• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

Overclocking - stability (56k beware)

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

||Matt

Registered
Joined
Mar 13, 2005
Location
Northampton, England
I brought a 3500 939 Winchester a while back and overclocked it up to 2.4Ghz a few days after I installed it. Runs fine in games, runs SuperPI and is Prime95 stable overnight (about 8 hours).

I have come to have another bash at overclocking it a little more, but before I go any further I will give you the specs FYI.

CPU: Winchester 3500 S939 @ 2.2 default. CBBID 0451FPFW 1174365A50167
Mobo: Asus A8N-SLI Bios version 1006-001
Mem: Corsair value select 2x512 cas3.0

Basically I have added a divider so that my memory doesn't interfere with my overclocking of the CPU, because the memory is poor. And I have set the PCI Express to 101Mhz so that it doesn't get locked to my CPU frequency whilst I'm trying to overclock.

At the moment I am running at 240x10 at 1.475V, the memory is at 2.7V and 333Mhz 3.8.3.3 on 2T timings. The HTT is set to Auto - I don't know much about this.

In this state the system runs windows and everything fine as I stated above and is prime95 stable for 8 hours or so. The problem comes when I raise the CPU frequency up to about 250x10, it will run SuperPI but it will error after six minutes on the prime95 torture test (top one).
The computer generally seems to run alright but I would like it to be more stable than 6 minutes on Prime95.

I am confused as to what may be causing this as the temperature on idle is down at (35C and load 50MAX) so I don't think the temperature is the problem. I can't see the memory or PCI E card being the problem as they aren't affected when I raise my CPU freqency because they are locked lower.

Anyone have any ideas? or is this just what happens naturally when a CPU reaches its max? This is the first CPU I have overclocked.

Some pics:
cpu_z_2400_cpu.JPG
cpu_z_2400_cache.JPG

cpu_z_2400_mainboard.JPG
cpu_z_2400_memory.JPG

cpu_z_2400_spd.JPG


||Matt
 
Thanks SuperNade for the links!

I have up'ed the HTT to x3 and nothing noticable happened.

If I am raising the CPU frequency to work faster should I also raise the CPU voltage? If so when should I stop? I don't want to blow this CPU up.

||Matt
 
http://www.neoseeker.com/Articles/Hardware/Guides/a64ochowto/4.html

Very good explaination of what the divider does and how to use one there. Thanks to SuperNade.

I have a SuperPI and two 10-minute torture test stable system at the moment running at 250x10 @ 1.5V
Will run Prime95 overnight and with luck it will still be running tomorrow morning, thanks for the links SuperNade they explain it in simple terms :)

Can anyone recommend me a good program for checking my CPU & Mobo temperature in Windows?

||Matt
 
Guys,guys!

Please thank Reefa Madness!! He pm'd me with those links. I'm just spreading the good news :D.

Matt, with your Winchester, 1.6V would be as far as I would go at first. I see that you have a CBBID Winny. Statistically speaking, CBBID's top out close to 2.6GHz. Keeping this in mind, see how close you can get to 2.6GHz. Run 32M Super Pi for a quick estimate on stability.

You can use speedfan to check your Temps in Windows:-
http://www.almico.com/speedfan.php

More useful links (Please drop a line to felinusz for writing up such an excellent guide):-
http://www.ocforums.com/showthread.php?t=335813
(Pay particular attention to my post about the GROMACS based stress tester.)
http://www.ocforums.com/showpost.php?p=3569801&postcount=51
 
Thanks for the info!

One last question, I have calculated that I could get to 2.6Ghz by using 289x9 or 248x10.5 would is generally better to use, higher CPU freqency or higher multiplyer? Or is just the trial and adjust method best practiced?

||Matt
 
Mate,

One thing to remember is that half Multipliers don't exist on the A64. I don't know why MoBo makers have included the half multiplier option, but regardless of that, it is not supported by the CPU.

With the VS, I'd say 289 x 9 or even 260 x 10 is better than a 248 x 10.5. One of the articles lined to in my previous post, states that you don't lose too much in benchmarks by running on a divider. If so, you should not see any real world difference.

Make sure you bump up VDimm to 2.9 V.

Good Luck!
 
Thanks SuperNade.

Yes Im running at divider as that information recommends if you have poor memory, as I do. I will give 289x9 a try and will lower the HTT down to 3 so it doesn't go over 1000.

Thanks for all the help!

||Matt
 
Back