards and does allow 100% loading of the CPU. The benefit of first testing with it off is it removes it as an issue of stability while testing CPU and memory.
Which is my point. You can't just stress test with it disabled, find stable settings, and then turn around and enable it.
If you want it enabled 24/7, then you have to stress test with it enabled.
Maybe it's a nit pick at all these guides written on the net because they're wrong.
Testing with it off would be no different then lowering the ram frequency while testing CPU stability. Some people find it easier to test each individually then test all combined. Guess it comes down to what works best for each person. I doubt many guides recommend testing with CxE off and then enabling it without a final round of testing.
Like Praz said, if it is eist, c1e or c3,6,7 affecting stability, you will wont to know that, since first, it is much easier to work out one variable at a time, and second if you have to increase vcore to compensate, that may negate some or most of the piddly ~3% power savings you get by using them while overclocking. Manually entering in vcore on most boards (ocing) disables the ability for those power saving features to decrease vcore, so only decreases frequency, though some boards still regulate it but those are often the ones that then lose stability with them, requiring more vcore negating the piddly power savings. I would only enable them if my bios forced me, by linking their use to turbo or TDP limit disabling.
For example, my entire computer, idles OCed at 350W. With eist, c1e, c3,6,7 enabled it idles at 340W (vcore not dropped just frequency). Somehow I just cant rationalize that as power savings, and add in the annoyance it slows spi times and other benches/apps as it takes a second for cpu to get to full power...the annoyance outweighs the piddly power savings.
Using S3 while my computer is not in use, now that decreases power down to 38W (most of that is UPS/router/cable modem). Or some turn the computer off. Now that is power savings you will notice and does not affect performance.