Quote:
Originally posted by EvilCrow
OK then...
Why are they saying it's the fastest personal computer?? |
Computation is faster. Clocks are a big part, but totally
not the whole story.
G2-G4 PPC (RISC) chips used a setup that kind of equates to the old math coprocessor setup for 80486 and lower - a second data processing area, sending instructions to the AltiVec chip.
On the G2-G4, only certain instructions are processed in the AltiVec, but the instructions that are processed are very complex ones that x86 chips do not farm out. As I understood the issue, the AltiVec chip would execute a routine in something like 1/4th the clock cycles of the main CPU. Hence, AltiVec makes PPC execute certain code with fewer clock cycles versus x86. So, if you send the CPU the instructions which AltiVec is going to process, you pick up speed.
I have not fully investigated the G5 - it seems to have silicon to handle the AltiVec routines (see Velocity Engine statement, below).
Oh, and 1GHz FSB??? That is
wild!
G5 Architecture discussion Quote:
Fastest bus in the west
The G5 features a scalable design that enables it to run at clock speeds up to 2GHz. This represents a 600MHz jump, the largest in PowerPC history, over the fastest G4 processor at 1.4GHz. But all the megahertz in the world wouldn’t mean squat if the G5 were stuck talking to the rest of the machine at the 167Mhz bus speed of the Power Mac G4. That’s why the G5 features two unidirectional 32-bit data paths: one traveling into the processor and one traveling from the processor, unlike previous designs. Its frontside bus works at speeds up to 1GHz for an astounding 8 GBps of total bandwidth. That makes it 200MHz faster than even the latest Intel 875 chipset, which sputters out at 800MHz.
|
I know that the G3 CPU cards (for a Mac 7500 that I own) can be purchased with L2 cache of 2 megs.... Intel chooses to make low cached CPUs; usually 512k or less, for consumers.
Finally, the G5 isa 64 bit CPU, in a consumer box. nyaah nyaah.
G5 page at Apple Quote:
|
Apple collaborated with IBM to leverage this industry-leading design for the G5, combining an optimized Velocity Engine with a new superscalar, superpipelined execution core that supports more than 200 simultaneous in-flight instructions.
|
Ouch!