Alright I can't take it anymore this thread won't die and it keeps taunting me on the front page....

Well I can tell you that no self respecting computer scientist would argue real world performance in terms of "FLOPS potential". This is the realm of marketing droids of various hardware manufacturers.

I for one stopped reading the OP when he said the SPEs were DSPs. No. Follow his IBM link, then wiki DSP. You will find that they are not the same. Not even close. These people are just putting words --whose meaning they don't understand-- together to support their POV in this bizarre manifestation of identity politics the console wars have evolved into.

You can safely ignore the OP as it's a pretty safe assumption that it's based purely on emotion.

Now look at all the posts that say the SPEs are floating point blah, not integer blah. Lets pretend for a moment that you can't read the damn pictures on the posted link which has a box in the even pipeline which has both floating point and fixed point arithmetic in it. What exactly do people think floating point numbers are represented with? You can safely ignore these posts as well.

Nobody, certainly not from those emotionally attached to Sony, seem to mention that the xbox360 has dedicated hardware for AA. I take that back. Some of those who are emotionally attached to MS refer to it as 10MB of some kind of super special magic ram.

What does it matter how biased or unbiased any opinions offered here are? I see very little evidence that understanding of the facts have significant influence on any of the conclusions I read (didn't read every post, but I'm pretty sure I read enough). Therefore, if it wasn't for bias the conclusions would just have a gaussian distribution. Thus I think I can give you a picture of what your bias free vgchartz would look like as we can closely emulate this theoretical universe by flipping a coin. Heads person X says the CBE is "more powerfull". Tails person X says Xenon is "more powerfull".

@alephnull

Agreed... However its the only thing that makes 'sense' for the crowd. :)

Well, there are good points like usage of SPUs as CELL in PS3 do not use all of them. Other than that it theoretically BS that has very little to do in real cases.

Yes, daughter die is 'made' for AA, but you can also use it for other stuff too. Anyway you are right. Its not a some kind of super special ram and I admit I was wrong. Its seems to exists for postprocessing pixels etc.

(Heh, hopefully you weren't referring to me as emotionally attached to any of console as I am more like PC purist/elitist. If I am wrong about something just correct me, I try to remember it next time. :) )

http://www.extremetech.com/article2/0,1697,1818140,00.asp

Both are quite powerful machines. Both of them have their own strengths and both of them can do some things that other can't.

PS. Its good that your back. I was getting tired with MikeBs propaganda posts. Hopefully youre still around when I try to make something for PS3 as I am sure that you might be able to help me with it, if I have probs. ;)

(However it seems that with current prices and pricecuts it will take a long time before I own one.)

I'm not crying - see point 1) of my post!

I just though maybe, just maybe, there was really an unbiased view of each machine I could peruse with interest.  But no.

Although I will admit that there's a bit of LBP that has me crying with frustration right now but that's another story.

As for your post, that's more or less the conclusion I've come to.

I just re-read what he said and the only good-ish points I see involve the GPU. The rest are nonsense.

The stuff about the OS's memory footprint on these machines is just plain silly as even if we had a game machine with a 96 meg kernel as those pages not in use would be swaped out. Completely rediculous! It's almost a certainty most of that consists of user space apps, maybe I just haven't been paying attention for awhile, but I don't think I can build a linux kernel that big even if I compile every possible module into the stupid thing. And seeing as both systems run the same flavors of the same kernel anyways (unless MS finally ported some stripped down version of NT to Power with the new NXE patch thing or whatever its called out of embarrasement).

The stuff about the SPEs being floating point cores (someone should tell him that most DSPs are not floating point friendly, btw) is nonsense. The fact that one is reserved for the OS should say something about that. The biggest problem is branching, and that's not THAT big of a problem, basically they just start loose a bit of there performance edge over normal cores. And you'd be surprised how many tricks there are to avoid branching. Also the SPEs are quite good at searching trees believe it or not. Anyways alot of those integer ops you are going to want to convert to bit-perumtations/ops. Although, I will admit I initially thought the same before I started to get decent experience coding on the damn thing.

I actually wasn't referring to you, but someone in the middle of this thread (If I had known YOU had said something to that effect I would have been way more condescending... I kid, I kid).

Anyways my thinking has evolved on this daughter die :P (Yes, I am a terrible hippocrite)

Initially I figured this thing was just a simple ASIC hardwired with a bunch of ALUs surrounding the DRAM cells connected to do fast AA. All of these "reviews" (which are usually just lame attempts to parrot back what manufacterer X said about the new thing they are selling) seemed to imply this by stating 4xAA was "free". Then I came across this..

http://forum.beyond3d.com/showthread.php?t=4624

So now I have a new theory. In order to make the GPU fit into the unified shader model without loosing performance, they moved most of the things that would have stood the most to loose performance-wise in the switchoff to the unified shader model eg. most of the algorithms that use pixel shaders off onto the daughter die. I bet there is more logic on this thing than disclosed and game devs are using it as basically a pixel shader core. That's the only explanation I can thing of for all these xbox games not having AA. You realize the irony of this don't you

Of course this is all just speculation as I've never actually coded on the 360. For some reason it lack linux support at the moment... hmmm.

No, I was refering to the usual suspects, selnor, the captain kirk guy, his sidekick the anime girl with an ak, crazzyguy or something, your arch-nemesis MikeB, the guys whose names end in J, etc.

I am pretty much a turned based strategy gamer at heart so, yeah. Ok, that's not entirely true I find myself going back to star control 2/(Urquan Masters) rather frequently.

Heh, don't lump me in with MikeB. I was an AtariST guy, if you think xbox versus ps wars are bad, the Amiga versus ST wars were worse since it was more like a civil war.

