Full Load All The Way: AMD Vs. Intel, Continued
The idea of conducting a stress test began to take shape at a time when comparison tests of processors in conjunction with appropriate motherboards were often fraught with serious problems . On several occasions the lab engineers in Tom's Munich lab were nearly driven to despair. The endurance test contenders have now been set up in a separate room in the THG lab in Munich. We figured there could be no better candidates than the new platforms from AMD and Intel, outfitted with a host of powerful components. The special feature of this test is that all peripherals were purchased from retail computer stores, rather than using production samples, to avoid introducing potential unknown errors.
High time to ask the burning question: which of these two platforms, both of which deliver truly impressive performance stats in short tests, can withstand the unprecedented endurance test?
Both the AMD and Intel platforms have been equipped with state-of-the-art components. The big question is: which system can pass the stress test without crashes or other serious glitches?
The key issue with Intel remains heat dissipation. P4 processors with a Prescott core (3.6 and 3.8 GHz) in particular become extremely hot - so hot in fact you could jokingly call the Intel Pentium 4 "the world's smallest heating plant". (Insiders used to talk about the "cogeneration unit" - meaning in this case, however, the old 1.4 GHz version of the AMD Athlon 1400 with a Thunderbird core.)
And then there's the question that nags both would-be buyers and enthusiasts most of all: can the system hold out for days on end running at full load without crashing after just a few hours - as required in the "business environment"? After all, it was only a few weeks ago that we published the article The P4-560's Heat Can Crash and Kill , in which we showed that the Intel Pentium 3.6 GHz is anything but stable even in a system that complies with all of Intel's specs.
Of course, we always assume a maximum CPU load - a situation rarely if ever encountered by normal users performing normal desktop tasks. So the focus here is clearly not on "casual users", but rather those who run applications that tax computing resources to their limits.
