The Whole PS3 Debacle
Veteran TPS staffer Robbie Otal collects some of the more recent happenings with Sony's multi-core beast.
Published: January 12, 2006
When Kotaku first reported this story, my first gut reaction was to take all of this with a grain of salt. But when Sony neglected to show off any morsel of PS3 hardware during this year's Consumer Electronics Show, it made me wonder whether or not these problems may actually be real. Fact is, I've always known from the start that the Cell processor would be a real challenge for the developers. Because it uses vector processing versus scalar processing - which is the industry standard, developers must now take more time to get used to this new form of computing architecture. In this case, both the PowerPC core as well as those seven Synergistic Processing Elements (SPE's) found within the PS3 must all be working in tandem with one another, with the core deciding which SPE will be performing the calculations for which task. The core must ultimately be responsible for properly distributing the tasks evenly to each of the seven SPE's.
More specifically, these tasks for the SPE's will include number crunching for video compression and decompression, encryption and decryption of copyrighted content and the rendering and modifying of graphics. These SPE's will all be performing billions of single-precision floating point calculations in order to produce the desired results. In this regard, many of the tasks performed by the Cell processor will be broken up into varying portions, and performed in parallel with one another - hence parallel processing.
It is this parallel processing that will likely be the biggest obstacle that developers will likely face when developing PS3 software. And this is also the likely obstacle that could be wreaking havoc for a number of PS3 titles, due to inexperience in dealing with so many processors all performing work at the same time. As Craig Steffen, a senior research scientist at the National Center for Supercomputing Applications told IEEE Spectrum earlier this month, working with the Cell processor will be like working with eight engines running at full speed without them constantly stopping to wait for data - a serious hurdle for any gaming programmer. But he later went on in the interview to say that parallel processing will likely become fairly routine, and that in the next few years, most programmers won't consider asymmetrical processing as anything different from the ordinary.
Software Engineers for IBM have already begun creating development tools that will likely make it easier for programmers to divide tasks between the power element (in this case, the PowerPC core) and the synergistic cores. The company has already released over 700 pages of documentation targeted to ease the burden on the developers and programmers as they make use of this new technology. The question, is whether or not all of this will end up producing the AAA titles that we (and most gamers) will be expecting to see on the PlayStation 3.
Right now, Sony has remained rather mum on the PS3 release date, still going by their original "Spring 2006" target. You might remember, however, that when Sony first announced the Japanese launch date for the PlayStation 2, they did so in 1999 during the Tokyo Game Show. They ended up releasing the console on March 4th, 2000. The console didn't arrive here in North America until October 26th, 2000; some seven months later. It can then be said that when the PlayStation 3 does finally arrive, it will likely be released in Japan first, and in North America second, with a lengthy time period in between.
I predict that the way things are going right now - specifically referring to the way Sony has not divulged any further details regarding their Spring 2006 release date, and the fact that the PS3 was a no-show at this year's CES, the system will likely be released either this Summer, or probably this Fall in Japan. With the current Spring 2006 release date in mind, North American's shouldn't expect to see the PS3 until this holiday season. However, if the PS3 is pushed back to the Fall season in Japan, then we probably won't be seeing it until 2007.
Would a 2007 release date in North America really hurt the chances for the systems' success? Well, it would certainly be a much longer wait than anticipated for gamers who are still waiting for the PS3 to arrive. It would also give Microsoft and Nintendo (who will likely release their Revolution console later this year) a chance to further erode Sony's market with their respective consoles. But with the XBox 360's most eagerly anticipated title; Halo 3, likely to arrive either late this year or sometime in 2007, I'd say that Sony still has a chance to release the PS3 next year and still get away with it. It would certainly be a long wait - maybe too much for some - but it would give Sony all the time they need to iron out every wrinkle and ensure a successful launch with the AAA titles that everyone expects to see.
Whether or not the scoopers over at Kotaku or right or wrong about the ongoing PS3 development woes, we can confirm for sure that developing anything using the Cell processor will not be a walk-in-the-park, and that many complexities will likely be waiting for most programmers.
More specifically, these tasks for the SPE's will include number crunching for video compression and decompression, encryption and decryption of copyrighted content and the rendering and modifying of graphics. These SPE's will all be performing billions of single-precision floating point calculations in order to produce the desired results. In this regard, many of the tasks performed by the Cell processor will be broken up into varying portions, and performed in parallel with one another - hence parallel processing.
It is this parallel processing that will likely be the biggest obstacle that developers will likely face when developing PS3 software. And this is also the likely obstacle that could be wreaking havoc for a number of PS3 titles, due to inexperience in dealing with so many processors all performing work at the same time. As Craig Steffen, a senior research scientist at the National Center for Supercomputing Applications told IEEE Spectrum earlier this month, working with the Cell processor will be like working with eight engines running at full speed without them constantly stopping to wait for data - a serious hurdle for any gaming programmer. But he later went on in the interview to say that parallel processing will likely become fairly routine, and that in the next few years, most programmers won't consider asymmetrical processing as anything different from the ordinary.
Software Engineers for IBM have already begun creating development tools that will likely make it easier for programmers to divide tasks between the power element (in this case, the PowerPC core) and the synergistic cores. The company has already released over 700 pages of documentation targeted to ease the burden on the developers and programmers as they make use of this new technology. The question, is whether or not all of this will end up producing the AAA titles that we (and most gamers) will be expecting to see on the PlayStation 3.
Right now, Sony has remained rather mum on the PS3 release date, still going by their original "Spring 2006" target. You might remember, however, that when Sony first announced the Japanese launch date for the PlayStation 2, they did so in 1999 during the Tokyo Game Show. They ended up releasing the console on March 4th, 2000. The console didn't arrive here in North America until October 26th, 2000; some seven months later. It can then be said that when the PlayStation 3 does finally arrive, it will likely be released in Japan first, and in North America second, with a lengthy time period in between.
I predict that the way things are going right now - specifically referring to the way Sony has not divulged any further details regarding their Spring 2006 release date, and the fact that the PS3 was a no-show at this year's CES, the system will likely be released either this Summer, or probably this Fall in Japan. With the current Spring 2006 release date in mind, North American's shouldn't expect to see the PS3 until this holiday season. However, if the PS3 is pushed back to the Fall season in Japan, then we probably won't be seeing it until 2007.
Would a 2007 release date in North America really hurt the chances for the systems' success? Well, it would certainly be a much longer wait than anticipated for gamers who are still waiting for the PS3 to arrive. It would also give Microsoft and Nintendo (who will likely release their Revolution console later this year) a chance to further erode Sony's market with their respective consoles. But with the XBox 360's most eagerly anticipated title; Halo 3, likely to arrive either late this year or sometime in 2007, I'd say that Sony still has a chance to release the PS3 next year and still get away with it. It would certainly be a long wait - maybe too much for some - but it would give Sony all the time they need to iron out every wrinkle and ensure a successful launch with the AAA titles that everyone expects to see.
Whether or not the scoopers over at Kotaku or right or wrong about the ongoing PS3 development woes, we can confirm for sure that developing anything using the Cell processor will not be a walk-in-the-park, and that many complexities will likely be waiting for most programmers.
