Zmierzch: Intel - hyperthreading "PRESCOTT"

[Gość: bzykbzyk.]

In a processor that's as deeply pipelined as Prescott, waiting on a cache
access means multiple dead cycles, especially considering the fact that
Prescott's L1 and L2 cache latencies are worse than those of Northwood
(significantly worse in the case of the L1); and waiting on a main memory
access means lots of dead cycles for a processor that's running at upwards of
3GHz.

So what the Intel rep was really saying was this:

Because we got obsessed with MHz as a marketing number and made Prescott's
pipeline so ridiculously long, Prescott benefits much more from a latency-
hiding technique like hyperthreading than a saner design like the Pentium M.

Intel's Pentium 4 architecture (a.k.a. Netburst) is predicated on the
assumption that the former approach — shrink the core size and raise the
clock speed — will translate into both better performance and ----->>>>>>
better sales, because it's easier to sell MHz than it is to sell added
functionality. <<<<<<------

The problem with this approach is twofold and can be summed up with two
terms: transistor leakage and power density.
......
So the dual-core Prescott design is Intel's way of making lemons out of
lemonade.
Because Intel can no longer rely on clock speed increases to bring
performance gains, they plan to take their flagship processor design and head
off down the same multicore path as IBM, AMD, Sun, and the rest of the
industry. And of course, the rest of the industry is going multicore because
of the power-related problems outlined above. The whole industry is suffering
from the same headache, and now everyone has settled on the same cure; Intel
was so addicted to MHz that they were a bit late to come around, but come
around they have.

The perceptive reader will have picked up on the larger point that I'm
making: a dual-core Prescott is a hack, made necessary by the fact that the
single-core Prescott is going nowhere fast.

So Intel is hoping that a 64-bit dual-core Prescott with hyperthreading will
make for an attractive workstation offering, especially since they'll still
have a clock speed advantage over whatever AMD has out at the time.

Whether they'll have a performance advantage is hard to tell, but if this
chip does see the light of day and if MHz still matters at all, it might help
Intel squeeze more life out of what is an otherwise ----->>>> moribund
architecture. <<<<-----








arstechnica.com/cpu/004/prescott-future/prescott-1.html