Slashdot Mirror
IBM's Radical Cell Processor
Rouslan Solomakhin writes "Forbes has recently posted an article on IBM's new revolutionary Cell processor. Cell is going to enable PS3 developers to create movie-quality games with blazing-speed graphics. Applications in other areas are also considered." From the article: "Some techies say PlayStation 3, which may debut by midyear and could end up in 100 million homes in five years, will usher in the next microchip revolution. The Sony system owes its prowess to a microprocessor called Cell, which was cooked up by chip wizards at IBM (with help from Sony and Toshiba) at a cost of $400 million over five years."
With Apple no longer buying chips from them, they really need to prove themselves.
I lost my signature... help!
**Outside the Sony Booth being handed fisherman's waders**
[Gabe]: What are these waders for?
[Tycho]: My guess? All the bullshit
It's not that I don't think this chip might be as fantastic as everyone says but since Sony has basically lied out its ass for its past 3 consoles, I'm not giving it the benefit of the doubt with the PS3 and god save any journalist who gets sucked into their schilling.
So, does this mean the PS3 will have more games based on movies?
Forbes has recently posted an article on IBM's new revolutionary Cell
Damn, the enemy within. I can't believe they've infiltrated IBM. Is nowhere safe?
You need to login with your "FREE" Forbes.com account :)
I lost my signature... help!
>Cell is going to enable PS3 developers to create movie-quality game
hum...
more like:
Rumours and hype about playstation 3 intended to reduce sales of Xbox 360.
nothing to see here...
Open Source Drum Kit, LPLC deve board - mjhdesigns.com
Really? Just like the PS1/2 could do on the fly Toy Story quality graphics? Or did you just get around that by saying movie quality games, rather than games that look like movies, but still implied it?
I have no doubt the cell is going to be impressive, but we are quite along way away from an affordable processor than can replace a render farm (I believe thats what there refered as).
"I may be full of crap about this game, and I may be wrong, and that's fine." -Jack Thompson
More information about the Cell processor directly from the source : The Cell project at IBM Research
-- javaDragon is an instance of JavaDragon.
I'd say almost everyone is in agreement that the Cell processor is a very powerful design, but I don't believe the PS3 will be the best example of what it can really do.
Sifting through what I've read about the PS3, the Cell processor is bottlenecked by a few things including but not limited to memory bandwidth, and a fairly generic pc graphics solution from nvidia (by generic I mean, one of their standard pc products tweaked slightly for use on the PS3).
The "movie quality" games that I'm assuming the article is referring to are the demos shown at places such as e3, which are nothing more than either pre-rendered movies or carefully programmed, high end pc demos (Epic demo with high end pc and 7800 sli config).
I'm not trying to disparage the ps3, nvidia, or IBM. Frankly, I'm a fan of Nvidia and the Cell processor and I truly believe (drm jokes aside) the ps3 will be a solid console, but I think saying that the PS3 with Cell, "...is going to enable PS3 developers to create movie-quality games with blazing-speed graphics" is misleading, ignorant and sensationlist journalism.
$sys$droids
I think I've heard of this line couple year back, sometime before or around PS2's lunch date possiblly.
Movies take several years to generate two hours of content. Games are often ten times that long, with a much smaller budget. How can they possibly be of comparable visual quality? and why do people try?
I would much rather have games that concentrate on art instead of graphics. (Rez and Darwinia come to mind as examples of visually impressive games with non-realistic styles. Unfortunately, it doesn't seem to work well in terms of sales...)
Who wants to develop games that are the quality of holywood movies...?
I guess we will be seeing Cell in servers at some point as well, though not as cheap as in a PS3.
Sony probably won't want anyone to run Linux on the PS3, lest geeks start cranking out PS3 server farms, but hopefully Sony will leave enough backdoors so can we can see the PS3 run Linux (or FreeBSD, or some beta OpenSolaris distro). Knoppix running on the PS3 just about removes the need for a home computer.
Hopefully Sony will create such a backdoor. I mean if they can screw up with a rootkit...
*ducks*
READY.
PRINT ""+-0
Motion pictures made in the 1930s are also, technically, "movie quality", seeing as they're, well, movies....
What exactly does the reporter (and Sony) mean by that statement?
