The 360 has 512 of general purpose memory which is 1/8th a decent gaming rig.  The 360 has 0 dedicated video memory which pretty much guarantees you of running pc games on low settings.  Both the 360 and the PS3 are good gaming machines that will always fail comparisons to a pc gaming rig.

Hmm, maybe I wasn't entirely clear. I wasn't saying that the PS3 isn't a good machine - it clearly is.

My point was that the PS3 has a much better processor than it does RAM, whereas for gaming you traditionally need alot of RAM, but the processor isn't as important. And I was using that to explain why in pretty much all PS3 games, only 60% or whatever of the SPE will be used, but that we will never see games that do use 100% because of the insufficient RAM

Although I would agree that both systems could utilize more RAM, as we move to a multi-core/thread environment the amount of memory is no longer the only design factor.  The increase in excecution cores in the traditional pc design brings an increase in memory latencies due to limitations with the bus architecture and more importantly the speed of the memory.  This is why the comparisons between the xenon and the cell don't necessarily work.  If you look at the specs of each using the standard metrics of amount of memory, core frequency, and clock speed both systems would look similar.  If you take into account the overall design including bandwidth and memory speed, the cell would be viewed more favorably. 

In short the amount of memory becomes irrelevant if the access mechanics are not adequately sized. 

Statements like X game uses Y % of CPU power mean absolutely nothing.

One could write an infinite loop, catch any integer overflow exceptions and reset the counter back to zero and that would use 100% of the PPU or SPU continously. The same code could be forked across all SPU's and now the Cell would be completely maxed. Using all CPU cycles does not equal good or efficient programming. Optimization is about using less cycles not more. The PS3 is not impervious to inefficient programming. I am not saying that GG are inefficient programmers, what I am saying is that their claim does not give any indication of how much untapped potential is still left in the PS3. Unless Killzone 2 is 100% optimized in every aspect of development wrt to the ps3 architecture, it is entirely possible that even more than 40% of the PS3 has been untapped.

Let's pretend I have a ferrari that goes 200MPH. Unfortunately 99% of the time I can't drive at full speed utilizing the full power of my Ferrari's engine. Why? Because variable levels of traffic, stop lights, toll booths, and cops limit how fast I can go. Just like the Ferrari engine the Cell will most of the time not be fully utilized. Most of the time system bottlenecks are not CPU bound. Slower drive speeds, slower or limited availbility of RAM, and shared data that other threads are operating on result in wait states for idling threads.

Just my 2 cents

@MisterBlonde: That is all correct.

I think the thread came to this conclusion on page 2.

A better example of inefficient programming would be, running 12 vgchartz.com tabs in internet explorer 7.  This seems to consume 50% of cpu that for some reason brings Windows xp to a crawl.

 Yes it certainly does. And I think Uncharted 2 is going to be amazing.

@NJ5 and dbot

Thanks for pointing out page 2...I missed that the first time and ended up repeating what you guys stated. I was more or less replying to statements like Kameo, Gears, and MGS used X% of console power that I saw later in the thread.

As far as the 12 vgchartz example, that is a good example also. Could be an IE problem, vgchartz website problem, or 3rd party plugin like Flash or something. Or a combination of 2 or all 3. You could attach windbg and find out pretty quickly. Just google windbg if you are interested in finding root cause.

With my example I was just trying to point out that utilizing all resources is for the most part not considered a good thing, and with bad programming is easier than one would think.

*See Gebx sig*.

Here's my perspective from before this thread even started:

"

*Online* Message Add Friend Posts: 3,458 vg$ 2,073.26 (12/05/08, 13:13)

Further game engine optimisations can bring down the 60% SPU load figure considerably. So potentially there should be even more headroom to do additional stuff. Current processor loads can be easily determined through benchmark software, but future wizardly coding optimisation cannot be quantified.

http://www.vgchartz.com/forum/thread.php?id=51752&page=1

It's nice to see NJ5 changed his perspectives on this issue, regarding a much earlier thread.

http://www.vgchartz.com/forum/thread.php?id=49421&page=5

BTW, like I suspected in that thread this diagram would be obsolete.