In the other thread you asked when (in project time lines) might be a good
time to approach Gabe. I think you just answered that. ;) Get the simulator
working smoothly and the basic building blocks "working" so to speak, then
once you have a good estimate of projected power envelope / performance
characteristics you are good to go.

That being said, a Ping there way is certainly tempting.

Gary


On Sun, Dec 9, 2012 at 12:37 PM, Timothy Normand Miller
<[email protected]>wrote:

> I have different short-term goals for the simulator and reference design.
>  For the simulator, the number of stream multiprocessors is just a
> configuration parameter.  We can scale it to be as large as we want and
> simulate any existing GPU's timing characteristics.
>
> For the reference design, I'm going to code up one SP.  Then convert that
> to one SM.  Then build multiple SM's into a thread processor.  In steps.
>  In this case, the target is something like PowerVR's 4-core GPU used in
> smartphones.  My reference design will execute 64 threads at once in a
> single SM, 4-wide SIMT, where the pipeline depth is 14 and there are 16
> quads of threads run round-robin in a barrel processor.  This is all
> RTL-level stuff, but with some work, it could be synthesized and run on
> FPGA or merged into another design on ASIC.
>
>
> On Sun, Dec 9, 2012 at 3:31 PM, gary sheppard <[email protected]> wrote:
>
>> Just need to get a Reference Point end goal jotted down, so you know what
>> it is you are working towards. How much performance does this GPU need? How
>> energy efficient does it need to be? Does it need Compute for physics? Once
>> those end goals are jotted down then it is time to hit the grind. Digging
>> in before you have a set of goals might actually slow things down?
>>
>> When Can I Buy One ;)
>> Gary
>>
>>
>>
>> On Sun, Dec 9, 2012 at 11:19 AM, Timothy Normand Miller <
>> [email protected]> wrote:
>>
>>> Steam is a business that wants to sell games, and they have found that
>>> Linux is a good platform.  They have gotten help from the graphics card
>>> vendors, but they've also had complaints.  If there were an
>>> open-architecture GPU, that they could directly influence, this MIGHT
>>> provide them with some advantages that could positively impact their bottom
>>> line.  Or not.  They still have to support other graphics cards with
>>> proprietary drivers.  Is it worth their time to support yet another, even
>>> if it's easier to support?  But it's not easier, because to maximize their
>>> benefit, they'd have to help develop it.
>>>
>>> Whatever deal we strike with them, our graphics solution has to be a
>>> _direct_ source of revenue for them.  They should not only profit from
>>> selling more games but also from selling graphics cards.  I'm uncomfortable
>>> with the idea giving them full ownership.  But having them profit from
>>> every chip sold (with limitations) is a viable approach.  I think that we'd
>>> actually make MORE money from licensing the design as an IP core -- if they
>>> got in on the ground floor, they could also make millions in revenue from
>>> licensing from companies like Apple, Nokia, and Samsung embedding the
>>> design in their smart phones.
>>>
>>> But do you see how far ahead of things I'm getting?  A lot of this
>>> discussion is nothing but hot air.  I have a reasonable path towards
>>> developing something with short-term utility, and what I need right now is
>>> help on development.  Steam might be sensible to not listen to us until we
>>> have a complete set of masks for a complete GPU in 32nm process technology.
>>>  Nothing is stopping us from getting there besides a bunch of distractions.
>>>
>>> Mind you, someone probably thought the whole Steam idea was a
>>> distraction, while I think it's a good idea, so I'm not saying to stop
>>> brain-storming.  I'm just saying that we should divert more of our energy
>>> towards developing an energy-efficient GPU.  The more we get distracted by
>>> pie-in-the-sky ambitions, the longer it'll take us to make something even
>>> mediocre.  First, we do the things we can do now, with the resources we
>>> have, and then we use those to achieve something else.
>>>
>>>
>>> On Sat, Dec 8, 2012 at 1:23 AM, Gregory Carter <[email protected]>wrote:
>>>
>>>> Itemized List
>>>>
>>>> 1) I would not suggest asking for profits of the game revenue.
>>>> 2) I would suggest the method for accumulating capital for the design
>>>> process be directly integrated into the Steam Checkout process for gamers.
>>>> The idea is to get Windows gamers to donate as well as Linux.  :-)
>>>> 3) Cobranding possibility with Steam games (Nice Animation that comes
>>>> up that says "Works Best with Open-Graphics") for Steam games to make
>>>> average joe look for that card with the projects name on it. Also bring up
>>>> the possibility of selling the card with games made by Steam developers.
>>>>
>>>> 4) I would offer to provide free cards to Steam developers early. In
>>>> consideration of the fact open-graphics will produce a wide range of cards,
>>>> low end, middle end, and high end.  We want driver feed back from the