(Oh, yeah, I forgot: "well if they'll be the same quality as some of the movies Hollywood pumps out recently, I'm not buying it...")
If you want to read more about the CELL heres a link for you...
http://www.research.ibm.com/cell/home.html
http://www-128.ibm.com/developerworks/power/librar y/pa-fpfxbox/?ca=dgr-lnxw09XBoxDesign
All your base are belong to Google.
Since the first cell product is already shipping. http://www.linuxdevices.com/news/NS3591350722.html we should be able to benchmark the processor pretty soon and find out if it is all a hype or this really is the second coming :-)
The Cell won't be terribly well suited for AI either, so you probably don't have much to look forward to. Game AI is typically notoriously branch-heavy and often tends to be mostly integer code (seeing how it is mostly search problems and at worst a neural net or two, no heavy stuff like machine vision since all information is already available). Which the Cell is more or less worthless for.
It annoys me greatly that the Cell is getting the hype it does, not only it is very specialized and as such hard to use, it is not even very innovative. One of the very first proper vector computers, the ILLIAC IV, was based on pretty much exactly the same approach. The Cell would at any rate be absolutely horrible as a general computing chip.
The IEEE Spectrum magazine (surely a better source for Slashdot readers) predicts that Cell will be a winner in the multimedia space, noting that already its going into TVs made by Toshiba.
They also mention Linux on page 2.
An Eye for an Eye will make the whole world blind - Gandhi
Multicore CPUs, and multiprocessor systems, are only going to be as fast as the software can make them. Concurrency is a major focus of software programming research at the moment.
For some more info, check out:
The Free Lunch is Over, the article that sparked the discussion.
A talk Herb Sutter did on the Concur project, a research project into abstracting concurrency, sorry IE only but it's worth it
mistake?
a mistake is forgetting to tighten a bolt, or carry the two, resulting in problems down the line.
sony's rootkit was an intenional and corporate level decision. DRM itself has no justification for existence at all. When confronted with the fact that it does not stop piracy, executives often come clean by putting forth a "positively spun" statement which pans out to, and i paraphrase: "we want to deny the technologically unsavvy of flexibility theyre used to in order to screw them out of money we dont really deserve"
Sony's products arent that great either. most sony stuff i've owned has broken (not broken down.. literally broken like plates break), so i have no trouble with the idea of not buying their flimsy and anticonsumer products.
VLC FOR MAC IS DYING! IF YOU DEVELOP, PLEASE SAVE IT!!
Don't be fooled by the prints you see of early pictures. The original negatives used for movies in the 1930s were high-resolution monochrome film. The prints you see of them today are marred by age, repeated copying and sometimes a mismatched frame-rate.
The restored print of Fritz Lang's Metropolis is exquisite. The resolution is far beyond anything a playstation will generate, and that's after reconstruction. The original 1927 negative would have been even better.
Metropolis frame
This is a low-resolution capture, but you can see how detailed and high-contrast the frames were. The vignetting around the edges is the major picture issue.
Of course, the effects in those early movies weren't often brilliant, but on a console effects are easy. It's the subtle shading and curves that challenge a playstation.
Writing concurrent software isn't that much more difficult than writing single threaded software, as long as you do a good job of partitioning the system into seperate control loops early on. The main difference will be a period of tweaking and adjusting the interplay of the different threads of execution in the system towards the end of development. It's not uncommon for this last stage to take more time than writing the code initially. A tactic that will help a lot is to build an event log into the software from the beginning that can be used to record when each thread finishes doing some processing task. The later version of the freescale 7400 series processor have many features for just this purpose, I would think the ibm 7400 core used in the cell would have the same features, but I am not sure.
A good language to look at for how concurrency can be supported is Ada. There is a lot of good stuff in Ada and a lot of bad stuff in Ada, but the designers did a very good job on the concurrency model.
As I understood it, the Cell comes with a VM on top, which will coordinate the actual work done on all pipes. The VM is supposed to use intelligent allocation algo's to schedule compiled code on each Cell. the code can of course be compiled using specific platform compilers for the ps3, just like they did for the ps2. So in fact what is said about the ps3 is true. It's going to be hell to program for it *directly* (where most of the power can be found) but I wouldn`t say it`s going to be impossible for sony to build a line-up of titles at the start.
;)
If there's anybody who has more details about Cell programming, I`m interested!
