""And yet where are the super-fast PPCs?"
"it is possible that they don't exist for the same reason 1.13 ghz p3 doesn't exist but a p4 can run at 1.7 ghz at the same process. it is possible to design the micro-architecture to do more in a cycle instead of frequency scaleability."
Sure, this is one way those of us with 400 MHz G4 Macs explain to our Windows friends that our 400 MHz (or 500, or as fast as 733) machines can actually do Photoshop tasks as fast as a nominally faster Intel CPU. However, this argument was a lot more convincing when Moto/IBM were shipping 300- and 400-MHz parts and Intel was at around 600-700 MHz. Now that Moto/IBM seem stuck at the 500 MHz range for most shipping products, and AMD and Intel are up at 1 GHz or even 1.4, the argument is being lost on the Moto/IBM side.
So, I repeat my point, phrased in terms of the "more in a cycle" point you raise: If Motorola and IBM are being beaten on so hard by Apple to get faster versions of the PPC out, and if Moto and IBM risk losing the PPC architecture completely because of the perceived lag in speed, AND if SOI is so fast and so manufacturable, WHY, THEN, is SOI not being used to boost the speed of the PPC? Why is IBM not selling a 1.5 GHz version?
Between the "more stuff per cycle" and a faster clock rate, IBM and its partner Motorola, could presumably stave off the onslaught from the Intel processors.
Could it be that very large die sizes (even with a shrink to .18 or .13 micron) are just not yielding on SOI? That would be my bet.
--Tim May |
