Well, you can always find developers that will say the PS3 is amazing and you'll find other devs that say the PS3 is garbage.  You can find top tier developers that say both.  They could be speaking from experience or there could be other motivation.  Some programmers are very much idealists.  Some are tolerant.

I have two primary issues with the Cell when compared to the 360's CPU.  The first is that the SPE performance is misleading for general gaming.  Anyone who has worked with high end computing (especially Cray/SGI top-end machines) knows that peak theoretical numbers are bullshit, and that practical sustained numbers for your specific application are a much better measurement.  

If you want to approach the numbers with the Cell, you need an application that can provide a tight pipeline of Floating-Point calculations.  With game engines, you often want A*B=C and E*F=D, or something along that line, then C/D=H.  The issue here is that you'd need to load two SPEs with your first two calculations, then you'd have to wait many cycles until those SPEs are done *and* you can fetch the result from those SPEs. In the mean while, you may be sitting with 5/7 of your SPEs completely unused and you're unable to use them.

There are things you can do to minimize this, of course, but they require a lot of forethought and that particular performance issue cannot be mitigated 100%.  Nor does massive floating point performance gaurantee good performnace for most gaming engines.  Even if you could keep the SPEs crunching on pertinent information as opposed to busy work like generating 1000 of some same simple thing just to use the cell.

The second issue I have with the Cell is that they cut it down to a single core with SMT.  SMT's benefit depends on the hardware design as well as specific usage examples.  The 360 offers a version of SMT as well, and because it's not predictably useful in many circumstances, I will ignore it. This may be a flaw of my analysis.

Consider how game engines render frames.  The engine I wrote was fairly simple (although my implementation of Carmack's Reverse was good ;), but the basic ideas are the same.   Say your engine has to do the following each frame (please look past the gross simplification):

Deal with input

Calculate physics 

Sound 

Calculate positions of entities (including animation)

Calculate AI (for the next frame)

Render the frame 

Say your target is to spend 50% of your CPU time rendering the frame, and split the rest between calculating physics, dealing with input, calculating positions of entities and AI, etc.  Each "big loop" your engine takes, it will end rendering the frame and it will begin with the AI.  You also need to buffer at least one frame so you can render over the current frame.  The frame being displayed is the frame a user is responding to, but there is a frame being rendered into a buffer that the computer cannot change and will be rendered regardless of how the user responds to the current frame.  Therefore, the user's input does not affect the next frame, but rather the frame after that.  This is assuming the simplest case of buffering: double buffering.  If you're doing VSYNC, you have to wait to display a rendered frame until you sync, so there may be cases when you must wait to begin rendering the frame as well.  This is part of the reason it is necessary to render frames fast: faster framerates equal better game experience because the game responds to input more quickly and accurately.

The interesting aspect of having multiple processors and a multithreaded engine is that you can dedicate processors to specific tasks.  There can be all kinds of nasty race conditions to worry about, but assuming you plan for multiple processors from the beginning you can avoid many of them.  You can dedicate one processor to handling:

 Input

 Physics

 Positions/Animation

You can dedicate another processor to:

 Sound

 AI 

You can dedicate a final processor to:

 Rendering

The goal is to keep the third processor rendering 100% time, rather than the 50% time you'd be doing with a single processor.  This means your frames take 1/2 as much time to render.  You can run processor 1, 2 and 3 all at once, providing the data from processor 1 and 2 to processor 3 after processor 3 finishes the current frame.  Your gains come from being able to render twice as quickly.

The Cell originally had two cores and would have been a much better processor for game engines if that was the case.  Many of the tasks associated with game engines are searches and cannot be helped with the SPEs.

@ TheBigFatJ

This has been talked about here:

http://www.vgchartz.com/forum/thread.php?id=11075#end

If you want to revive the thread I would be happy to discuss the points you've just made.

@TheBigFatJ:

That's exactly what I think about the Cell even though I'm not a professional games developer (and I likely won't be since I'm in a different career path now).

If the Cell had two or three PPEs, it would be no harder to program than the 360's CPU, and the SPEs would still be there to help with whatever tasks are better suited for DSP type work. However, the cost of doing this would probably have been prohibitive at this time. Maybe in the next gen.

