Originally posted by gopher
jrv, you think you can compare Apple's to PCs. Read http://www.macvspc.info/
The speed is not in the FSB on the Mac. It is the L3 cache, and the RISC processing, the lower number of pipeline stages making processes finish their time on the CPU faster, and the lower heat in the CPUs themselves being actually smaller CPUs. Go back and read my prior posts on the issue. The problem with speed is in software and not hardware.
Also for Mac OS X speed visit my FAQ:
http://www.macmaps.com/Macosxspeed.html
Then explain to us why the latest single processor 3.06 GHz Pentium 4s beat the crap out of dual processor 1.25 GHz, at a factor close to 2.
Have you ever seen this infamous one:
http://www.digitalvideoediting.com/2002/11_nov/reviews/cw_macvspciii.htm ?
FSB is always an issue. The faster the FSB, the more memory bandwidth it has. You should take a look at the PowerPC 970. That's right 900 MHz dual'ed up, 450 MHz per channel.
Pentium 4s are at 533 now, four 133 MHz channels. Feeding gigabytes and gigabytes more data in the processor than the G4 ever can take. However, they're approaching the 800 MHz bus, which is comparable to the 970s, because IIRC, 970 has more instructions for memory transfers and therefore its really effective around 800 MHz. The more bandwidth you have, the faster you'll be able to work with huge datasets.
And what's that? Pentium 4 3.06 GHzs don't even have L2s. L3s only matter if you have slow memory, READ: PC133/gimmick DDR. Dual-channeled DDR400 can very well beat DDR L3 in the Power Mac by a factor or more given the right circumstances. Besides L3 only helps if you stay in one application and one dataset. Huge databases [gigabytes, mind you] or even large Photoshop files [100 MB] won't fit on the L3 and therefore unable to take full advantage of the L3.
RISC is very irrelevant these days. Did you know that the Pentium Pro/II/III/4 are RISC? The Athlon is RISC? Yep they are, they just have CISC interfacing the RISC, which is a great design decision, IMO. Make it easy to program for with CISC, but still maintain the power of RISC.
Pipelines are important, yes, but as Intel has proved, you can ramp up the processor with long pipelines and get extremely fast processor speeds.
That guy who runs that site needs to be informed, IMO. He says NT hasn't changed that much since its first incarnation. ?!?!?!?! Ok, his credibility has been lost.
He also points out that when "Macs are properly set up, they won't have problems. The same is true can be said for the PC.
So OS X now has the industrial strength benefits of UNIX (preemptive multitasking, full symmetric multiprocessing, memory protection, etc.), combined with significant digital media architectures (QuickTime, OpenGL, PDF, etc.), and all this power has the world renowned, intuitive Mac interface. Quite a package!
The NT platform has been delivering these since 1993. I think we all can agree Apple was late to the game, but they do indeed have a very nicely and well done OS. [edit: oh don't forget about the free Unices... here long before Mac OS X--guess what Mac OS X is based on? FreeBSD! And that is primarily an x86 OS.
I don't think the solution to today's problems are bitching and moaning about either Mac or PC. I think today's problems should be heeded to compatibility, standards--of which both Mac OS X and Windows XP are going for. A mix-and-match environment is best, because of course, if everyone used Macs, there would be just as many "problems" with Macs as there are on PCs nowadays.
Can't Mac and PC users just get along and interoperate. That's what I do.
Isn't the original topic about new Power Macs?

And yes, bring the damned prices down and up the performance!
[edit: a deceiving spelling error didn't catch my eye]