new AMD dually: K7D and dual T-bre burn-in questions ...

[wquiles]

I got my K7D with dual 1800's T-breds Prime95 stable (two instances) and then proceeded to increase the voltage to start the burn-in process. I have been keeping careful notes as to the iddle temperatures, whether it posted and whether Win2K loaded without errors at ever voltage that I tried. I am currently running at 15x150 for about 2250 MHz stable, at least outside the case

The initial Primer95 temperatures that I got at Vcore of 1.80 (higher than required for stability) were about 56/59 (case 33) so I decided to stop there and not increase voltage any higher as that was close to 60dec (at least for CPU #2). I am using Motherboard monitor to read these numbers as per cmcquistion's guide.

Question 1:
1) As reported by others, CPU#2's temp is higher than CPU#1's by 3-4 degrees. I guess this is normal, but which one is the "correct" one, CPU #1 or #2?

2) I have been assuming worst case and using CPU#2 as a guide for temperatures. Since under Prime95 (priority 10) CPU #2 got to 59 degrees I decided to stop increasing the voltage as that is close to 60 which I considered max. Should I use 60 degrees for burn-in, or should I go even higher? I could go a little bit higher if I use CPU#1 as a guide, but I already have about two days at this level with ZERO errors/problems.

3) The system has been running on 1.80 volts for almost 48 hours. It apears to me that the temps are now a tad lower than initial. Case temp (per the motherboard monitor) has changed 1-2 degrees but even after accounting for this difference the CPU temps consistently do look a little bit lower (about 2 degrees or so) at the times that I have looked from their initial values when I got started. Is this normal, or is it that the Prime95 temps are not always max (in other words the "stress" on the CPU's goes up/down depending on the test being performed)?

4) How do I know when to stop the burn-in process? Just a couple of days is enough? As you can imagine I am looking forward to really using my new monster !!!

5) Once I am done with burn-in, I am planning to put everything in the case, close the case, add the case fans (two 120mm fans in the LX-6A19 case recommended by cmcquistion), and then start lowering Vcore and look how low I can go while maintaining Prime95 stability. Are there other programs (links would be apretiated) that would also help me determine stability of the system once it is in the closed case?

6) Should I be expecting lower performance/stability once everything is inside the case? I guess temps should go up a little, but I have no idea what to expect.

Thanks again for all of the GREAT help I have gotten so far. This forum is totally awesome!

William
(almost done with his first-ever dually!)

 

[Deathknight]

Well I think that the answer to which temp is the right temp is both. Now if you need to determine if your machine is running too hot then the higher of the 2 is going to be the one that hits your predetermined limit first (heh duh ).

As for burn in well if you are burning in with the purpose of increasing your max oc well the objective is not to cook your processors (burn in is probably not the best of terms for it). The goal is to run as much voltage as possible with as little heat as possible (optimal burn in is to crank the voltage and to underclock, thus reducing heat). Not as easy to do on a k7d master since you have so little control over voltage and mult compared to many single cpu motherboards.

 

[wquiles]

Deathnigh,

Yes, I am doing the burn-in with the hopes of being able to run stable OC with the lowest voltage possible.

So I guess (if I understood you right) that I should stop this test, change the board multiplier to 100 (15x100 = should be a comfortable 1500Mhz) , and then increase the voltage as much as it will go or until I hit 60 deg on CPU#2, right?

William

 

[cmcquistion]

So I guess (if I understood you right) that I should stop this test, change the board multiplier to 100 (15x100 = should be a comfortable 1500Mhz) , and then increase the voltage as much as it will go or until I hit 60 deg on CPU#2, right?

William

Exactly.

 

[diehrd]

I am curious as to why 60c is the magic number here.In my experence AMD chips can get real goofy from 54c and up and I have always targeted 50c as a max temp for these Processors..Just curious as to how you chose 60c

 

[cmcquistion]

60C is just a number. In my experience with the K7D, you're ok until you get to about 60C. 50C is better, of course, but instability on this board, in my experience, doesn't kick in till after 60C.

 

[diehrd]

Ok well from my experence the processor temps affect the processor not the main board. SO I would guess that the actual temps are below what is reported,And leaving head room no matter if 50c is your number or 60c is very important because we all know dust and other factors like room temp will make temps rise in short order.So when overclocking it might be a good idea to consider these factors before setting the processors at the ultimate max speed and leaving them there. Just my opinion....

 

[cmcquistion]

We should also take into account that this board reads CPU diode temps and it does NOT use a CPU thermistor, like many other boards, therefore the temps reported on this board are bound to be higher than the temps reported on boards such as the Asus A7M266-D, which uses a traditional under-the-CPU thermistor.