A64 3500+ Winchester OC

mauri

Guest
Jun 4, 2005
139
0
0
I Just bought a new Zalman CNPS9500 for my setup to replace my broken CPU-cooler, and I wanted to see how well does my A64 3500+ Winchester overclock. Here is what I got so far:

11*240 = 2640 MHz @ 1.5 Vcore
11*230 = 2530 MHz @ 1.45 Vcore

I had to set my memory to 166 (333) Divider, since my memory isnt that good for overclocking (on 1:1) and I set HTT multiplier at 4. How much effect do those have on overall performance? the memory timigs are 2.5-3-3-8.

The system was stable for 12 hours of Prime95 torture test with 2530 MHZ and stable after 14 hours of priming at 2640 MHz (had to turn off, since i'm leaving town for a few days, but I will try to run full 24h when I come back).

My question is that is it safe to run the system with 1.5 vcore with those results? With those settings my temperatures are 34 C while idling and prime95 temperatures are stable 45 C (1,5 Vcore). My MBM indikates the motherboard temps at 30-32 C. Are those temps safe to run in the long run and how are the results so far?

Is there any other good methods to test the system/CPU stability than Prime95 torture test, or is it enough for indikating if the result is stable? So far I've been running the Prime95 torture test.

I would like to try to push the processor more (by raising HTT?), but I don't really feel comfortable to raise the Vcore any more (from 1.5) if it is needed. Any suggestions or advices to how to continue from here? What is the affect on lowering the multiplier of the CPU?

my other components in my setup are:

ASUS A8N-SLI Deluxe, 1014 Bios
A64 3500+ Winchester, Zalman CNPS9500
2*1BG Kingston DDR400 Valueram (2.5-3-3-8)
Radeon X800XL, Zalman VF700-Cu
Antec TPII 480W
Hitachi T7K250 160GB

thanks!
 

T101

Senior member
Oct 13, 1999
558
0
76
No dangerous temps there. But I would not worry that much about the temps, more about the voltage. Overvolting electronic circuits kills them faster than the heat can. It is however true that cool circuits survive overvoltage better.
When it comes to overvolting, you better check what the manufacturer have set as a "safe" range for the chips operation. At the moment I do not remember what that is with the AMD Winchester, but it should not be hard to find on the net :)
 

fbrdphreak

Lifer
Apr 17, 2004
17,555
1
0
1.5V is on the high end. Can you run 2.5GHz @ a lower voltage than 1.45? You may even be able run it @ 1.4V if you're looking. Since your max is 2.6, losing 100MHz to reduce voltage is a worthy tradeoff to me.
 

Icepick

Diamond Member
Nov 1, 2004
3,663
4
81
Nice overclock. It's good to see a fellow 3500+ Winchester owner is having good luck. I've got mine running at 2607 with 1.49V, memory divider at 166 and 2-3-2-10 timings. One thing interesting thing I've found is that my mobo raises the voltage to 1.53V when I run the Prime95 Large FFT Heat and Stability test. Anyway, I've run this overclock with these settings for a few months. I have run the following Prime 95 tests:
Small FFT - 29 hours, max temp: 51C
Large FFT - 25 hours, max temp: 53C
Blend - 14 hours, max temp 51C ( I had to stop this test early )

No failures or warnings on any of the tests.

In my opinion, your 2640 overclock at 1.5V is excellent. You can push up the voltage as far as 1.54V and be very safe. The reasoning behind this is that the CPU manufacturers determine the default voltage within + or - 10% of the guaranteed safe operating range. Default voltage for Winchesters is 1.4V, therefore, 10% increase is 1.54V. If you simply must squeeze as much overclock out of your processor as possible then I'd recommend raising the voltage to 1.54V and seeing how fast you can run it stable. Verify stability by running Prime95 Small and Large FFT for at least 24 hours each. Then you can rest assured that the CPU will be able to handle anything you throw at it.

Setting HT to 4x has a very negligible impact on the performance. Also, using dividers on the RAM has very little impact on performance because Athlon64s are not bandwidth starved due to the memory controller being integrated on the chip instead of on the north bridge like Intel procs.
 

VStrom

Senior member
Dec 27, 2004
423
0
71
I have my 3500+ winny @2640 as well (8x330) w/a 150mhz divider @1.5v. Couldn't get it stable anywhere less than 1.5v. My temps on air max at 49 but are usually at 47-48c under load. Idle I'm looking at 37-38c with both of these being at an ambient room temp of ~68F. I've gotten windows to load @ 2850 with 1.575v but not stable for most apps.

I just ordered an Opty 148 and some OCZ DDR600 to see how far that will go :evil:
 

mauri

Guest
Jun 4, 2005
139
0
0
Thank you all for replying me.

Icepik: Thank you, it was nice to see that my CPU overclocked so well and that someone else also had the good experiences with overclocking the 3500+ winchester. It went further than i expected. I will try to seek its limits when I get back home. Btw; I also had the similar mobo Vcore increase to the BIOS settings which seemed strange, but it worked alright in the end (I set it to 1.475 but my mobo altered it to 1.5 at load). It seemed bit worrying but it worked? And as I remembered it, (and as tou confirmed it) the HTT multiplier and RAM dividers have not too bad impact on the performance. thank you for confirming that, I can leave it as it is now if I feel like it. I also have to test the stability with more intence Prime runs (so far I've only runned it 14 hrs with similar tests as you have done).

T101 and fbrdphreak: I'm aware that the voltage kills the processor more easily than the temperature. that is the reason why i asked about it. I appreciate the input greatly, so far as i've got it, the CPU losed its stability closing to 240 HTT at 1.45V, but I also have to check the stability at 2.5GHZ @ 1.4V (by lowering the voltage at 11*230), it is very good point indeed: 100MHz vs 0.1V may indeed be a good trade off. But i'm so tempted to run it closer to the "stock" 1000 & 400 MHz HTT and RAM speeds, which I achieve by raising the HTT with the dividers (and again the voltage).

Now I still I have to check it how far it goes with 1.4V (230 HTT, hopefully..), what is the highest overclock and seek the lowest voltage/overclock. All the advices have helped me alot. Thanks.