forums.ps2dev.org

I've just posted a message to the main page.

http://ps2dev.org/News/Is_Sony_blocking_3D_access%3F

Have a read and let me know what you think?

David. aka Oobles.

Right now, Sony is split into two factions: the "Sony-BGM" faction, who wants to release content on highly controlled and proprietary media, root-kit your computer, and sue you at the least sign of trouble; and the "Sony-SCE" faction, who wants to release content across the net to run on memsticks, and put linux on their machines. Right now, the BGM faction has more power in the upper levels of Sony, making the SCE faction tiptoe around the shadows doing what they can from behind the scenes. Hopefully, the BGM faction will lose power as their method of doing business loses steam. There are signs that is happening.

There are a couple of possibilities at play here:

1. The 3D access could be on a list of pre-determined things that contractually must be blocked. What if, for example, there is an arrangement between Sony and Nvidia such that RSX access license is rolled up into Sony's developer license fees? Not saying this is the case, but an example of the kind of thing that will always force an annoying response.

2. 3D RSX access is on a list of things pre-determined no one should be allowed access to. Absent any counter-directive, the OtherOS/firmware developers must move to plug any holes found on this pre-determined list. Again, the not saying this is it, but another example of something that forces an automatic response.

3. The thing most people think is at play: Sony is afraid of losing money. But like your rationale, it stretches the imagination Sony would lose out this way. Thats why I am less inclined to think this is it.

4. Knee-jerk "they found it we plug it whatever it is" reaction. This is more likely the case if someone is in a position to look bad just because something like this was discovered, or if someone is directed to "watch" these forums and report things found, things that could be closed, as a matter of course, without any thought to whether it should be cosed.

I do agree, a "ombudsman" within Sony that will communicate with unlicensed developers as to what and why things happen the way they do could work wonders. The PS2Linux forums worked similar to this since there were actual Sony employees paid to operate, among other things, in such a fashion. But then, they were there for the people that forked over a smaller amount of money and still signed legal contracts/licenses regarding their usage of the materials they received.

The problem is, someone at a high enough level within Sony is not likely to be watching these boards and would depend on someone lower-level to advocate this to their attention enough (somewhat sticking out their neck) to be considered.

And if they did, my vote would be on Oobles as an upstanding representative on behalf of the legal dev community to communicate with Sony. :)

It seems that if Sony wanted to deter people from using Linux to boot their game ISO's they should simply make their games use every core of the cell processor, even for padding operations, just so the games will skip and buck if anything else is running on the system (Linux, RSX access workaround, OpenGL, etc.) I can see the possibility of somebody making an iso loader to run under Linux, but it's hard to imagine an iso loader which shuts down Linux and itself, making the game the only thing running on the machine.

boxbuilder wrote:It seems that if Sony wanted to deter people from using Linux to boot their game ISO's they should simply make their games use every core of the cell processor...

Actually we can be pretty sure Sony isn't worried about anything like this, at least in the context of dev community's RSX hacks. Its not even on the table as a possibility for any of us to worry about at the moment, and thus isn't figuring into the discussion in this thread.

I believe into the strong internal war between salesmen and engineers at Sony. PS3's father and his first lieutenant got sent into golden closets.

But it's a tragedy, really.

Engineers wanted to avoid hardware hacks by allowing homebrew.

Salesmen are all money oriented but are lacking technical intelligence (and relying too much on possible actions or threats using lawyers).

So the worst will now happen and all the efforts of engineers will fail.

Tradegy!

(from article) By blocking access to 3D aspects of the GPU, Sony will encourage developers to find exploits that bypass the hypervisor security.

Again, I like to point out the primary conclusions in http://www.xbox-linux.org/wiki/17_Mista ... ity_System - that by forcing crackers and homebrew teams to work together, security on the xbox was broken more quickly than it would have been if they'd allowed homebrew developers more access.

Given that Sony was in the position where homebrew developers were mostly concerned with developing fun stuff and not at all interested in helping the crackers, it seems foolish for Sony to have rocked the boat.

Great post Oobles.

How it will end? The PS3 will end up completely hacked from every possible side. Sony just turned on a big "HACK ME!!!" sign over the PS3.

Hackers don't care until there's a challenge, now there's one...

The first XBOX was a challenge, it was supposed to be ultra-secure from the starts and hackers jumped on the oportunity right away, it can now be modded with the click of a button without even opening the unit!
Same is happening with the 360, slowly but surely.

Another of their own bullets in their foot... how many can they take?

I think if people has access to BR and GPU they would be able to virtualize the whole PS3 and play copied games on it. Everything over Linux. Something as running Windows over VMWare on i386 Linux.

Virtualizing the PS3 and playing copied games is bogus. The firmware and all games are encrypted just like on the PSP, and having access to the BD and GPU doesn't at all lead to piracy.

The only reasons I could see for them to block it would be contractual obligations with NVIDIA (unlikely because the hardware isn't too much different from what people have in their PCs) and to prevent developers from making OtherOS games without paying Sony a royalty (unlikely because it's easy to pirate, so nobody's going to put a lot of money into developing a game like that).