As it stands, the architecture of the Cell is probably a burden, since you have to force the SPEs to do things you'd rather use the PPE for. I imagine this would be especially true in a team with lots of developers, where each one would like to fight for a bit of PPE power, but will often have to settle for programming their components on the SPEs.

Excellent post RocketPig. 

Respect people when they are working. MikeB is just shilling to raise some $$ to feed his family - just let him do it to justify the salary.

Btw Mike, did you post here already the distorted dvd shot of the 360's playing Ice Age vs the PS3's correct ratio blue-ray screencap? Or maybe it's on the to-do list yet. You've posted the same biased comparison several times at neogaf, why are you discriminating against vgchartgz? they need some sony bias too.

Haha, Bitmap he already did that here. I called him out on it and he told me to grow up. Hilarious stuff.

so, i'm new to this forum but not to gaming and ps3/360.
Here is my analysis :

360 :

The good
- Good MS backup that may help to fund a price war and acheiving faster the magic $199 price point or buy some exclusives from 3rd party.
- Some additionnalx room compared to ps3 to reduce price
- Strong american marketshare. Huge sales of software there.
- Ecellent US sales in the recent weeks
- Already some AAA games justifyig the purchase of the machine : Halo3, Gears of Wars, Mass effect, Bioshock
- Ease of port of PC games : any publisher with a pc version has much easier work to do to get the game running on 360. Result : lineup is bigger.
- regular flow of good games and momentum.
- Good press.

The bad
- No more defacto exclusives vs ps3 due to late port or non availability of ps3
- RROD and other hw problems not yet solved completely
- The image : too hardcore. If 360 sells 14m units by years end, it's not much more than the original xbox.
- HDDVD may loose the HD battle and can tarnish the 360 image.

The ugly :
- disaster in Japan. It may prohibid any further 360 exclusives from japan studios just so that they can sell on their homeground
- Slowing sales in Europe.
- No real ammo against ps3lineup for 2008.

ps3 :

The good :
- ps3 Games are startin to show some muscle. Uncharted and Ratchet set the bar for top quality gfx.
- killer lineup for 2008 : MGS4+GTA4+LBP+KZ2+GT5 is nothing short of outstanding
- the HD era rise : any HDTV sold means more potential buyer for the ps3.
- European & japaneese sales picked up recently.

The bad
- Still some bad press up and there
- Still not THE game that evrybody wants to play and exclusive to ps3.
- ps3 exclusives are bombing on the sales charts vs the budget they represent.

The ugly :
- A year of miscommunication to erase in the mind of players. They may never recover from this feeling of "ps3 is a failure"
- Apparently no room to engage a price war
- Still poor ports or late ports
- Trails badly the 360 in the US with a huge difference in installed base. They need nothing but a miracle to reverse the charts and
- Stalemate on HD war, which results in negating on of this big + of the ps3, its blu-ray player.

My predictions of sales per region :
2008 will make or break the ps3. Sony knows this, and MS too. 360 has nothing to lose. They already won. Sony will be unable to reach the 70%+ marketshare they had with ps2.

2 scenariis :

- Either ps3 matches 360 on january & february 2008 US sales. => Then, ps3 sales will go growing and overtake 360 write easily. after those months, there will be MGS4, KZ2 and some more to maintain momentum in US while Japan and Europe will makeup for the lost initial ground.

- ps3 does not match 360 on january and february US sales : the gap then is increasing in US, the 360 becomes the de-facto HD nexgen console of choice and US is lost for Sony. Problem is that, ps3 needs the US market real badly as its a platform on which a dev wants to put ambitious project. Without significant installed base (majority), then no publisher will take the risk, thus resulting in the "death" of ps3, even if ps3 manages to sell more or less. Gamecubified as some would say.

 Most posts that long I just don't read because there full of fanboy ranting, but that was actually a very interesting post. 

PS3 can't match US Xbox360 sales in Jan/Feb

Which games/events of Jan/Feb would make the PS3 sell better ?

nothing ...

The 2:1 ratio should be kept ...