Would A Kickstarter Open-Source GPU Work?

[artivision]

Straight answer, NO!!! If you do need our money for a -=good=- reason then you must produce an ultimate Chip inside an ultimate SoC. The Chip must be like something NISC, EDGE, "an ASIC equal" FPGA like Tabula's, Memristor based, spintronic based (like known memory) or something similar - and as simple as it gets. For me its better the Chip to decide the best configuration for MESA and each source-algorithm-program (the less transistor needed), and not to have General + ASIC units. It must have backends for all open compilers and emulation(qemu-acceleration) capability of other ISAs. Must be fusion and have at least 1Tflop/Watt or 10 times that (Toshiba had 1Tflop/Watt dsp on 65nm). Then the SoC must have anything open from Bios to Slots. Also the SoC needs an ultimate Antenna, THz or near signal, laser-like strait line signals Like 5G, gbps or tbps speeds. I imagine the future with antennas(share routers) on every house top, communication between them like metropolitan networks and share with everyone. Long pikes inside a field when houses are not available, even phones can exchange information. To all that, add router internet-cache and you can take down services like Facebook, with also full anonymity. You can also trace your self with known places when they cannot trace you and destroy GPS. I think we can provide you with 5-10 million box if you want to do at least 1/10 of that.

[Delgarde]

Well we really need more details on what they are aiming to do to make any worthwhile comments. However I would assume they would definatly not be aiming for a top end GPU.

Ok, then... are they aiming for a *bottom end* GPU? Because producing something competitive with even the worst graphics hardware available today is still a huge effort, huge cost. And while there are no doubt some hobbyists keen to fund such a thing just for kicks, I figure it's going to be a much smaller number than something like Parallella, which is actually useful to people.

[Ibidem]

First, they obviously already have something that can do OpenGL:
"been around for a number of years...contemplating open-sourcing their 2D/3D graphics engine"

I am curious about some things:
1) What level of OpenGL?
If it's 2.1 or better, I'm interested.

2) How many LUTs/slices? Is this going to work on a Spartan, or will it take a Virtex?
I presume that it's semi-reasonable from this:
"Their open-sourcing would be in the form of Verilog with test benches and the whole shebang, including designs for the Xilinx and Altera FPGAs."

But if it takes so much space that it costs $200+ for a compatible FPGA, you might have difficulties.

3) What license? If it's GPL or another Free license rather than some custom one, that makes a big difference. And in hardware, I'm fine with a dual-license model.

4) What price? Do they need $500,000 for review, or is it $50,000,000?

5) What performance? Just a couple tests of 2d/3d or a comparison to some mainstream gpu will be enough.

6) Are there other features?

7) Current status of Linux support?

Answer these questions and it will be much more interesting. Without performance and GL level, it's not something we can really answer.
Also, if they do go ahead, I'd be willing to go for boards from their clients.

[Caledar]

I think we should do this, even if its only for kicks. How awesome would it be to have open source hardware? Its been done with software and smaller hardware perhaps this is the begining of the completly open source computer?
This could be the begining of a game changer folks.

The question is, can it be done? Theres only one way to find out.

[Ragas]

The problem with Open Hardware in comparison to Open Source is that you can't just build and change it in your backyard a few times a day.

At least if you don't use an FPGA but the tradeoff to that would be crappy performance. (Although there are some intriguing things going on in that space)

[duby229]

I think we should do this, even if its only for kicks. How awesome would it be to have open source hardware? Its been done with software and smaller hardware perhaps this is the begining of the completly open source computer?
This could be the begining of a game changer folks.

The question is, can it be done? Theres only one way to find out.

Thats all fine and good, but we are talking about a financial investment of tens of millions of dollars. If the amount of money required to make such a thing doesnt get raised then the product never gets made. Who cares if it's open source i it isnt being fabricated and manufactured, then marketed and sold.

[timothyja]

The problem with Open Hardware in comparison to Open Source is that you can't just build and change it in your backyard a few times a day.

While it is true that you wont be getting free upgrades or be able to do to much tweaking on the hardware side. However how many of us actually develop and modifiy kernel code other than applying patches (developed by others) here and there. Most of the kernel work is done by commercial companys, this project would be similar in that changes would be mainly worked on by a commercial companys. The point is all those companies would be contributing back to the project with the possible help of PhD students, Uni professors, etc. You could in theory (if you have the knowledge) apply customised changes and get your own batch manufactued who knows maybe even with the help of your own kickstarter.

The other advantage I see is that if this takes off and uses Mesa as a base that would equal another batch of contributors/testers which can only help boost the project.

[timothyja]

First, they obviously already have something that can do OpenGL:
"been around for a number of years...contemplating open-sourcing their 2D/3D graphics engine"

I am curious about some things:
1) What level of OpenGL?
If it's 2.1 or better, I'm interested.

2) How many LUTs/slices? Is this going to work on a Spartan, or will it take a Virtex?
I presume that it's semi-reasonable from this:
"Their open-sourcing would be in the form of Verilog with test benches and the whole shebang, including designs for the Xilinx and Altera FPGAs."

But if it takes so much space that it costs $200+ for a compatible FPGA, you might have difficulties.

3) What license? If it's GPL or another Free license rather than some custom one, that makes a big difference. And in hardware, I'm fine with a dual-license model.

4) What price? Do they need $500,000 for review, or is it $50,000,000?

5) What performance? Just a couple tests of 2d/3d or a comparison to some mainstream gpu will be enough.

6) Are there other features?

7) Current status of Linux support?

Answer these questions and it will be much more interesting. Without performance and GL level, it's not something we can really answer.
Also, if they do go ahead, I'd be willing to go for boards from their clients.

Some great questions maybe Michael could pass them on and report back what they are willing to share at this point.

Ok, then... are they aiming for a *bottom end* GPU? Because producing something competitive with even the worst graphics hardware available today is still a huge effort, huge cost. And while there are no doubt some hobbyists keen to fund such a thing just for kicks, I figure it's going to be a much smaller number than something like Parallella, which is actually useful to people.

Some answers to the questions from Ibidem would be helpful before speculating further. As for Parallella the initial board that was being funded was actually not that impressive its the future planned boards that become more useful so I dont see this project from being to different in that regard.

As said by many, I also don't see much benefit in having a completely open source hardware that will never catch up to the closed ones.

No one can know how far a project can go. Will it ever catch up to the closed source ones who knows, all you can do is support projects you are interested in and see what happens.

1991, Linus announces Linux to the world:
"Hello everybody out there using minix -

I'm doing a (free) operating system (just a hobby, won't be big and professional like gnu) for 386(486) AT clones. This has been brewing since april, and is starting to get ready. I'd like any feedback on things people like/dislike in minix, as my OS resembles it somewhat (same physical layout of the file-system (due to practical reasons) among other things).

I've currently ported bash(1.08) and gcc(1.40), and things seem to work. This implies that I'll get something practical within a few months, and I'd like to know what features most people would want. Any suggestions are welcome, but I won't promise I'll implement them :-)

Linus ([email protected])
PS. Yes ? it's free of any minix code, and it has a multi-threaded fs. It is NOT portable (uses 386 task switching etc), and it probably never will support anything other than AT-harddisks, as that's all I have :-(.
?Linus Torvalds [12]"

[johnc]

waste of time

[timothyja]

waste of time

Much like all of your comments