As for the "they found it we plug it whatever it is"... I doubt that. It wasn't changed right away, even though the lv1_gpu_memory_allocate(0) thing was an easy fix (just assume they meant 252*1024*1024 when they passed 0, which would still allow old kernels to work fine). Plus, I have experience where I've notified Sony of a security hole in GameOS almost year ago now, and it's still not fixed. It sounds like they were making other changes in this area (increasing min FIFO size) so it probably just got fixed as part of that rework.

As for the direct RAMIN access, IronPeter has said for a while that this is basically an unintended error on their part, and the proper way to provide the functionality would be through a lv1 call. Such a lv1 call almost certainly exists, after all the lv1 interface should be basically the same for each LPAR. It's just not documented.

I agree with jimparis here. Sony has demoed a Cell blade that's a Cell + RSX with 1 GB of ram, and they've made some minor noises about selling it. But for that to be even slightly interesting, there will have to be software that works with both the Cell and RSX, and the best way to get that is to allow access to the RSX in a consistent way on both the PS3 and the blade.

Someone also noted the reason the new kernels were having trouble booting was due to the command buffer being increased to 2 MB. Does a dumb framebuffer need a 2 MB command buffer?

The HV apparently requires a 2MB command buffer now. Dumb framebuffer only needs a little bit (<64K), so what's done in ps3-linux.git is to overlap the framebuffer with the last 1984K of the command buffer so that the memory isn't wasted.

As I have found out recently, Microsoft (a MAJOR competitor to Sony in this field ;) ) has released XNA. It is a framework that allows people like us to create homebrew on the XBox 360 and Windows XP (possible Vista) platforms for, I think, $99 per-year subscription or something like that. It is held to the public, and I have seen books on how to work with XNA in bookstores, so I'm guessing it has to be. I have also seen information on the web claiming that Microsoft is going to be offering licencing. That could possibly lead to financial gain for homebrew developer in some degree. Now, this framework only works the C# language. C++ is out the window, so you won't really have the great control of optimization like C++ has to offer. Also, the XNA is more like a framework much like DirectX, so there goes even more optimization. One could possibly say it's more like a text-based editor than game development.

Now, if Sony has given us the 'otheros' feature, why not use it for good? Why not provide documentation, and openly allow others to develope on it, Sony? Why not compete with what Microsoft has to offer? If they could offer mini sdk's to homebrew developers to develope little games that could be hosted at the PlayStation Store, that would be good enough as well. I could see something good coming out of it for both Sony and the homebrew developers. That could lead to a nice resum'e, possible job offers, or allow Sony to scout out new aspiring game developers and designers. You never know, there could be an awesom, undiscovered franchise right in front of us... It's just hiding in plane sight.

Vincent_M wrote:As I have found out recently, Microsoft (a MAJOR competitor to Sony in this field ;) ) has released XNA. It is a framework that allows people like us to create homebrew on the XBox 360 and Windows XP (possible Vista) platforms for, I think, $99 per-year subscription or something like that. It is held to the public, and I have seen books on how to work with XNA in bookstores, so I'm guessing it has to be. I have also seen information on the web claiming that Microsoft is going to be offering licencing. That could possibly lead to financial gain for homebrew developer in some degree. Now, this framework only works the C# language. C++ is out the window, so you won't really have the great control of optimization like C++ has to offer. Also, the XNA is more like a framework much like DirectX, so there goes even more optimization. One could possibly say it's more like a text-based editor than game development.

It's not so limited. Granted, you have to use C#/.Net and you are required to use the hardware through the XNA framework but .Net is almost as fast as native. .Net was selected because it is VM based (so Microsoft can change/fix security settings very easily.)

Building programs for Windows is free. The XNA SDK is free, Visual Studio XNA 2005 Express is free, and all docs are on their website. (And lots of ressources elsewhere)

The $99 fee you refer to is have the right to upload programs to your Xbox. (a "megatons of premium developper content" according to MS)

Yes, it's harsh to pay to use an hardware you already paid for, but MS is the only one to have given a so strong impulse on homebrew on game consoles. (I haven't seen anyone else going to schools to start Homebrew contests between students)

If they could offer mini sdk's to homebrew developers to develope little games that could be hosted at the PlayStation Store, that would be good enough as well. I could see something good coming out of it for both Sony and the homebrew developers. That could lead to a nice resum'e, possible job offers, or allow Sony to scout out new aspiring game developers and designers. You never know, there could be an awesom, undiscovered franchise right in front of us... It's just hiding in plane sight.

... and they already use their contest to "hunt" for gems to be published. (and get devs a share of the download fee)

theBrave wrote:The $99 fee you refer to is have the right to upload programs to your Xbox. (a "megatons of premium developper content" according to MS)

I have been a Creator Club subscriber for a year now. The "megatons of premium developer content" amounts to some sample codes (w/ MSPL license), a license of GarageGames' TorqueX Builder license, and a 4 month membership extension because there were no content. Everything else can be freely downloaded by non CC members. With a bit digging, you can probably find the MSPL codes on internet too. The biggest selling point for $99 CC is ability to run your code on xbox. Your friends won't be able to see/play your games on their xbox until they get CC membership. When your CC expires, you won't be able to run your games on xbox anymore. The Dream/Build/Play contests aren't even limited to Xbox only, so if you want to compete, you can still do it on PC. With the additional resitrictions on things you can do on the xbox (no text chat, lack of libraries), not sure why anyone wants to pay $99 for that.

theBrave wrote:(I haven't seen anyone else going to schools to start Homebrew contests between students)

I think Sony had some shader contests, and PS2 Basic contests back in the days. They definitely didn't throw as much marketing money into their PlayStation computing effort as MS did.

