Linux : PS3's downfall?

[Fafalada]

see, Faf, as much as i agree with you about the psp being sony movies' wild card, i'm not so sure ps3 is so detached from sony movies on its turn (i.e. you saying it being a SCE wild card only) - blu-ray is too much of a hassle to be a clearly-SCE initiative

PS3 is Ken's baby - and not even Sony Pictures can one up him on that. It's obvious where BluRay came from, but there's enough common benefits for other usage that it just makes sense (aside perhaps for potential price issues).
PSP on the other hand has been under iron grip of suits with nothing but DRM in their eyes, from day one.
It's more like they are 'allowing' SCE to put games on it, then the other way around. The attitude towards security issues on the two platforms is like night and day, and trust me, end users don't even know most of it.
While you may have been disappointed with closedness of PSP, devs have often been infuriated by it.

so SCE, on their turn, have zero incentive to make theirs open in response.

That's a good point, but then again they had zero bussiness incentive for programs like PS2 Linux also.

that does not mean though the original psp couldn't have been much more sainly protected and homebrew-friendly if desired, no?

Of course, but as I said, things aren't the way they are because of some technical issues, it comes down to politics.

in the content of creating legally-clean homebrew for the device, it isn't, by far.

True - but what would be the benefit of releasing low level documentation to homebrew coders when APIs are enough to work with, and 99% wouldn't even bother going lower if they could?
I don't see ATI or NVidia usually releasing lowlevel details even to professional developers, they are more in the mode of guarding it like some stupid state secrets - hell for awhile we were in danger of ending up with RSX as just a black box too...

 

[Panajev2001a]

PS3 is Ken's baby - and not even Sony Pictures can one up him on that. It's obvious where BluRay came from, but there's enough common benefits for other usage that it just makes sense (aside perhaps for potential price issues).
PSP on the other hand has been under iron grip of suits with nothing but DRM in their eyes, from day one.
It's more like they are 'allowing' SCE to put games on it, then the other way around. The attitude towards security issues on the two platforms is like night and day, and trust me, end users don't even know most of it.
While you may have been disappointed with closedness of PSP, devs have often been infuriated by it.

Not true, VFPU's special moving body-trajectory-to-rigid-scenery-body collision instruction should at least have made one developer happy.

Just kidding man .

That's a good point, but then again they had zero bussiness incentive for programs like PS2 Linux also.

I disagree, the more their home console plans have grown in ambition the more such programs do make sense. PS2Linux shipped at a ridiculous fraction of the cost of the original PlayStation Net Yaroze which was in Europe a little too close to 1,000 Euros to my linking .

I think that PS2Linux did pay off in terms of helping people arrive in the industry with previous PS2 experience (R5900i programming and MMI instructions, VU's and VIF's programming, DMAC programming, GIF registers usage, GS usage, etc...) and that would have two effects:

1.) help the architecture's survival as basis of a more open platform back when the thoughts of employing the EE and the GS in a series of more and more powerful workstations and maybe personal computers... at a point they had that dream inside SCE.

2.) last but definitely not least, help filling industry with more PlayStation 2 experienced coders which could help SCE (if those joined the company as someone did... strangely enough it ended up being the creator of the famous library SPS2) and much more importantly third parties which might ease a bit more (the massive PlayStation 2 user-base did most of the job here) the focus shift towards PlayStation 2

PLAYSTATION 3 seems to be no simpler platform compared to PlayStation 2 and reaching its peak efficiency and performance is not easier so point 2.) would apply ever better in this case, but point 1.) is coming back strongly as they really seem to convinced to be able to push CELL and PLAYSTATION 3's basic architecture much farther than the EE+GS combo ever dreamed of being. The entire STI group sees a good reward in making sure that the CBE Architecture is wildly deployed and that as many programmers worldwide give it support. SCE certainly did not seem opposed to the Linux on CELL plan and seems to be supporting the Linux idea in an even stronger way than they did on PlayStation 2 (they want to arrive on day 1, at no extra cost to the user since it seems to be included on every console shipped... to say the whole truth even PS2Linux was free in a way they said that they were only asking the money for the HDD, for the USB Mouse+Keyboard, the Network Adapter and the DVD's as well as the shipping cost for all the material they would give you).

 

[Panajev2001a]

darkblu,

I definately agree with you that homebrew is VERY strong with PSP and sadly I did enjoy a lot playing with PSP once FW 1.50 was still viable for me.

Unfortunately I want to play games on it, Flash support, AVC codec support, RSS feed support, Locoroco support , LocationFreeTV support (and PLAYSTATION 3 compatibility), etc... make it kinda impossible for me to keep FW 1.50 in fact I just upgraded to FW 2.70.

I cared about Homebrew, but even if I struck a compromise and stayed with FW 2.00-to-FW 2.60 I would be limited in the choice of homebrew programs and especially in actual programming as for a usable debugger FW 1.50 is the only solution right now.

Maybe if I had the budget for a second PSP, but I prefer saving money for PLAYSTATION 3, Wii and games for my consoles than waste it on a simple standard PSP... if they released a new PSP with a screen that had the same resolution, size, color depth, but a lower refresh rate massively reducing ghosting then... I would probably consider trading mine in and add the difference for the new model or simply buying the new model.

It would be great if SCE found a way to allow homebrew coding on the PSP in a way that got the green flag from the Sony suits too. Unsupported homebrew programming is not nearly as fun as supported homebrew programming IMHO.

 

[Arwin]

they released a new PSP with a screen that had the same resolution, size, color depth, but a lower refresh rate massively reducing ghosting then... I would probably consider trading mine in and add the difference for the new model or simply buying the new model.

