Xbox 360 1 Teraflop of Performance Explained

I suppose a similar but fairer analogy would be a SuperMini car versus a human runner in travelling 26 miles through London at rushhour. One has far more power (about 240x in this example), but due to the bottlebecks and limitations of the road infrastructure it gets beaten by 10 minutes by a 1/4 horsepower bloke. Peak processing power, like peak horsepower, doesn't give the overall picture as to how well the hardware will accomplish it's tasks. Which I believe is what powderkeg was getting at.
 
Shifty Geezer said:
I suppose a similar but fairer analogy would be a SuperMini car versus a human runner in travelling 26 miles through London at rushhour. One has far more power (about 240x in this example), but due to the bottlebecks and limitations of the road infrastructure it gets beaten by 10 minutes by a 1/4 horsepower bloke. Peak processing power, like peak horsepower, doesn't give the overall picture as to how well the hardware will accomplish it's tasks. Which I believe is what powderkeg was getting at.

Yeah, but that's not necessarily an analogy that fits here.

People seem keen to polarise one system as "efficient" and the other as "inefficient" but I don't think there's as much contrast there as many people seem to think.
 
Titanio said:
Yeah, but that's not necessarily an analogy that fits here.

People seem keen to polarise one system as "efficient" and the other as "inefficient" but I don't think there's as much contrast there as many people seem to think.


Both systems are inefficient. The question is, is there a real difference between the two once the inefficiencies are factored in?

And we'll probably have to wait until 3rd-4th generation 1st party games are out to make that call.
 
london-boy said:
That's a completely useless comparison. One could say the 18-wheeler can carry a load that's many times bigger than your fast car.


Absolutely.

But that wasn't the job I asked it to do, was it?


Nothing to do with what's being talked out here, a car is made to be driven fast round the city, and an 18-wheeler is made to carry huge loads. Each does what they are supposed to do better than the other obviously.

And the 360 CPU was tailored specifically for a gaming console, while Cell is a processor made for many different purposes, not just gaming. Cell would blow the doors off the 360 CPU in doing weather pattern prediction models, but that isn't the job we are asking it to do, is it?
 
Powderkeg said:
And the 360 CPU was tailored specifically for a gaming console, while Cell is a processor made for many different purposes, not just gaming. Cell would blow the doors off the 360 CPU in doing weather pattern prediction models, but that isn't the job we are asking it to do, is it?

PS3 was the raison detre for Cell, despite its diversification into other markets. It can be applied in different areas, but so could XeCPU (although it is not, afaik..). The requirements of a processor for games are not completely disjoint from those of other areas.

I think it should emerge to be as well suited to games as XeCPU, if not more so. That's the vibe I'm getting at the moment, anyway. For both systems there is a degree of mapping your software to the hardware rather than the other way around, however (but there should be a payoff for that).

Performance is a different point also.
 
Better analogy

human runner versus throughbred horse in a 100 meter dash.

The Human WILL win the 100 meter (at least a fast human ;-)) but the horse will bust his ass in a quarter mile.

The Xenon is the human.
The Cell is the thoroughbred.

Now for GPU's the analogies are much more difficult.

While Xenos is an exotic roadster coupe like a carrera s, RSX is a high end muscle car (viper). Top end speeds dont matter because they each have governors... carrera is best for curvy roads that require agility yet control and high speed... while the viper is massive on a wide flat straight road....
 
blakjedi said:
Better analogy

human runner versus throughbred horse in a 100 meter dash.

The Human WILL win the 100 meter (at least a fast human ;-)) but the horse will bust his ass in a quarter mile.

The Xenon is the human.
The Cell is the thoroughbred.

Now for GPU's the analogies are much more difficult.

While Xenos is an exotic roadster coupe like a carrera s, RSX is a high end muscle car (viper). Top end speeds dont matter because they each have governors... carrera is best for curvy roads that require agility yet control and high speed... while the viper is massive on a wide flat straight road....

i'd take the human, he's able to do more than just runnning and i'd take the carrera S because it comes with much much more immenities, reliability, and style. :devilish:
 
blakjedi said:
Better analogy

human runner versus throughbred horse in a 100 meter dash.

The Human WILL win the 100 meter (at least a fast human ;-)) but the horse will bust his ass in a quarter mile.

The Xenon is the human.
The Cell is the thoroughbred.
I wasn't presenting an analogy for the two consoles, but showing how peak figures don't really mean much. My analogy was based ona real race broadcast on BBC a few weeks back. The main point is whatever the peak performance of a part, that's no guarentee of final performance to a given task.

If we were to liken XB360 and PS3 to automobiles I don't think anyone's in a position to say what each platform represents (though it may make an interesting topic!). Apart from pure HP which PS3 wins, we can't say which has better handling, acceleration, yadayadayada without having worked on the two platforms, and we can't say which would win any of a dozen different racecourses (where a racecourse is a different game genre or effect).
 
Shifty Geezer said:
blakjedi said:
Better analogy

human runner versus throughbred horse in a 100 meter dash.

The Human WILL win the 100 meter (at least a fast human ;-)) but the horse will bust his ass in a quarter mile.

The Xenon is the human.
The Cell is the thoroughbred.
I wasn't presenting an analogy for the two consoles, but showing how peak figures don't really mean much. My analogy was based ona real race broadcast on BBC a few weeks back. The main point is whatever the peak performance of a part, that's no guarentee of final performance to a given task.

Hey no fair I saw a man beat a horse on TV too... :p

The xb vs ps comparison were simple analogies based on the raw horsepower versus elegance arguments attributed to the two consoles. dont let it bugger you my friend! :D
 
Please can we stop with the comparisons?!

What next, X360 is Jordan and PS3 is Pamela Anderson? (Not sure who's bigger to be honest)
 
london-boy said:
Please can we stop with the comparisons?!

What next, X360 is Jordan and PS3 is Pamela Anderson? (Not sure who's bigger to be honest)

Jordan wins hands up
Pam wins hands down ;-)
 
Well, well... It looks like this thread is done, time to stick a fork in it...

The thread is now seriously off-topic. I'm not locking it since there's nothing offensive (Flamebait/name calling) about it, but let's either discuss the original topic, or let the topic die on its own.

A lot of words just to say, stop the off-topic jokes/ comparisons. :devilish:
 
OK OK vysez... Can someone explain non programmable flops and why RSX has twice as many as Xenos? For some reason Im NOT getting that part. Is it just spec inflation? Is it and estimated ratio? (ie 400 Programmable flops = 500 non programmable therefore 800 programmable flops = 1000 Non programmable?)

I remember J Allard using the term "targetted" to describe the teraflop performance x360... would that be the equivalent of z only pass? Lots of questions and the topic stays on course. :D
 
blakjedi said:
OK OK vysez... Can someone explain non programmable flops and why RSX has twice as many as Xenos? For some reason Im NOT getting that part. Is it just spec inflation? Is it and estimated ratio? (ie 400 Programmable flops = 500 non programmable therefore 800 programmable flops = 1000 Non programmable?)

However they (NVidia & ATi) count those figures, that's how they count them..there's little tranparency in the process. I'm sure technically they could sit you down and walk you through it, but they're not likely to do so ;) Don't worry about them. If you wish, just focus on what you can derive (programmable flops)..that's what a lot of people do.

blakjedi said:
I remember J Allard using the term "targetted" to describe the teraflop performance x360... would that be the equivalent of z only pass?

I think "targetted" was simply a flattering adjective thrown in there, it's not a technical term relating to anything. Targetted suggests something focussed etc. They just wanted to communicate the idea that this power is well focussed.
 
Back
Top