With great power comes great electricity bills.
Could we all just remember what IBM had in mind when they designed Cell? If you have a read of the Introduction [Pdf warning], you can see they identified the primary bottlenecks to performance, back in 2000, one of the most important problems being memory latency. Now, if you've done some work with assembler, you should know that every time you touch main memory, you loose about 20-30 clock cycles through your memory's low speed. If you want an example, I have a 3GHz computer, but the memory goes at 400Mhz. Just think how much time it must spend waiting for that memory?
Cell counters this problem by using SIMD in combination with what they call "Local Storage". Instead of having to wait for every single memory transfer, threads can read blocks of memory into storage actually on the SPE, process it, and then read it back. All with a couple of instructions, and execution continues even while the memory is been read/written.
The closest that present-day multi-processor computers can get to that is by caching the data. However, that still means that a cache miss will halt execution for many cycles, and each processor / core has to constantly check what other processors / cores have in their caches, ocassionally invalidating them.
What this all adds up to, is a level of efficiency that hasn't been seen before. However, I don't think it's gonna be anyway near "movie quality" graphics, you'd need a farm of Cells for that.
My impression is that the technology surrounding the x86, e.g., the system bus, etc., has had a lot more development geared towards whatever PC's are typically used for (like desktop and server stuff), whereas the PowerPC line, and certainly the Cell, have not. This would have been due to x86 having such a large market share and many companies and people targetting there development towards it. This resulted in the x86 system architectures outpacing smaller-share competitors and being able to deliver much more performance, as far as a desktop, server, or scientific user is concerned. So Apple, which produces computers mainly targetted towards the desktop, decided to go with x86 CPUs to finally stop being hamstrung by "the rest of the computer". Apple can design a basically good motherboard, but they're the only ones doing it for the PowerPC on the desktop. x86 has 20 companies doing it, and with that amount of effort and competition, the results are better.
Apple's other problem with selling computers is that low third-party software availability makes the hardware less attractive. I assume that while it won't necessarily be any easier to port a Windows/x86 program to OSX/x86, it won't be any harder. It will be easier though for them to put some kind of emulation layer in there to make it easier for Windows progams or "half-ported" programs to run, or at least have a virtual PC that will run Windows to run Windows programs (or dual-boot). Reports from friends who run current Macs though are that OS/X is not the most stable operating system. I believe it probably has a better general software architecture and code, just that it hasn't gotten as much testing or rigorous use due to the smaller size of Apple and its user base. Windows (surprisingly) seems much more stable than OS/X. So the wisdom of running Windows inside of a virtual PC inside OS/X is debatable. With the increase in market share for Apple that I think will be coming, their reliability may improve. I'd certainly prefer to work with a Unix-based system and clean OS API over the pile of closed spaghetti code that is Windows. I think they're counting on people deciding that an Apple computer that can run either OS/X or Windows is better than a non-Apple computer that can only run Windows.
(Sorry, that went offtopic).
I hear it also will cure cancer! Go Sony!
My friend at a video game company has been saying everyone much prefers working on XBox 360 than the PS3, and the biggest complaint is no one really knows how to write the high performance code Sony boasts about. Also, he says Sony's developer support has gone down hill and Microsoft has been bending over backwards to help developers working on 360 games.
Anyone else in the game industry care to confirm/refute this?
The bitter lessons of a veteran coder: http://bitterprogrammer.blogspot.com
From a pedantic point of view, Sony has only released two consoles so far. How can you say they've lied out their ass about 3 consoles? Is it because you somehow know the new one doesn't measure up? Is it your special fanboy sixth sense that gives you the ability to know that?
Additionally, I don't remember PS1 being a disappointment at all. Toshinden was ready for PS1 at launch in JPN, and look great. It played well too, but it had 3-d fighters and 3-D backgrounds. By the time the US PS1 launch rolled around months later Battle Arena: Toshinden was ready with further improved graphics including use of transparency in the backgrounds (waterfall). Meanwhile on Saturn, they had Virtua Fighter which by the US release wasn't even texture mapped! Do you remember receiving your free "VF 1.5" CD in the mail from Sega? I do. And it still didn't look as good because the backgrounds were not 3-D and the platform itself couldn't even do transparency (it used stipples instead).
PS1 handily beat the competition on technical merit and games.
