I'm such a liar!!

Techniques, tips, and advice for the performance PC enthusiast on overclocking.

Moderator: The Mod Squad

I'm such a liar!!

Postby Peanya » Mon Nov 29, 2004 10:25 pm

Although I could get my computer stable at 2.5GHz for the most part, it wasn't 100%. Everquest and Prime95 would crash, but only those 2. Everything else ran good, including HL2 and Far Cry.
So anyway, I finally took my own advice on overclocking and found out something. I really think the AMD 64's (at least the new ones) have a current limiter built in.
I first tested the LT bus and had no issues, games ran just fine. Next I bumped the memory to 250MHz and still stable too. I was even under rated voltage w/ 2.5-3-3-10 timings. So off to the CPU I went. 2.2GHz was stable at stock speeds. 2.3 GHz made prime95 fail almost exactly at 2 hours. Bumping the voltage up yielded successful results, had to go w/ 1.475V.
Next I tried 2.4GHz. 1.475 wouldn't work, so I tried many others: 1.5, 1.525, 1.55. 1.575, 1.6, 1.625, 1.65. I stopped there. That's a lot of voltage increase without any successful results. I do know it wasn't the memory or LT bus holding me back.
Peanya
Black Belt 5th Degree
Black Belt 5th Degree
 
Posts: 6055
Joined: Fri Feb 07, 2003 7:43 pm

Postby Twisty » Thu Dec 09, 2004 7:39 am

Is it possible that the limit simply is not affected by the power that is put into the CPU or the extra heat counteracts the extra voltage.

I don't see how they would put in a current limiter on the CPU.
I have now left the Building :tongue8:
Twisty
Black Belt 1st Degree
Black Belt 1st Degree
 
Posts: 1970
Joined: Fri Mar 02, 2001 1:01 am
Location: UK / Singapore

Postby Rabidwerewolf » Thu Dec 09, 2004 8:49 am

That is an interesting and plausible hypothesis. I have noticed when oc'ing my s754, .130nm A64 3000+, the first time I tried oc'ing through the bios(not using clock gen) highest stable oc I could to achieve was 2.357ghz with a vCore of 1.53 and htt at 235.8. I then decided to bump up the vCore to 1.55, and set the htt to 237.7, getting a high stable oc of 2.377ghz, but that is the highest I have been able to achieve. Anything at 2.4ghz and above, no matter what voltage you pump into it, it is just not stable. I even opened back up clockgen to see if I could get a higher oc, but I couldn't get past 2.4ghz either. I believe my 3-3-3-8 Kingston Value Ram and cooling may be holding be back over your system Peanya, but it is interesting that you have the memory, power and a better hs/f and can't get it stable at 2.4ghz or higher.
Image
Rabidwerewolf
Black Belt 1st Degree
Black Belt 1st Degree
 
Posts: 1519
Joined: Mon May 17, 2004 10:37 am
Location: lost in Georgia

Postby Peanya » Thu Dec 09, 2004 10:42 am

I have my LDT bus at 255x4 and my memory at 255MHz w/ a multiplier of 9, and I'm rock solid stable, so clearly the CPU isn't clocking. Maybe it's just as far as it will go? Either way, I'm happy, it is faster than my P4C 3.4.
Peanya
Black Belt 5th Degree
Black Belt 5th Degree
 
Posts: 6055
Joined: Fri Feb 07, 2003 7:43 pm

Postby Rabidwerewolf » Thu Dec 09, 2004 12:43 pm

:mb_iagree:: :mb_woohoo::

It makes me wonder when Anandtech got their's above 2.4ghz how long they ran it for at that speed.
Image
Rabidwerewolf
Black Belt 1st Degree
Black Belt 1st Degree
 
Posts: 1519
Joined: Mon May 17, 2004 10:37 am
Location: lost in Georgia

Postby Peanya » Fri Dec 10, 2004 1:35 am

I can get it beyond that too, and I can even run games and many other apps. However Prime95 does crash after about 2 hours, and also EQ will crash (heck it'll crash with any system)
Peanya
Black Belt 5th Degree
Black Belt 5th Degree
 
Posts: 6055
Joined: Fri Feb 07, 2003 7:43 pm

Postby Rabidwerewolf » Fri Dec 10, 2004 5:27 pm

That is also weird that it crashes with Prime95. Prime95 must really pick up some minute errors that doesn't bother other apps.

and also EQ will crash (heck it'll crash with any system)
True and SoE is no fun to deal with either :mb_pcmad:: , but I don't know if those problems have improved since I closed my EQ acct a couple of years ago.

It'll be interesting to see what other overclocker's will achieve, and how, on the s754 and s939 procs. I really wish their was a more detailed cpu database than the one over at overclockers.com. Some people have recorded overclocks of 2.5+ on the s754 .130nm Athlon64 3000+, but I question the validity of the posted results of some individuals. It would be nice to see a database showing results of s939 cpus.
Image
Rabidwerewolf
Black Belt 1st Degree
Black Belt 1st Degree
 
Posts: 1519
Joined: Mon May 17, 2004 10:37 am
Location: lost in Georgia

Postby Twisty » Fri Dec 10, 2004 5:38 pm

That is exactly what hte prime95 torture test is good for, showing up the slightest bit of instability. It's not uncommon to have a computer that otherwise appears to be completly stable but prime95 picks out a problem.
I have now left the Building :tongue8:
Twisty
Black Belt 1st Degree
Black Belt 1st Degree
 
Posts: 1970
Joined: Fri Mar 02, 2001 1:01 am
Location: UK / Singapore


Return to Overclockers Anonymous

Who is online

Users browsing this forum: Exabot [Bot] and 1 guest