![[H]ard|Forum](/styles/hardforum/xenforo/logo_dark.png){kind=logo}
woah this test is indeed more stressful than orthos. It failed after only 10min 5 sec. whereas Orthos failed at 27min 6 sec. Both were on the blend.
@OP: Good find
We need cputest86 and systest86 haha Even if it's beta, the test can be trusted now. if it reports an error, there was one with your hardware I've spent a year in the US, but my english is still far from perfect Only a Single core A64. So i can't get Tat to run on my computer... Hard to code it the support You need to spend less time at Slashdot. Seriously.
Since 1.1.0.b06 or something like that. Including a better testing algorithm, lots of bugfixes and monitoring options 
This has been an interesting read so far but I have a few questions. If you run Folding @ Home you are using 100% of the CPU’s capacity assuming you don’t touch the keyboard since Folding will instantly stop and give up it’s cycles to what ever event you create.
Prime-95 runs in much the same way using only idle cycles, however Prime seems to push my temps about 2c higher then folding.
Stress CPU also uses 100% of any core it’s running on and also pushes my CPU temps about 1c higher then prime.
Otrhos is simply a simpler way to run Prime on both cores and use 100%, and my temps on Orthos run about the same as dual instances of Prime.
I tried OCCT and it was fine for 30 min, ran my CPU cores at 100% however it did raise my CPU temps a full 7c.
Strange how all these stress programs push the CPU to 100% but all differ in the Temps they produce. That translates to more work being done or less work being done by the processor even though it is theoretically already doing all it can at 100%.
My point is, a stress program could probably be written to crash a perfectly normal CPU almost instantly. Perhaps what we need is a new or different way of defining what 100% CPU usage means. On the other hand perhaps some of the test programs push the CPU beyond 100% (if in fact that’s possible). 100% of stock settings equals one thing while over clocking that same CPU to 1ghz over it’s rating, very common now with a CD2 then 100%, in my mind at least means something quite different.
Thoughts?
Even if SSE/SSE2 and FPU are, imho, the very same unit, using SSE you activate many more transistors, and thus generate more heat Sorry if it isn't, just ask, i'll try to clarify But this is really technical stuff, and i tried to stay as simple as possible.
Yes. This is because, mostly, of branching prediction and pipelining, and the kind of instructions you use.
In some cases, the CPU needs to wait for the result of an instruction before starting his work on the next one. This happens, for instance, in branching ("if" instructions).
During this wait time, your CPU does virtually nothing, it is just waiting for the result of another instruction. However, during this time, nothing can be executed.
So, in all those cases, CPU will be reported as busy, even if it is waiting for a result, and thus, doing nothing.
Obviously, doing nothing is less likely to generate heat than computing. That's why some programs are more stressfull than others.
As OCCT checks for computing errors, i can not avoid all of those cases. That's why i will always be less effective than Intel TAT, which is just loading the CPU, and not checking the results. as OCCT is doing verification, it is loosing a bit of efficiency.
Moreover, some instructions will generate much more heat as they will activate more transistors than others. for instance, it is much more efficient, for heat geeration, to compute using the floating point unit, rather than the regular ALU (basic additions). The first one being much more complicated and using much more transistors than the latter. And it is much better to use SSE & SSE2 than FPU
So it depends on alot of things. in summary :
*Which unit you are using
*How many bubbles you are introducing in the CPU
OS-wise, the CPU is always viewed as 100% busy.... but that doesn't mean 100% of the CPU units are busy, nor even it is computing as its full capacity
PS : i don't if that message is clear... english not being my native language