Yeah, I think sony was planning on having the 45nm cell by now from the old road map here:

http://www.lanl.gov/orgs/hpc/roadrun...7%20(LAUR).pdf

But here's the newest roadmap leaked...

http://img411.imageshack.us/img411/9205/roadmap1kh0.png

Notice how the pushed back date coincides with when the rumored pricecuts will take effect.

Also notice how the prototype cell2 will be done around when sony will probably start on the ps4.

PS. I'm usually lurking, just to busy to post.

I honestly can't see any difference between the Xbox 360's and the PS3's graphics. And I'm sure most others can't either.

All CPU Gflops ratings are theoretical Maximum's and generally as noted you will be very lucky to get 50% of this actual figure in most cases ! .
As for the Xenon vs Cell being dramatically easier to use - Game software even on the PC where dual/quad core is now common is rarely using more than 1 cpu fully ! - it's not easy at all - parallelisim adds complexity - but once you do good parallel code the number of cores will provide a linear speedup in execution - 2 cores is twice as fast as one and 4 is twice as fast a 2 and 8 is twice as fast as 4 - this is how GPU's are so fast but so limited by design usually - it's difficult to write good parallel code even for simple repeditive calaculation of graphics - but once you can do it 7 cores is as easy do as 3 or 4.

"When physically tested however for theoretical peak performance, only 155.5 GFLOP’s were actually achieved (see Table 4) with a total efficiency rate of 75.9%.
Because of manufacturing yield issues, the PS3 only uses 7 SPE’s with the theoretical peak for the PS3’s Cell processor being reduced to 176 GFLOP’s, each running at 25.12 GFLOP’s. Utilizing the same 75.9% efficiency, it is easily interpolated that the PS3’s Cell CPU will only be capable of 133.6 GFLOP’s.
The Xbox 360 has 3 general-purpose 2-threaded CPU's, which generates a proven 115.2 GFLOP’s which is dramatically easier for developers to utilize."

Yes the 115.2G figure is also a theoretical Maximum ( is the post author that gullable or just a fanboy ?) - and unlike the cell the efficinecy of the Xenon is even lower than 75.9% - Cell efficiency is generall higher than other CPU's.
Xenon probably manages around 50G actual (being generous here) - thats assuming all 3 cores - 6 threads are being used - And a carefull balanced mix of Altivec and Floating processor code on 6 threads - which i doubt would be the case -It's just as difficult as using the multicore Cell SPU's to get peak performance from the Xenon like this.

All up IBM do pretty good CPU's for all 3 current consoles (Wii,360,Ps3) - but the Cell is the pick of the Bunch - if only Sony were as good as MS with software support .

 This

It would be valid if it were remotely true. It has already been proven that the PS3 is more powerfull. Is the 360 powerfull as well? yes. Does it have some great looking games? yes. It does not however have the best looking games. That title is still held by the PS3.

The 360 needs to eat up a good chunk of its processing power decompressing its data. The PS3 does not. The developers have already managed to get better looking games on the PS3 without even using all of the SPEs.

http://faculty.ksu.edu.sa/mhussain/Publications/IJMS2005.pdf

Their paper deals purely in normals for their error metric, so any vertex movement that dramatically changes the normals of all triangles connected to it gets flagged as "high error", and hence ends up very far down in the edge collapse priority list. Since it's just normals, no extra artist involvement is needed. The reason it works at preserving contours is that a contour implies local geometry at somewhat different angles from immediate surrounding geometry, and hence any play in those verities generates a high error value. Like if you have a cube mesh, collapsing an edge starting from a corner vertex would cause mass error, as all surrounding triangles suddenly change their normals dramatically. Note as well that their technique is purely subtractive, it's meant to start from the best quality mesh and work down from there.

If left unchecked, such as if you try collapsing 70% of all edges, then for sure in some mesh contours will disappear. But you can mitigate this by setting an error cap. For example, collapse edges until the lowest error exceeds a certain 'max allowed error' threshold, then stop. So if you fed a cube mesh into the system, it would emerge on the other end untouched, however a piece of terrain may drop 40% of it's verts, etc...

Source: http://forum.beyond3d.com/showthread.php?t=50486&highlight=tessellation

By, Joker454

Laymans terms: Faster, Less memory, Works on SPUs.

August 2, 2007 - In a conference call set to begin at 5:30pm EST today, Rockstar Games will announce that it will delay the release of its highly-anticipated title Grand Theft Auto IV. Previously slated to hit stores on October 17th, the game will now make its way to gamers some point early next year.

In technical, company jargon, Rockstar has moved its release from the fourth quarter of fiscal 2007 to the second quarter of its fiscal 2008. In real-world terms, this means the title will ship sometime between February 1st and April 30th, 2008.

The reason for the delay is simple: Rockstar North needs more time to polish it. Given the game's exceptionally high expectations, Rockstar has chosen to push back the release a handful of months rather than rushing the title out to market.

Sam Houser, founder and executive producer of Rockstar Games, had this to say: "The new consoles are allowing us to create the Grand Theft Auto game we always dreamed about. Every aspect of the game and its design has been completely transformed. The game is huge and is pushing the hardware platforms to their absolute limits. The top engineers from Sony and Microsoft are working closely with the team in Edinburgh right now, helping us to fully leverage the power of both platforms. As always, our goal is to surpass even the wildest expectations of the game's fans, and to create the ultimate high definition video game experience."

So unless the MS "team" is better than the Sony one, both platforms ran the game as well as it could be run.

Source: http://forum.beyond3d.com/showpost.php?p=1234932&postcount=954

So by the one objective standard the 360 runs better (Higher frame rates/resolution/AA applied) Install issues are another matter entirely.