[vega24]

I'm taking the CompTIA A+ course at New Horizons and we were talking about processors. Which is faster(more powerful), a RISC or a CISC processor?

Well, I answered it this way, which is true. "AMD and G4 processors are RISC and Intel's are CISC. Look at any benchmark compairing AMD and G4 processors against Intel and AMD and the G4 always do better, even though they have a slower clock speed." Oh, man. I shouldn't have said that because the Intel fanatics went of on that same old discussion. They either couldn't understand or didn't want to hear how more MHz doesn't mean better performance. So, then we had to go back and talk about the Data Bus for each processor, the difference between RISC and CISC and everything else. Needless to say, we wasted half the frick'n class, however; the Intel fans in that room left with their tails tucked between their legs lastnight.

btw/ we got to the chapter about Firewire and on the side of the page it says "Apple currently receives a $1 royalty for every Firewire port." I thought that was pretty interesting.

[M�lum]

You did a good job in converting those pagans!

[Nimisys]

dumbass... all intels have been risc internally just like the AMD's ANd the PowerPC since the Pentium Pro... they run CISC in emulation.


as for the G4 vs INtel chips... with the exception of photoshop filters the Intel cores do a remarkable job of betting out the g4 because of frequency even with their low IPC...

High IPC means dick if you got no frequency... As much as you may not like it the P4 is a remarkable chip, just as is the G4 and ther Amd's... hell cyrix was amazing (at being a space heater) on owns right.

try not be a zealot all the time.

[vega24]

Unlike you Nimisys, I don't feel the need to insult others and I won't be suckered into responding in a similar way.

The Pentium Pro had a internal architecture that was RISC, but it used a CISC to RISC translator. It was designed to satisfy the demands of 32-bit server operating systems and high-end workstations, not home PC's. No other Intel chip shared in it's RISC architecture.

The P4 is nither RISC or CISC, it's what Intel calls EPIC, which is similar in design to VLIW. The only thing remarkable about this chip is it's ability to scale up in clock speed, which it needs.

If the G4 was scaled up to 1.4GHz it would literally crush the current line of P4's much in the same way the Thunderbird has already done.

If you want to argue and call people dirty names, please get you're facts straight first.

[The Godfather]

Whether RISC or CISC is the fastest one can't be answered by just citing 3 examples: Pentium, Athlon and PowerPC. A theoretical question like that can not be fully satisfied with an empirical solution, you need to prove it with mathematical certainty. I don't think it is possible to ever know.

The problem gets harder due to the fact that the PowerPC started as a low transistor-count RISC but with Altivec instructions, it is converging to be as CISC as a Pentium.

You can compare chip makers' engineering philosophies: Intel and Athlon do CPUs for personal computers and they take marketability VERY VERY seriously. Intel and AMD would never let themselves be left behind in the MHz race because most buyers know it is important. Motorola, however, doesn't mind too much because their biggest market is in the embedded computer world, where watts count more than GHz.

It is very likely that the low IPC in the Pentium 4 is intentional. Sacrificing IPC for a faster clock is a good engineering move if your final result is higher performance, and you get to sell more. The P4 is not the case, it is a horrible chip now. However, when bugs are fixed and all the planned features are implemented, the P4 will be the BeOS of CPUs.

It is also very likely that Motorola's low clock is intentional too. They want to keep their product as low power as possible so they don't care about MHz. Unless somebody corrects me, power use in a chip is proportional to the number of transistors and the square of the frequency. Motorola will never want to sacrifice IPC for high clock speed.

CISC and RISC don't matter. The company with the best engineers will, in the end, win (unless the marketing dept. sets their foot on them).

I think these 3 chipmakers are equally freaking geniuses, but I would get an Athlon first.

[ 12-01-2001: Message edited by: The Godfather ]

[vega24]

Unless somebody corrects me, power use in a chip is proportional to the number of transistors and the square of the frequency.

The power usage of a chip depends heavily on the engraving size of the transistors on the die. Using the same amount of transistors, a chip manufactured using a .25 micron process uses more power than one made at .18 microns. Smaller transistors provide less resistance and therefore require less current which reduces the wattage and heat. Smaller transitors will allow a chip to run at higher clock speeds. Also aluminium transitors have more resistance than copper transistors and will cause a chip to use more power and produce more heat.

[Raman]

Originally posted by M�lum:
<STRONG>You did a good job in converting those pagans!</STRONG>

Isn't A+ "certification" for people that want to work at BestBuy and CompUSA fixing PeeCee's?

[The Godfather]

Originally posted by vega24:
<STRONG>

The power usage of a chip depends heavily on the engraving size of the transistors on the die. Using the same amount of transistors, a chip manufactured using a .25 micron process uses more power than one made at .18 microns. Smaller transistors provide less resistance and therefore require less current which reduces the wattage and heat. Smaller transitors will allow a chip to run at higher clock speeds. Also aluminium transitors have more resistance than copper transistors and will cause a chip to use more power and produce more heat.</STRONG>

I will add that to my formula:

generated heat is proportional to the frequency, dye size and voltage.

Given that dye size is proportional to number of transistors and fabrication resolution.
Empirically, overclockers have always increased CPU voltages before cranking up the MHz.

[Scotttheking]

Originally posted by Raman:
<STRONG>

Isn't A+ "certification" for people that want to work at BestBuy and CompUSA fixing PeeCee's?</STRONG>

No, it's way more then that.
Too tired to explain.

[Nimisys]

Originally posted by Raman:
<STRONG>

Isn't A+ "certification" for people that want to work at BestBuy and CompUSA fixing PeeCee's?</STRONG>

o its an overly easy to get certification trying to be shoe horned intot he PC equvalent of ASE in the auto world by HR depts and marketing firm world wide trying to make since and control what they know nothing about.

i got resepct for the ASE for theirs is basicly a masters in auto tech... the A+ however can be acheived easiyl without ever opwing a machine up, and gets no respect from me