Unless those Athlon systems were Palominos, I don't see why Mr. McComas expected any throttling in the first place, given that T-bird doesn't have a thermal diode.
I think the point is, (according to his tests) the P4 did throttle indicating an active throttling mechanism. Now, the very strong implication is that had Intel not provided this throttling mechanism, something very bad (lock up/system crash) might have happened. Remember, this was not software specifically designed to tax and crash the processor. This was Quake3. Do you agree with this?? Please answer this specific question. Also, he claimed the Athlon systems under similar testing conditions did not show any throttling which is expected since these systems have no
active means of throttling. But, it is also clear that the systems did not lock up or crash either. It will interesting to perform the same tests on higher MHz Palominos.
Also, it seems the throttling went away with one of three trivial fixes.
Give me a break. It was clear from the article, that this throttling was seen on more than one system and processor. He also made it clear that the throttling mechanism seemed to vary quite a bit among different systems and processors.
Funny you should mention that, since I honestly can't tell the difference between 60 FPS and anything higher. And I don't think anyone else can tell the difference, either
Again, you didn't respond to my point. The point was that, according to his tests, the throttling mechanism kicked in with even of the shelf software. This was not any exotic benchmarking application specifically designed to test the stability of the processor. Given this, do you now think that this throttling issue is more of a concern and must be understood??
However, there are plenty of people out there who will be using Pentium 4 workstations for hefty-duty applications running 24/7. Surely these people would be the first to notice any instance of throttling, no? I'm sure my coworkers and I would, given that we'll soon be getting 1.7 GHz P4 Linux boxes for running our simulations 24/7.
Well, I think any CTO worth his salt would want to fully understand this throttling issue BEFORE he committed a large IT purchase for P4 workstations. Personally, I'd drag Intel in and force them to explain in excruciating detail how and why this throttling takes place. I'd want to see demonstrations of it occurring on both Intel test applications as well as on my internal applications. I'd want to have a reasonably clear picture of the implications of this clock throttling mechanism to the overall decision of going with these P4 systems. Of course, I'd do the same with competitive systems. I don't see where Intel mentions this possible clock throttling mechanism and its implications anywhere in its marketing of these systems. If I was a corporate CTO, I wouldn't want to be the first to find out.
THE WATSONYOUTH |
