Cell SDK released
If anybody wants to start programming on Cell, here's the SDK:
http://www-128.ibm.com/developerworks/power/cell/
Once programmers get some experience with Cell programming it will be possible to inject a little reality into the "Cell rules! Cell sucks!" threads.
(It's not future Apple hardware, but it is future hardware.)
http://www-128.ibm.com/developerworks/power/cell/
Once programmers get some experience with Cell programming it will be possible to inject a little reality into the "Cell rules! Cell sucks!" threads.
(It's not future Apple hardware, but it is future hardware.)
Comments
Originally posted by wmf
Once programmers get some experience with Cell programming it will be possible to inject a little reality into the "Cell rules! Cell sucks!" threads.
You presume that the programmers in question have a clue.
Heck, Carmack and Newell are stumped when it comes to multi-threading a game engine. Yet Sweeney has got the Unreal 3 engine working well with multi-CPU systems.
But Carmack and Newell are supposed to be expert programmers. Not every programmer knows every aspect of programming...some are totally clueless to anything but their branch of expertise.
Originally posted by Programmer
You presume that the programmers in question have a clue.
Well, not being a low level programmer (I'm mostly a Java-Head) I count myself as almost clueless on this front: however, Mr. P ... let's suppose you where going to be putting together a kick-ass media computer in, say, a years time, something that can handle oodles of DSP and SIMD, for apps such as Final Cut, Motion, Logic and the like ... which platform (the fact that we really don't know what the future holds notwithstanding)is more likely to grind through media data like Superman at a cheese grater?
Basically, I suspect Apple is going to keep PPC code in their OS to hedge their bets for some time to come (just as they kept Intel code previously), though as to how well the currrent PPC code translates to Cell is - for me - an open question. (along with, is it even worth it for Apple to persue hedging their media bets with Cell if Intel is going to be 'good enough')
Thoughts?
Originally posted by theapplegenius
i really cannot trust sony anymore...it's a shame.
... may the ghost of Celine throw herself down your stairs!
Originally posted by OverToasty
Well, not being a low level programmer (I'm mostly a Java-Head) I count myself as almost clueless on this front: however, Mr. P ... let's suppose you where going to be putting together a kick-ass media computer in, say, a years time, something that can handle oodles of DSP and SIMD, for apps such as Final Cut, Motion, Logic and the like ... which platform (the fact that we really don't know what the future holds notwithstanding)is more likely to grind through media data like Superman at a cheese grater?
Basically, I suspect Apple is going to keep PPC code in their OS to hedge their bets for some time to come (just as they kept Intel code previously), though as to how well the currrent PPC code translates to Cell is - for me - an open question. (along with, is it even worth it for Apple to persue hedging their media bets with Cell if Intel is going to be 'good enough')
Thoughts?
I expect Apple to keep their PPC code alive for a good long while, but that is mostly because of the installed base and the general healthiness of having cross platform code (its just a good thing to do). I don't expect to ever see them migrate back to PPC, unless something really stunning happens to Intel and AMD. In fact, if anything that big happens we've all got bigger things to worry about!
Which platform would I choose for media processing in the next 2-3 years? Well, if I didn't have a big pile of crappy legacy code to worry about I would choose Cell. More than 3 years out, however, and Intel will have had enough time to change its course (again) to compete so the picture will have become more complex. Personally I hate the x86 ISA so I'd still prefer Cell, but how nice the ISA is generally isn't the dominant factor in these issues. Furthermore, unless what I was doing would ensure the success of the processor family and fund future development, a major factor would be how well STI has driven Cell into their markets.