So I don't get your complaints there. Perhaps they are with PS2?
PS2 isn't as clear cut, but as a performance thing, I have to say it works for me, despite a truly bad architecture (very little VRAM) that could have sunk it. It is long in the tooth right now, but it is at the end of its life-cycle.
But is it a failure? Did Sony lie out their ass? Not that I noticed. Yea, they hyped it a bit. Who doesn't? There were references to Toy Story-graphics made, but Toy Story was the big thing at that time and MS made them also for Xbox. It was the first all-CG movie ever. Should I bitch at Bally-Midway because they made two TRON arcade games back when it was new, and neither was even 3D? That's a much more major failing at matching movies. Yes the two games were quite fun (one is a saught-after classic), so why complain?
I would note that except for Dolby Digital, PS2 has actually kept up with the times quite well. This was a platform that wasn't even advertised or planned to do 480p when it came out, and yet does 480p in a fair number of games now and even does 1080i in one (GT4)! It even bested Xbox in the Sim-racing graphics wars of 2005. GT4 definitely has better/fancier graphics than Forza (and has 1080i support while Forza maxes out at 480p), although Rallisport Challenge 2 is still the best looking racing game of its generation (maxes out at 480p, although it looks so good there's no way to complain). And the biggest/most fun racing game of 2005 turned out to be neither GT4 nor Forza but Burnout 3.
As to Sony claiming the CPU would be the new PC CPU, I don't remember that. I don't think Sony thought they would unseat x86. I do remember them saying the PS2 chipset (I'll call it the EE although it's really more than that) would be used in other things. For the most part this wasn't true, but they did ship a PVR using the EE in JPN (the PSX). They also attempted to license the chipset for use in TVs and set-top boxes, but no one took them up on it, probably wisely.
Sony also has plans to use the new (Cell) chipset in TVs/media devices again. Go back and find the Digital Reality Creation 2 announcement, it sounds a lot like it uses a Cell chipset.
Why they keep saying this stuff is basically because they partner with Toshiba to make a custom chipset for them (in this case they even built a new fab specifically for it). When you make that kind of investment, both companies tend to get thinking about how they could use the chip even more, thus making even more profit off a fixed asset (the fab). It is a Japanese tendency to wax poetic about the future of a significant new design/advance like this, and often it doesn't come true. I mean, you can't go buy an Asimo down at your Honda dealer, can you?
I like my 360. The games are almost universally awful, but the hardware is good. I have high hopes for it. But I also have high hopes for PS3. PS1 and PS2 have been very good consoles, and had plenty of titles worth buying the consoles for. So I expect PS3 will be good too. I have to say I find the $400-$500 price ridiculous, but then again, I did buy a 360 at that ridiculous price.
http://lkml.org/lkml/2005/8/20/95
Try this link to the printable version (should work without being logged in and is nicer anyway, all three pages in one): http://www.forbes.com/forbes/2006/0130/076_print.h tml
A World in a Grain of Sand / Heaven in a Wild Flower,
Infinity in the Palm of your Hand / And Eternity in an Hour.
Um... no. If Valve (not Blizzard) go out of business, you won't be able to play your games online anymore. You'll theoretically still be able to play LAN games and single-player. But I share your concern.
However, as far as DRM goes, Steam seems pretty inoffensive to me. You can make & restore backups of the data, you can install the game on more than one computer (but you can only play multiplayer on one computer at a time). I don't think Valve could have made the DRM any weaker without having Half-Life 2 cracked and illegally distributed on a massive scale within minutes of release.
I think the case for DRM on games is a lot stronger than for DRM on movies. In the case of games, most are bought by teenagers who would have the time and motivation to go and hunt down a pirated/cracked copy with pretty much no remorse. I know very few of my peer-group at school would have gone out and bought a game if there was any way on earth to avoid paying. In the case of movies, the majority of sales are to adults, the majority of whom I believe will go for what's most convenient -- and heavily DRM'd movies aren't convenient in the slightest.
The big reason I like Steam is that it makes it possible for small studios to distribute their games worldwide without having to worry about fickle publishers, and Darwinia's release on Steam is a good example of why that's a good thing; before the Valve deal it had very few sales outside the UK, and since then many thousands of copies have been sold worldwide. And over 50% of the retail price goes to Introversion, rather than the typical ~20% that the studio receives in a normal publishing deal.
