Announcement

Collapse
No announcement yet.

Would A Kickstarter Open-Source GPU Work?

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • glococo
    replied
    I agree with fbruno, should be GPL, (avoid lgpl and weak licences), so each manufacturer/developer should release the improvements to the community.
    If we support an opensource Gpu, they should also do the same.

    Keep strong fbruno !

    Leave a comment:


  • insink71
    replied
    Support

    I am in favor of running the gambit on Kickstarter. If I was one of the lucky three that just won the Powerball, I'd even kick in big time. Unfortunately, I'm a poor Android developer who encourages openness every chance I get. So, those opposed say they get enough out of an AMD or Nvidea card with open source drivers. Those for say imagine the possibilities of open hardware. Well, in a way, they're both a little off, but I side with the "imagine the possibility" crowd.
    First off open source drivers are good, but do you know how much work that takes? When an AMD or Nvidea (or Qualcomm in my case usually) changes the architecture of their video cards, its binary or bust. For gamers, this is typically your plight. When these events occur (and they occur quite frequently), the respective efforts and teams go about reverse engineering the drivers the best they can, implimenting shaders, then 3d if possible, etc. It is a ton of work. I applaud these open source guys for their work.
    In the case before you guys, these guys have done all that work on a development platform. A few have said I don't know what a FPGA is.. Welp, its a Field Programmable Gate Array. Its made for hardware developing and (once the hardware is developed like in this case) it is also made for rapid prototyping of hardware modules. Now, one thing, the major FPGA companies are for developing whatever kind of hardware you want to develop (including open hardware), but the FPGA itself is probably not open hardware. Alterra, Xilinx, etc.. while they promote openness; they also protect their own respective trade secrets (much like the GPU companies do); so, I doubt the initial release (the FPGA) would be open hardware (although it could be used to develop open hardware indeed).
    Why would I be in favor when there are so many who don't seem to care? Welp, the GPU to me is the single biggest hurdle in gaining a fully open sourced CPU. Yes, Oracle and OpenCores (and others) have made open source chipsets. But those that said they would buy in if it were a fully open sourced system, but wouldn't for just a GPU are short-sighted. With an open sourced GPU (which could be used to develop open hardware), you'd have the last piece of that puzzle. How are you going to have your (fully open sourced) computer with no display? To reiterate the frustration level. I will cite that JBQ (head of AOSP development at Google) is quitting; yep quitting, over this very issue. They are having to rely on binaries from Qualcomm, and their flagship device(s), the new Nexus 7, are not booting because the GPU driver isn't working. Its sad and frustrating and you won't find open source drivers for the latest and greatest GPUs out the box. They have to be painstakingly reverse-engineered as mentioned before.
    So, I am glad someone with the knowledge has decided they'd like to contribute back to the community. I hope that they can convince their partner. And I hope people see this opportunity and chip in on a Kickstarter event. Post the thread to said Kickstarter if the partner is convinced and the project is considered for open source. I'll kick in a wee bit because a wee bit is all I have to contribute.

    Regards,
    Rob

    Leave a comment:


  • fbruno
    replied
    Thanks for the ideas. I will definitely get in touch with both RMS and Jim. I'd like to get some of our materials ready to show them, so it'll probably be a couple of weeks.

    One of our goals for this is to gauge support. We'd like to do an OpenCL core if this is successful. Maybe we'll even make it a stretch goal. The only downside is that the cores I'm talking about exist and we can clean them up and release them in a month or so. Any new development will take time.

    Thanks for the support!

    Leave a comment:


  • kyzh
    replied
    Actually you might want to contact people in the open source world if they would be keen to step in and contribute.
    You could probably find support in the open source hardware, ultimatly they try to achieve what you propose.

    There is actually some profitable company versed in open source, asked them if they can back you/ help you.
    Ask any of your connection in companies/organisation like Red Hat, Mozilla, Google, FSF, O'Reilly etc... you might find someone with some cash or some publicity capability.
    Jim whitehurst said once that he is always looking for good idea to back, i'm sure that he would give you 140 caracter to pitch your idea, he will read it.

    Also RMS should be over the moon that you think about open sourcing hardware. Did you try to get in touch with him ?
    I would imagine that he is the best person to get people to put their money where their mouth is and get the word out .

    I would spend some time on lobbying those guys a little bit, worth case you lost a month, best case you got some influancial backers.
    Some of the afromentioned company can definitly help you with licensing and finding the right one for you that also work for the rest of the world.

    I hope you will be successful, and i'll do my bit to help

    Leave a comment:


  • fbruno
    replied
    Originally posted by glococo View Post
    How many kLE are necessary for an OpenRisc + FBrunoVGA + peripherials implementation ?
    For what I see in OR2K document and fbruno specs, 150kLE will be more than good for starting point of a dev board, is it right ?
    Unfortunally these fpga are not very cheap but not expensive at all for what we want.



    Which is the minimum trace width and layers for a PCB with FPGA like this?
    150Kish luts should be enough to build a system w/ 3D. 2D would be a lot less.
    We are currently looking at the Altera or Xilinx SOC boards for the hard processors and converting the IP to use AXI. With the conversion it will also be easy to convert it to Avalon or open interface.

    We are still deciding whether we will build a board initially or support some eval boards from Xilinx and Altera, possibly building a board later.

    The good news is that my business partner has jumped aboard and we will be doing a video and putting together the kickstarter in the next month. I think the videos will show it all as we have numerous (windows) 3D benchmarks we can show. The linux driver needs work to get the 3D working, but we will show the 2D running.

    Leave a comment:


  • glococo
    replied
    How many kLE are necessary for an OpenRisc + FBrunoVGA + peripherials implementation ?
    For what I see in OR2K document and fbruno specs, 150kLE will be more than good for starting point of a dev board, is it right ?
    Unfortunally these fpga are not very cheap but not expensive at all for what we want.



    Which is the minimum trace width and layers for a PCB with FPGA like this?

    Leave a comment:


  • fbruno
    replied
    Originally posted by Setlec View Post
    curious about the cost of those boards, why are they expensive and what do they offer as specifications?
    They are generally expensive as they support many peripherals to show off the FPGA capabilities. They also have very high density connectors which allow expansion. Typically they include FPGA development tools. Also, a part large enough for OpenRISC and the Graphics would be on the order of the largest inexpensive parts.

    Expensive is relative. Here are some examples: http://www.mouser.com/Embedded-Solut...ols/_/N-8mwzo/

    Bear in mind the board required would be 100K LE or greater, Cyclone or Spartan family.

    Leave a comment:


  • Setlec
    replied
    Originally posted by fbruno View Post
    That is true. The boards are expensive, though. So possibly it comes down to making our own w/ just the components we need to make it work. Something to think about.
    curious about the cost of those boards, why are they expensive and what do they offer as specifications?

    Leave a comment:


  • fbruno
    replied
    Originally posted by curaga View Post
    If there already exist suitable boards, then that's less work for you, no? You could get a bulk discount on one, then offer it preinstalled with the opened core as a kickstarter perk.
    That is true. The boards are expensive, though. So possibly it comes down to making our own w/ just the components we need to make it work. Something to think about.

    Leave a comment:


  • curaga
    replied
    If there already exist suitable boards, then that's less work for you, no? You could get a bulk discount on one, then offer it preinstalled with the opened core as a kickstarter perk.

    Leave a comment:

Working...
X