The biggest problem is MS wants to control all contents on Live, so they won't allow just any user content to be accessible by anyone. They want to implement some censor/approval process. Sony took a different philosophy on PSN (see how you have to agree to a different EULA for each different demo/download?). Sony seems to want to push for more user-generated-content (UT3/LBP). This may make youtube-like-game-content easier to realize on the PS3 side. Even without Sony official support, it's possible to have an user created portal via OtherOS route. On XNA/XBox side, it won't be very easy without MS opening some doors.

Back on Topic.

I had an interesting though unproductive conversation with someone who claims to have access to LV1 source, and the conversation turned to this issue:

...
<Strontium> Understanding what every parameter of every HV function does without documentation from Sony, that is the hard part.
<pinskia> yes
<pinskia> and there is a reason why it is not documented
<Strontium> Sony are slack?
...
<pinskia> no
<pinskia> there is documentation, just internal
* pinskia has access to LV1 (and most of LV2) sources
...
<pinskia> because LV1 can change and will
...
<Strontium> Like taking away RSX access? Pretty spiteful really.
<pinskia> no
<Strontium> no? That wasn't a LV1 change?
<pinskia> what changed was the interface
<pinskia> since that was never an official supported HV call
<pinskia> only officially supported HV calls will never change
<Strontium> None of the HV calls are "officialy supported".
<pinskia> not true
<Strontium> Thats why they aren't documented.
<pinskia> all the ones which Linux are officially supported
<pinskia> and will not change
<Strontium> The RSX calls used to get RSX access were in linux, they just used different parameters.
<Strontium> So we couldn't rely on lv1_gpu_memory_allocate, so what lv1 functions can we rely on?

Conversation ended at this point due to it being very late for one of the parties, If I get any more information I will post it.

BTW, as is the way with IRC, we were carrying on about 3 conversations at once, this is the stuff limited to HV interfaces and RSX we discussed. I have not edited this to change the meaning of what is said, and if anyone wants I can send them my full log of this conversation by PM.

This means there is still rsx access but its all changed?

Rich43 wrote:This means there is still rsx access but its all changed?

That's the implication - and it makes some sense. Sony would want developers staying away from the hardware as that allows them to change the hardware over time. Added functions, reduced cost measures, changes due to changes in manufacturers... all can be handled by making people use official functions rather than banging on the hardware. Ideally, programmers only use CPU resources directly, and the OS handles everything else. Given Sony wants the PS3 to last ten years rather than five, there will probably be changes over the lifespan of the PS3 that would compromise compatibility if programs were allowed to bang on the hardware directly.

To say the least this is sony's stance, this has made the BIGGEST "hack me" sign. When it is, it will be BIG and bad!. Just like the PSP. Just a thought to all the people here....if there was a way to run code without restrictions of otheros, would you guys use it? if it wasn't legal?

just a thought! happy deving

Why they do not provide tools for the development of everyone?

The RSX may be "underpowered".The claim that it has 1.8 Tflops may be fiction!There is no official S0ny doc that says it has 1.8 Tfs!The 1.8Tflops count,counts every flop S0ny could find on the chip!

As we all know (assume), RSX has 24 shader "pipes", two alus each, each alu is capable of 4 flop madds. (vector4, vector3+scaler, vector2+vector2, etc..)
And since madd=multiply + add. (a 2 flop op) So, each alu is capable of 8 flops per clock. So, 24 pixel shader "pipes", would yield: 24x2=48x(4x2)= 384 flops per clock. (192 gflops per second, at 500mhz)
We don't add in the flops contributed by the free fp16 normalize operation, nor do we add the fog alu, nor the special function alus. (Some of which may have been there for fixed function pipelines to begin with) Or any other automated function that just happens to involve flops (there are lots).
Nor, (suspiciously) do we add the flops from vertex shaders to that number. (would be 80 flops per clock from those, if there were 8)
Which are listed on a separate slide. (There may only be 4 vertex shader alus btw, but no (official) linkable proof...yet)
(464 flops = 232 gflops total programmable from both pixel and vertex shaders, assuming 500mhz, and 8 vertex shaders)

Oh, and even if you give it credit for the fp16 flops, and the special function alus, and add the flops the supposed 8 vertex shaders, you're still only looking at 728 programmable flops per clock. Or, 364 gflops, at 500 mhz. And that's being generous

They are possibly afraid that the fact that the RSX is underpowered may become public! So, they could possibly be hiding the fact by removing RSX access frm homebrewers!

That's bogus. The upcoming ZEGO BCU-100 has the RSX and a driver, so it will be available for public use.