Well my interpretation of the Klamath tests is that it is a nice design, and a bit faster than I was expecting. Even giving up the fast L2 cache it consistantly outperforms the PPro. If it is available at speeds of 233-300, it may well be the fastest available CPU this year. Of course, if you run 16 bit applications, the M2 will be fastest by a wide margin. If you run heavy FPU work, who knows, but I would guess that Klamath or PPro would beat the M2 or K6 handily.
My guess (and it is only that) is that the M2-180 will rate as a K266+ (new rating system), and the K6-200 will rate as a K200+. Of course, an M2-200 would have a rating of K300+. Thus, rank ordering speeds:
1. Klamath 300, M2-200-K300+
2. Klamath 266, M2-180-K266+
3. Klamath 233, K6-225-K225+, M2-150-K233+
4. K6-200-K200+
5. K6-180-K180+
But what about an M2-187.5 @ 2.5x75? Or an M2-166 @ 2x83? It's too late to figure these, but I would guess that 2.5x75 would be as fast as 3x66, and that 2x83 might be that fast as well.
My guess is that Cyrix once again will be one step ahead of AMD in performance, and nipping at the heels of Intel, but this year they will all be fairly close.
Carl |
