AMD Athlon X2 4200+ Dual Core Overclocking
I think I'll try to keep this one away from the knife. Stay tuned for some results
Going to be running it on air for a little while yet.
Going to be running it on air for a little while yet.
0
Comments
I'll take it you not gonna popp the lid yet !
Hahaha *yet*
After my recent experiences with my Opteron 148, I've been turned off IHS removal--especially for phase-change. Just not worth it for 24/7 use. Definitely more useful for high-end air/water cooling. Probably the worst part is that it will be difficult to sell the 148 afterwards. Not much of a market for a chip that looks like a dog has been chewing on it.. although it does still work wonders
I would love to follow your progress, so if you can be detailed!
I am currently doing that to my Opty 170 - and I have it priming on 1.24 vcore.
If thats good (after about 24 hours) I am going lower.
Also a quick question if I may ... do you consider running a divider for the memory to be a big performance hit?
Thanks, and hope you get a great result with that processor with its lid on!
Hey Krazeyivan, very funny that you should mention burn-in, because I really wanted to give it a genuine shot with this processor. Generally, I have given new chips a good run at stock clocks and volts for a few days (running prime or F@H) and then I start to give em' heck. Some people swear by burn-in.. it really can make a difference in some situations. I'd also like to see if it can make a difference with this chip.
Were you following a specific 'burn-in' guide?
In regards to memory dividers, there is no performance degredation with any K8 (754/939) processor while using memory dividers. Unlike older platforms, K8 runs the memory at a divisor of the actual CPU frequency (i.e. CPU/11) for the 4200+ that I have. So as you can see, even at default clockspeed, your memory is already running on a divider. When you set your memory to '166MHz or 5/6' divider etc, you are basically just changing that CPU divisor. It would become CPU/13, effectively lowering the memory speed. Although your memory runs slower, there is no 'extra' penalty for changing the default divisor.
I was able to run prime stable 11x237MHz (2600MHz) at default vcore (1.33V in bios). 2.7 seems ok at about 1.4V, but I didn't test long enough to validate that. Core1 craps out pretty quickly at anything above 2.7.
I'm going to have to be patient. I'm using this chip for testing purposes at the moment. Working on an article that called for a 939 2x512K cache dual core
My memory burn-in is happening because I spent all Sunday rebuilding my PC, its neater cable-wise. I rebuilt it mainly to fit a different Northbridge cooler but the one I got was out by 1mm or so, shame but anyway I recoated in fresh Formula 5 (bit like Artic Silver 5) - and I seem to have got rid of my strange temp reading "PWM IC", more on this in my thread (some photo's to come too!)
anyway after the rebuild - I was having problems going up in vcore (don't ask) so I decided to go down as I want this running cool and with possibly a higher overclock.
I am lucky that the DFI board I have has options galore!
I started at stock and did a 24 hours max heat burn with prime 95 on both cores. No sweat it passed. So I dropped the vcore by 0.025 retesting for at least 10 hours. When that passed I dropped again after another 10 hour pass I dropped again this time I want at least 24 hours stable, I am going to keep going and see how low I can get. When it fails I will increase a step again and reburn. Hopefully this will enable a cool processor and fast overclock.
Will keep you informed...
I am priming mine now at 1.21v, if it passes 30+ hours of torture it goes into 1.1v territory...
Afraid not. I haven't had a chance to tweak much more yet. I've been playing with some other hardware that you guys will hear about on Tuesday
3GHz no problem..
Its too bad that Core1 is so much weaker than Core0. There is an almost 200MHz delta between the two. Core0 can do almost 3.2GHz on the exact settings listed above. This could probably be a 3.4GHz dual core if it wasn't for core1. Ah well, such is life
I had success at 3.1GHz as well (dual sp2004 stable for over 8 hours), but it needs about 1.56 volts to get there. This seems to be a better 24/7 clock. Going to keep tinkering but I'm pretty pleased for the time being.
Hand picked the stepping.. the CDBHE, LDBHE and LDBFE manchesters seem to do just fine under single stage phase-change.
Yes, Vapochill LS.