Pirate Party UK
The PowerPC was jointly designed and developed by this alliance.
a rs/1
You can read here:
http://arstechnica.com/articles/paedia/cpu/ppc-1.
Thanks, now I don't have to barter my soul to read somewhat clueless technical rubbish from a business & financial magazine ;)
"Christ what a design! I could eat a handful of iron filings and PUKE a better emergency pump than that!"
The 68060 was announced as the end of the 68000 line before it ever entered production. Apple had the choice of abandoning the 68000s or be regarded as a dead-end company.
Contribute to civilization: ari.aynrand.org/donate
There's no doubt about that in my mind.
I dug up an old review I did of Forza (esp. vs GT4), and reposted it. Here's a link.
On the Clarkson thing. I have to disagree, Forza did not get those things right. It got more right (see my review), but in Forza, turn 1 still doesn't exist, just like in GT4. Clarkson seems to refer to this. Now, in real life, I've never seen a vehicle that can ignore turn 1 on Laguna Seca. Even an underpowered car like a Spec Miata has to set up for turn 1 a little bit. I do feel that such a vehicle could perhaps exist. It would have to have slow acceleration and lots of lateral grip. In GT4 and Forza, you just drive right through (using the pit-out lane in Forza, you don't even have to do that in GT4).
Additionally, Clarkson is again right on with the area between turn 7 and 8 (8/8a is the Corkscrew, Clarkson refers to it as turn 7). In both games, you blast right over 7 (it's as much a bump as a turn), then do the braking after it. This is impractical in real life. First of all, as Clarkson alludes to, you'd simply fill your shorts. Cresting 7 at full bore, you'd have a wall about 300 feet ahead and a downhill zone to do your braking in. Additionally, most cars in the world just couldn't slow down in that short a distance in a downhill braking zone. Really, neither game gets good marks in this area.
Forza also for some reason has a very sticky turn 9. High-powered cars will drift out quite a bit in 9 in real life, and do somewhat in GT4. But not so in Forza. It's odd. Of course, in GT4, turn 10 is a real joke (like 6 is), so Forza still wins in this part of the course.
Referring to your comments about worthless cars being in GT4, often games do seem to bulk up on cars. None has an SUV race like PGR2 does though! I actually liked the SUV racing in PGR2, it was a change (like the Strana trucks in TOCA 2). But both are only gimmicks, I hope they don't expand upon them in the future. Drive PGR3 and you'll wish these shitboxes were back. PGR3 removes nearly every car any person could reasonably own from the game. I think the only ones left are the Mustang GT, Corvette C6 and Corvette ZR-1. I also know people who own Ferrari 355s (although not the F1 model) and Aston Martin DB9s, but I don't count those. The slowest car in the game is the Ferrari Testarossa. Annoying.
Lack of damage in GT4 doesn't bother me. Damage is mostly a hassle. Forza does try to keep the stupid AI mistakes from hurting your outcome, but it isn't completely successful. There's nothing worse than having to restart a race because the AI punched a hole in your car. This happened in TOCA 2 (most damage modelling yet), and it drove me nust. And TOCA 2 had fewer AI problems than Forza.
I do prefer that games try to keep you from driving on the wall and across the grass to reduce times, especially as online play becomes more important. Unfortunately, Forza didn't penalize you enough for hitting walls strategically, so it can be done to advantage. PGR3 adds slow-down penalties (like GT4 does in the rally races, but not as long) but left out the penalties for crossing the grass. So people cut several corners in the game. Very sad.
I don't feel GT4 feels a lot like GT3. Even GT4 Prologue had significant improvements over GT3. I'm not justifying their lack of online play or the zillion years it took to come out either, but to me it's quite a different game. Do you remember how bad GT3 was? It was greatly inferior to GT2 in game progression, because the removal of cars (and thus the removal of entry qualifications) made the game difficult and pointless at times.
I liked Forza, but as I mentioned in my review, there is plenty of room for improvement. I didn't finish Forza, the races become very difficult due to AI mauling at the high levels, and that's just not fun. And I didn't like the endurance racing in Forza. Despite the technological advancements of the Drivatar, it just didn't work for me.
http://lkml.org/lkml/2005/8/20/95
Friends don't help friends install M$ junk.