AMD combats ‘megahertz myth’ with Athlon XP 2000+
Battle of the CPUs
AMD today launched the Athlon XP 2000+, the industry's "highest performing desktop PC processor". It costs $339 in OEM quantities of 1,000.
Note the company does not say "industry's fastest" - when it comes to processing cycles, the Athlon XP 2000+ , at 1.666GHz, clocks up rather fewer per second than Intel's Pentium 4 2.2GHz, also launched today. The megahertz gap between Intel and AMD is set to widen, but AMD has two tactics to level the playing field.
The first is benchmarks, published here which purport to prove that AMD retains performance leadership over Intel. This plays well with the overclocking/games-playing/self-build/hardware review site-reading constituency which AMD more or less owns.
The second tactic is rather more controversial - the use of part numbers performance ratings to suggest a higher clock speed than is actually the case. This fudging helps AMD fight its corner in the PC consumer sector, another area of AMD strength, with part numbers suggesting performance equivalence with Intel megahertz figures.
We think this is a legitimate but risky marketing tactic: the performance crowd, and corporate buyers can check out the benchmarks; at the same time AMD profits in the mass market, while none of its customers in this sector actually suffers.
However, the use of performance ratings does expose AMD to sniping from competitors and tech publications. The chipmaker aims to counter this by exposing what it calls the "megahertz myth" through industry dialogue. Long term, the company wants to promote the True Performance Initiative (TPI).
"This is intended to define a new, more accurate measure of processor performance and assist customers in understanding the benefits of PC performance."
In other words, AMD wants the world to accept benchmarks (as opposed to megahertz) will place its processors in the best light. There is nothing wrong with this: benchmarks are a moveable feast, and controversies over the performance of the Transmeta Crusoe, and the Pentium 4 when it first launched, show that standardised benchmarks are often not much cop.
But it is precisely because standardised CPU benchmarks, measuring real world performance or whatever, can always subject to attack, that AMD will, we think, fail in its quest to replace the megahertz myth with its own more 'scientific' TPI in the minds of the public. ®