[INQ] IBM HPC Dropping Cell, Going to GPGPU

Quote:
This proves nothing when it comes to Valve's fanboyish statements. I would have loved to see Valve make that comment if the market share was held by Sony like they had with the PS2.

Quote:
Not to start a flame war but you calling Valve's comment fanboyish, is fanboyish in itself.

Valve's comments come from a company standpoint and a money-making standpoint. Cell takes extra man-hours to code for and doesn't have significant benefits in relation to the extra hours it takes. If you look closely, sometimes you'll see titles for Ps3 have a later release time than those for xbox. Even an extra month of development can translate to MILLIONS in extra labor. From a business standpoint, the decision to use Cell, as opposed to say an easier CPU and a stronger GPU was a bad one.

The problem was that Sony ANTICIPATED that they would have Ps2 market share and treated the Devs like they would. Any major business can tell you that you can mistreat your partners all day long, as long as you stay dominant. The second that you're not, they're leaving in droves. The same thing happens with how employees are treated when job conditions aren't good. They get mistreated and handed extra work. When market conditions improve, they bolt.

Quote:
OpenCL.

Quote:
Right, but AFAIK its current beta only works for CPU and there hasn't been any publicly released benchmarks with GPGPU apps that I know of. Which again makes me wonder why IBM would go with a technology still in development and one that hasn't yet produced any results for GPGPU computing .

Quote:
How does it come off as sounding fanboyish ? I think if it was such a monumental waste of time then just about all major developers would be taking the same stance as Valve shouldn't they ! Of course they aren't . Ask developers like IW if the PS3 is a waste of time . Valve is no longer a small Game Developer . If you want to talk in terms of investment and returns , do you think that a game like L4D2 would not be worth investing in on the PS3 platform . Their comments don't make sense business-wise . If they were to spend let's just throw out a figure ($3 million dollars) on porting L4d2 to the PS3 and were able to sell about half a million copies at $60 a piece.. .. Do the math !

Quote:
AMD and IBM have collaborated in the past.

Quote:
Yeah AMD and IBM have been working together since the original Opteron and Hyper Transport days.

Quote:
The PS3 hasn't magically evaporated. Valve are just lazy. Developing for the PS3 still makes sense seeing as people still play it.

My god.

EDIT:

Also, Valve didn't say "we're not developing for the PS3 because IBM are going to drop support for it in 6 months" Did they? So what the hell has this got to do with Valve?

Quote:
Valve isn't being lazy. They made a strategic business decision not to invest time, money, and employees to develop Cell games. They did their cost-benefit analysis and their calculated outcomes says... not worth it.

Quote:
Because Valve made that comment as a business, and they had valid reasons why, which i detailed. You, on the other hand, just decided to defend Sony for no good reason. Ignoring facts while mindlessly defending a brand due to loyalty is fanboy territory.

Quote:
There are quite a few developers who said the same thing in a less blunt manner. But even if Valve was the only one, how would that make the facts surrounding why Cell was a bad idea any less valid?

Quote:
You obviously don't comprehend the subject matter. The Ps3 as a whole is not a waste of time, the fact that Sony picked the Cell instead of an easier to code for CPU and a stronger GPU is the real waste of time. Couple this with the fact that Sony execs have come out saying that they purposely made the Ps3 hard to code for ( They're basically saying "We chose to run up programming costs on our partners and we don't care" ) and Yes, cell was a bad idea.

You seriously think it will only cost $3million to port L4D2?! Valve gave the port job for The Orange Box to EA, and it was 6 months later than the PC/Xbox version. If Valve could learn cell for only $3 million, I'm sure they would.More than likely it's going to take time for their guys to learn how to code AND require them to hire new programmers AND it's going to take more programming hours to make a game for Ps3 AND they're going to have to rebuild their gaming engine for Cell. So lets be more realistic:
It's probably going to take them anywhere between $10-$20 million to do all this for a game.
Now take $60 for a new game
-$10 for licensing
-$20 store markup
=Valve is only making $30 a game
x 500,000 copies
= and that's $15 million, which might be as much as them losing $5 million.

Take a look at all the Xbox-exclusive DLCs and you'll get a bigger picture on how the extra coding makes it not worth it.

Think about it, the cell choice has cost devs MILLIONS in extra programming. Millions to learn, millions in extra time and delays. And the Cell provides so little extra when compared to PC or Xbox CPUs. The only difference I've seen between the Xbox and the Ps3 has been that trash on the ground in Prototype has shadows.

Quote:
God, it's you.

Ok Kryten, so after at least a solid year of arguing this with you, you still choose to use the "lazy" excuse? After all the valid reasons that can be verified by facts on how the Ps3 is harder to code for which can run devs extra millions and provides no real tangible benefit......you still choose to use the same excuse?

Actually, Valve said it was a waste of time because it requires them to learn how to code for cell, and that coding is not going to be used on anything else, ever......so considering that, and considering how even the Ps4 may not have it now, Valve's comments regarding learning how to code for cell is more valid than ever.