>>>> developers early so grunts like me and others can fix things.
>>>>
>>>> 5) What will be the consequences of AMD or Nvidia pursuing a lawsuit
>>>> against OpenGraphics for silly patent infringement? Corporations think
>>>> competition is a sin, and prefer fascism. Take note of the Apple/SAMSUNG
>>>> and other lawsuits based on design of product.
>>>>
>>>> Personally there is no way Nvidia and AMD will be able to compete
>>>> against a fully open hardware stack. I also believe secret proprietary
>>>> agreements are bullsh*t. The real reason why they want things secret is
>>>> because not only do corporations hate competition, they steal others hard
>>>> work like bandits.
>>>>
>>>> Keeping things proprietary covers their tracks and keeps them out of
>>>> court.
>>>>
>>>> So I cannot understate the gravity of #3.
>>>>
>>>> 6) What sorts of protections can we institute to protect the capital
>>>> from lawsuits so effectively if AMD or Nvidia don't like our design, they
>>>> will get jack if they sue and win? Meanwhile, the project can continue
>>>> without being affected from those a**holes. I mean where can the hosting
>>>> site be for the project that will be far enough removed from the US
>>>> Department of Commerce and other US corporations that want the project
>>>> shutdown?
>>>>
>>>> What sort of organizational controls as an organization will be
>>>> acceptable for the use of that capital/donations of the Steam
>>>> community/Linux Community?
>>>>
>>>> -gc
>>>>
>>>>
>>>> On 12/07/2012 11:17 PM, Dieter BSD wrote:
>>>>
>>>>> Gregory writes:
>>>>>
>>>>>> Maybe someone should write a letter to Steam and say:
>>>>>>
>>>>>> Dear Steam,
>>>>>>
>>>>>>    Given how Microsoft is trying to screw everyone this Xmas with
>>>>>> Windows 8, and seeing you had such bad luck with the Video blobs from
>>>>>> vendors, but good luck with the open source stuff...really only one
>>>>>> thing is missing.
>>>>>>
>>>>>> An open source card, which, you have a vested interested in helping
>>>>>> fund
>>>>>> or develop with the community.
>>>>>>
>>>>>> If you please put on your Steam page a percentage of how much of the
>>>>>> profits for the Linux games goes to a Chip Foundry fund so you and the
>>>>>> community can get Graphics hardware with 3D capabilities that are
>>>>>> decent, we would wuv you.
>>>>>>
>>>>> Not a bad idea. All we need is someone with ueber diplomatic skills.
>>>>> Which isn't me, unfortunately. Anyone have ueber diplomatic skills?
>>>>> Anyone happen to be friends with people from Steam?
>>>>>
>>>>> Here's my feeble attempt:
>>>>>
>>>>> Dear Steam,
>>>>>
>>>>>       We at the Open Graphics Project have read about the trouble Steam
>>>>> has been having with binary blobs, and your relative success with
>>>>> FLOSS.
>>>>> We are working on developing a graphics card that is completely
>>>>> documented,
>>>>> so that FLOSS OSes can have proper device drivers that work correctly.
>>>>>
>>>>>      As you are doubtlessly aware, the lack of a properly documented
>>>>> graphics card is a major sore point in the FLOSS community. Sponsorship
>>>>> of a "Steam Powered" graphics card would generate an enormous amount of
>>>>> goodwill. We look forward to discussing the possibility of working
>>>>> together
>>>>> with you at your convenience.
>>>>>
>>>>> ------
>>>>>
>>>>> Ok, perhaps some diplomatic wizard can fix that up.
>>>>>
>>>>> What do we need to do to prepare, in case they are actually interested?
>>>>> Will they want a demo of the OGD1? A detailed business plan?
>>>>> Something else?
>>>>> ______________________________**_________________
>>>>> Open-graphics mailing list
>>>>> [email protected]
>>>>> http://lists.duskglow.com/**mailman/listinfo/open-graphics<http://lists.duskglow.com/mailman/listinfo/open-graphics>
>>>>> List service provided by Duskglow Consulting, LLC (www.duskglow.com)
>>>>>
>>>>
>>>> ______________________________**_________________
>>>> Open-graphics mailing list
>>>> [email protected]
>>>> http://lists.duskglow.com/**mailman/listinfo/open-graphics<http://lists.duskglow.com/mailman/listinfo/open-graphics>
>>>> List service provided by Duskglow Consulting, LLC (www.duskglow.com)
>>>>
>>>
>>>
>>>
>>> --
>>> Timothy Normand Miller, PhD
>>> Assistant Professor of Computer Science, Binghamton University
>>> http://www.cs.binghamton.edu/~millerti/<http://www.cse.ohio-state.edu/~millerti>
>>> Open Graphics Project
>>>
>>>
>>> _______________________________________________
>>> Open-graphics mailing list
>>> [email protected]
>>> http://lists.duskglow.com/mailman/listinfo/open-graphics
>>> List service provided by Duskglow Consulting, LLC (www.duskglow.com)
>>>
>>
>>
>
>
> --
> Timothy Normand Miller, PhD
> Assistant Professor of Computer Science, Binghamton University
> http://www.cs.binghamton.edu/~millerti/<http://www.cse.ohio-state.edu/~millerti>
> Open Graphics Project
>
>
_______________________________________________
Open-graphics mailing list
[email protected]
http://lists.duskglow.com/mailman/listinfo/open-graphics
List service provided by Duskglow Consulting, LLC (www.duskglow.com)

Reply via email to