It would be great if SCE found a way to allow homebrew coding on the PSP in a way that got the green flag from the Sony suits too. Unsupported homebrew programming is not nearly as fun as supported homebrew programming IMHO.

I agree, am in the same pickle, but still not sure whether I will buy the second PSP or not. They will get cheaper as time goes on, and the FW1.5 rarer for a while (though there are rumors some people out there can flash any version down).

I just jumped back from 2.0 to 1.5, and a browser like Links2 makes that a lot of value. RSS readers are available too. But I'd love to be able to do stuff like LocoRoco and just buy Daxter. Tekken and MGS / Killzone coming may finish me off, but especially Tekken on the go will be ... :|

We'll see. My birthday is coming up next month, who knows I can get people to just give me money.

Anyway, the original PSP had a Sharp display. This is the one I have and after showing it to other PSP owners I'm starting to more and more appreciate I've been lucky (though I did pay hard). No ghosting. They should have stuck with that one, but I understand they'd go Samsung on us. They should improve that though, it's a shame. (Though the thing remains pretty awesome as it stands.)

But let's hope Linux PS3 works out, and the PSP can ride on that distribution platform.

 

[SPM]

I think access to the hardware from Linux will be through the PS3 hypervisor OS. This means the RSX bare metal and other bare hardware will be closed, but Linux itself will be fully open. In other words it will be a software hackers machine, not a games or hardware hacker's machine.

 

[Fafalada]

I disagree, the more their home console plans have grown in ambition the more such programs do make sense.

It's a nice way of growing fanbase and spreading platform awareness outside games, but in that sense it's basically marketting money - question is whether the suits believe the impact justifies the expenses.
There are no direct profits from distributing and supporting Linux on PS3 - unless they start selling 'desktop' software in the future, and that's a whole other can of worms.
On the other hand, it may very well be that the hefty price tag has something to do with this direction.

Anyway, the original PSP had a Sharp display.

To my knowledge screen refresh performance was the same on all retail units (and I've seen a large number of Japanese launch PSPs).
I suspect the Sharp->Samsung 'downgrade' rumour originated entirely from mixing up the manufacturer's 'reputation', with reports of early prototypes with better screen performance.

Anyway, if Samsung has a cheaper deal then Sharp on LCDs, I would reckon it has more to do with their manufacturing volumes being a lot higher, then product quality.

 

[patsu]

Hmmm.. but I know of a few OEM device companies that use embedded Linux and Open Source software to run their products (mostly embedded devices). If the software base and support is vibrant, and Cells with 6 SPUs (and fewer) are abundant, then would it be good business case for Sony ?

Naturally this will depend on the price of Cells, but the added software support has got to count for something given Cell's notorious reputation for being "hard" to program for.

 

[one]

Not even a desktop software suite, a commandline h.264 encoder optimized to Cell will sell 10 PS3 to 1 customer.

 

[patsu]

Well... short answer is there are many possibilities. Most of them need a lean httpd (for configuration), some need a database (so they stripped and ported MySQL), some needed DSP like functionality, and other proprietary stuff (e.g., finger print analysis). So more software availability is helpful provided the pricing for Cell fits into their BOM cost.

 

[SPM]

Hmmm.. but I know of a few OEM device companies that use embedded Linux and Open Source software to run their products (mostly embedded devices). If the software base and support is vibrant, and Cells with 6 SPUs (and fewer) are abundant, then would it be good business case for Sony ?

Naturally this will depend on the price of Cells, but the added software support has got to count for something given Cell's notorious reputation for being "hard" to program for.

The problem with using Cell for most desktop applications (including Linux and open source applications) is that the existing applications are coded for conventional CPUs and can't exploit Cell. This is not likely to change given the fact that most desktops will remain conventional and programmers won't bother to completely recode their applications including algorithms for a small number of Cell desktops given that the recoded applications will only run on Cell processors.

The only way I can see to exploit Cell's SPEs on the desktop on a widespread basis besides games and specialist image rendering applications, is to use Cell to emulate accelerated hardware and to access the SPEs via pseudo device drivers and this is the way IBM has chosen to make the SPEs available to Linux anyway.

I can see the possibility of very cheap but fast media PCs and set top boxes based on a single Cell chip to do mpeg decoding/encoding, sound, image manipulation, DRM, encryption/decryption, and maybe even acting as a low cost embedded 3D accelerated GPU at the same time by doing graphics rendering on it's own. Of course you can achieve the same thing or better using dedicated hardware for sound, mpeg decoders, and GPUs, but the Cell solution will be a lot cheaper and more flexible - the scale of production for the PS3 will make Cell very cheap. The PPE may be slower than the current x86 CPUs on non-floating point operations, but that won't matter in most cases because the graphics and multi-media aspects will be the bottleneck for most desktop applications.

As for servers, I can see Cell really taking off in floating point intensive parallel processing supercomputing servers once the double precision version of Cell becomes available. For conventional servers which just serve data, Cell is inferior to conventional processors because it is i/o and not CPU or floating point performance that is the limiting factor there, and Cell is particularly bad at buffering large amounts of data into main memory, or multi-tasking between a large number of threads because the SPE's use a limited local store, and have very expensive context switches.

 

[SPM]

If Sony could put all this in to PS3 plus OpenOffice (or AbiWord) plus photo management software plus maybe GIMP, Inkscape and a WebDav server/client to allow file access to/from a PC, and a bigger hard drive and maybe 512MB of XDR RAM it would make a really great Media PC.

http://www.crn.com/sections/breakin...QPP3YMQSNDLPSKH0CJUNN2JVN?articleId=189500212