Eeeasy there big fella!!JohnMK wrote:What do you take me for? I know what I'm talking about when I report a statistic such as that. I also run two instances of SETI@home overnight to maximize register utilization on a Hyperthreading P4. The highest temp I have seen is 54C, overnight, running SETI@home. My room is approximately 72 degrees Fahrenheit. Video encoding with XviD & Avisynth utilizes two threads heavily and distributes work well across both virtual CPUs, similiarly well utilizing the registers of the P4 and getting my temps up just as high as SETI@home.Ralf Hutter wrote:Ding-ding-ding!!! We have a winner here folks!!!Rusty075 wrote:Plus there's also the fact that JohnMK's temps are not true full-load temps. A few hours of video coding is not the same amount of stress as CPUBurn would be
I'm not capping on you either, just the fact that a DC client or even two instances of Prime95 will heat up your CPU to worst-case levels. Russ already explained the theory behind using CPUBurn and I'll give you some numbers as well. I build systems as a side business and have always been a real old lady about testing, testing, testing before I deliver something to the customer. One of my tests is to verify the cooling ability of the system and for that I use CPUBurn. It consistantly loads the CPU higher than any other application I've ever found. On P4 systems, the temp difference between running 2 instances of Prime95 vs. 2 instances of CPU burn runs from about 3-6°C hotter for CPUBurn testing. This is what I use for all my "worst-case" temp testing. I use Prime95 for stabiltiy testing and Memtest86 for memory testing but CPUBurn heats up the system like nothing I've ever found.
Case in point:
I built a P4 3.0 800MHz system a few weeks ago and used a Z-7000AlCu for cooling. Running it at 7V gave me 2xPrime95 load temps of 52-53°C. Running 2xCPUBurn ave me load temps of 58-59°C. This was at about 21-22°C ambient. This was on an Intel 875PBZ board, one that gives a fairly representive temp readout, as opposed to the Asus 875/865 boards that read about 8°C cooler than that, or an Abit 875/865 board that reads about 10°C hotter than that.
In regards your specific situation, I assume you're using the Asus P4C800-E that you referred to in an earlier thread. If so, did you add some temp compensation to your mobo, as I suggested in this post, or are you running at the default Asus readout? If that's the case, I would suggest adding some compensation to give a readout closer to the "average" P4 mobo.
I don't want to get into a pissing contest, but I've built a lot of P4 systems and am very familiar with the peculiarities of the various families of P4 boards. For a little backup, I'd point you towards the forums at OCforums and ABXzone. Search for posts from about Apr-Aug 2003 where people are asking how their CPU temps can be cooler than there mobo/case temps on their Asus 875/865 boards. It quickly became apparent that Asus' CPU temp readouts are too low. The same thing happened with Abit boards, but going the opposite way. There's some huge threads in the AbitUSA and OC forums where people are taking aAbit engineers to task over this issue.