Originally posted by: hokiealumnus
Well, I'm back. To start, thanks to all who have helped in the past and to anyone willing to read through this and attempt to answer my questions. Still relatively new to this and the questions just keep coming. Warning: this is LONG. My apologies, but as you've undoubtedly come to know, rather than have to go back & forth, I try to be as thorough as possible to save all of us some time in the end. (Posted this @ abit in OC forum also.)
Recently, I've abandoned the CPU core speed (reached the maximum I can feasably get with the stock cooler at 3.0Ghz on an E4400) in favor of seeing what I can get out of the FSB. This has been an interesting ride, with quite a few more CMOS clears than were ever required when just going for core speed!
In order to do this, obviously, I've needed to drop my multi and let the FSB "fly" as it were. It has prooved moderately frustrating.
To begin with, I have obtained two perfectly stable overclocks. One (
HERE) at 333MHz FSB w/ 8x Multi, & the other (
HERE) at 333Mhz FSB w/ 9x Multi; both with 1:1.25 mem ratio in order to keep 4-4-4-15-2T timings on my RAM without having to raise Vdimm. These were great, to start with. Both obtained with 1.345Vcore (just one step [.02V] up, which is preferrable, since I don't have C1E & EIST to help with idle temps after dropping the multi).
They were perfectly stable through orthos testing (in screens showing the speeds) as well as all Sandra tests, with screens as follows (compared to prior saved overclock results):
Memory Bandwidth Test:
@ 8x |
@ 9x
Memory Latency Test:
@ 8x |
@ 9x
Cache and Memory Test:
@ 8x |
@ 9x
Now, the problems (you just knew it was coming, didn't you?).
Each time I restarted the PC (normally, from Windows, using XP, SP2), it decided not to start up again. Just went down, did the usual double-post and never actually posts again. Just sits there w/ the HDD light on without beeping. This is exactly the same behavior as when I raise the FSB too high (we'll get to that in a bit, addressing what I've already gotten happily stable so far). So, I reset CMOS and decide to see what happened. Changed all settings back the way they were and, just to be sure, raise MCH by one notch, thinking the NB is needing a little boost to keep up. Boot up into Windows as normal, run Orthos to be sure and restart to see if it's successful. Nope, same problem. Tried boosing NB & SB, just in case. Same result, so put them both back to default. Tried boosting Vdimm, same result (also put back). Then, I try boosting Vcore, putting it up to 1.365 (two 0.02V steps). What do you know? ...it works, restarts and acts normal!
So, my main problem is my cooler. I know it needs replacing, just not feasable currently, so I'm trying to squeeze everything I can while still being safe.
Running Orthos large with the 333/9x combination yields 75C temps, too high for me to deal with; I like pushing the envelope, not being one degree from torching the envelope! So, at 333/8x/1.365Vcore, it peaks at 72C with Orthos large. It's warm, but doable since it won't touch that with standard (gaming) loads. The main problem with that is the idle temps; since C1E & EIST have to be disabled are about 3-5 degrees higher than they are using 300FSB/10x/1.345Vcore (because it idles @ 1.8GHz w/ lower Vcore).
Now, the questions:
1. Is it really impossible to use C1E & EIST with a lowered multi, ever? Why? Doesn't this seem pretty illogical?
2. Will idling 3-5ish degrees higher (44-47 as opposed to 40-43, depending on room temps) hurt the processor more than it otherwise would in the long run?
3. Similarly, will idling at 2.66GHz instead of 1.8GHz (and non-dropped Vcore) hurt the processor more than it otherwise would in the long run?
4. Why on earth would the desired FSB & Multi at only 1.345Vcore actually start up the first time after setting the BIOS and sucessfully bench all Sandra tests (ran the processor ones too) and run all Orthos tests (not just the blended test, but all of them) and then fail on a restart (repeatedly)? THAT, my friends, is illogical.
-----
Allright, I believe that ends my questions on those particular settings. My last few have to do with pushing the FSB a little more. I seem to have hit a hard brick wall at 333MHz FSB with anything (even just 340) involving a reasonable Vcore. After incrementally trying every combination of 1-step-increases on Vdimm/NB/SB and Vcore, it's all Vcore. 1.425 is the next stable Vcore at 350Mhz. Without it, the system just does the normal double-boot thing and just stops without posting. Seems like a big jump from 1.365 to 1.425 (required to be able to restart, it'll do the start-up-and-be-stable-once-but-not-restart thing at 1.405) just to go from 333 to 350. Wouldn't post at 366 with 1.425Vcore and I'm already uncomfortable, not trying to hurt my system here...remember, the key word is SAFE.
HERE is the result at 350FSB/8x/1:1.2/4-4-4-15-2T. Temps got to 75C running Orthos large, so I think I'm done here. Can't run 75C and feel safe. I tried 366MHz FSB at 1.425 and it wouldn't post.
So, questions:
5. It's not worth it, right? 1.425 just feels way too high without upgrading the CPU cooler.
6. Is it normal for such slight increases above 333MHz to require such a large jump in Vcore?
7. Well, more of a comment really. After hearing bad things about OCZ (after purchasing of course), I'm happy to see it's not the thing holding me back!
-----
Well, that's it. I thank anyone and everyone who even made it this far. Sorry for being verbose! Any and all help will be greatly appreciated!