A different take on Sony’s removal of the Other OS feature of fat PS3s. “The reality probably is that Sony loses money on every PS3 it sells, counting on game sales to make up for the loss in revenue. Academic institutions using PS3s for clusters aren’t likely to buy games or engage in online commerce. That’s why, when you hear that the US military was planning to buy 2200 PS3 consoles to upgrade an existing PS3-based supercomputing cluster, Sony doesn’t jump for joy. I suspect it was news like this, plus other sales for clustering, that prompted Sony to turn off the ‘Other OS’ feature for existing PS3s.”
I still think that hadn’t geohot broke their hypervisor they would have left the option.
after all, sony is a media conglomerate and as such it will do anything to protect its DRM
I don’t own a PS3, so I’m out of the loop, but …
Didn’t geohot do his thing back in January? So why delay until now? Why was Sony, even as recent as a month ago, still claiming to support OtherOS (except on new models for “financial reasons”)? And they say it’s now due to “security”?
No, I think Sony handled this the wrong way. Even assuming losing money was the main reason, how does removing the feature from pre-existing machines help? No, in that case, they should leave existing machines alone, and either raise the price or sell separately the OtherOS function for newer models. But they aren’t doing that.
I don’t understand it, it’s very odd and confusing. We’ll just have to wait and see what else Sony has to say.
Two weeks ago this (http://geohotps3.blogspot.com/2010/03/custom-themes.html) happened. They still have bad memories from the psp’s mods
They’re going to have even worse memories when a pissed off enthusiast community breaks their system wide open, much like the PSP.
Not likely as the PS3 has far more protection. All games are encrypted and new keys can be tied to firmware updates.
I don’t think it has anything to do with lost software revenue, as the summary implies. After all, yes they may have lost some money on each of those units sold into a cluster situation, and won’t make it back on software sales for those units, but its also the type of publicity that would normally require you to pay for it. Remember how many “Entity X buys Y Playstation 3 consoles for supercomputer.” stories there were when the PS3 came out?
Furthermore, if “subsidizing” these supercomputers was really a problem, they could have come up with a liscense that forbade (at least theoretically, in some countries) purchase of retail units for that purpose, and then sold specially-liscenced versions at a proper rate — or they could have made back that money some other way, such as by offering support, as most commercial Linux distributions do.
Honestly, I think the ‘Other OS’ option played a huge role in keeping the system secure for so long. By making the other OS available, they gave the folks that just wanted to play with the fancy hardware or run computations on it what they wanted, thereby diverting the effort that would have otherwise been poured into cracking it from the beginning.
Once it was cracked, they reacted entirely the wrong way — taking the feature away only ensures that the system will see even more attention from crackers wishing to retain the feature. As for Geohot himself, it may have been more prudent, from the standpoint of ensuring the continued support of the Other OS feature, so simply have said “I have an exploit. Here’s proof. I won’t release the details as long as Other OS remains supported.”
That said, it really would have been nice for Sony to have exposed more of the functionality of the PS3, such as the GPU. Agreements with nVidia surely forbade such a thing by exposing the chip directly, but it should have been possible for them to run a virtual GPU of sorts in the hypervisor layer, acting as a driver behind the scenes, and exposed to the linux kernel as a sort of idealized GPU — indeed that’s more or less what they do, except they only expose a framebuffer interface. Had they exposed even some of the GPU grunt — say half of what’s actually available, then they would have made the homebrew crowd happy, given small/indie studios the chance to prototype on *real* hardware, and taken away one more bit of incentive from the crackers.
I think they’d find much more positive feedback if they did more to actually embrace open-ness, rather than trying to bend it to their own designs.
Honestly, I think the ‘Other OS’ option played a huge role in keeping the system secure for so long.
Really? There was numerous attempts to break PS3 since it’s introduction. Geohot did this through Linux. So he choose to be “piracy starter” instead of just letting these enthusiasts to play with fancy platform.
Because of his move I need to spend $450 on another PS3.
The Other OS option was how he claims to have broken into it. Allowing a foreign OS to access the system is a security compromise, plain and simple. You don’t provide a command prompt unless the system requires it.
The whole supercomputer aspect was a marketing ploy that they never should have pushed in the first place. They make their money from games, not cluster sales.
I thought it was only universities who were using PS3 clusters; the US military must be terribly shortsighted to use PS3s as a cluster. Everyone knows that the PS3 as a product will not be sold forever, just like all other consoles… eventually Sony would stop selling them (in favour of the PS4) and organisations using them as a cluster would have difficulty sourcing replacement units and spare parts.
By that time the CELL is old hat anyway. so big deal
They could also by then have software optimized for PS3 architecture for which conversion costs to x64 would trump any hardware savings.
I don’t think procurement of spares and spare parts will be a problem. There are way too many units out there for it to be a problem for some time to come.
And, you have no idea how many of those 2200 uniys are slated as spares/spare parts donors.
NASA is still scouring eBay for old 1980’s hardware to keep the shuttles running, as much of the test equipment is based on that stuff.
And that stuff is much rarer than a PS3 is going to be for decades at least.
Well, as CELL is stopped being developed by IBM, Sony should have pushed the CELL processor at least to the market that they didn’t create, e.g. the scientific and high-performance computing.
I was going to buy a PayStation 3 (no PlayStation, name changed on purpose) just to have the CELL architecture around to tinker with its computational capabilities. An IBM CELL solution is much too expensive for me.
Now i’m going to reconsider this.
in itself and I don’t feel sorry for any IT manager that didn’t fully consider the ramifications. You’re buying locked-up hardware from a company that isn’t in the server business.
I also don’t feel sorry for anyone that gets burned by Sony. Sony is ran by pricks, more news at 11.
What about using the official PS3 SDK for this? At least for a big organization it should be affordable.
I think the point is that the PS3 allowed a small organization to get up to speed with a nice supercomputing cluster. Big organizations may well go beyond this now. Whatever happened to ibm’s second gen cell with enhanced double precision (wasn’t that “xcell” or something)?
It doesn’t matter any more – modern GPUs (STREAM and CUDA) already trump the PS3 CellBE for super-computing purposes.
Frankly, I could care less – after reading all the comments I’ve read in the recent months, I’m *glad* Sony pulled the OtherOS support. All the whiners who want to blame it on Geohot deserve to lose their OtherOS support anyway.
I once thought it would be nice to own a PS3 for non-gaming purposes, but Sony had already convinced me otherwise in the last year with their castrated Slim model – and now this.
Distributed computing apps are now focusing more of their energy on the GPUs, so the PS3 will quickly lose its luster among the DC crowds as they move onto greener pastures.
Anyone just now setting up a cluster of PS3s for super computing has showed up to the party late…
On some workloads the CELL is not that good. For instance, string pattern matching, the 3.2GHz CELL achieves something like 3GBit/sec. Whereas the 1.4GHz Niagara SPARC achieves like 49GBit/sec. So how good is the CELL in string pattern matching? Not at all if you ask me.
They can add litigation expense to see how that balances out, California and other states will nail them to the wall as many of us who have those machines have a guide that specifies that the feature is a part of the product.
Class actions are being prepared.