Let's not make this so overwhelmingly large that no one sees the end. That's one of the problems we had with the OGP originally.
Let's break this down into some steps: 1. Design simulator for GPU 2. Design RTL for GPU 3. Get it to work in RTL-level simulation 4. Get it to work on FPGA Once we've proven the design in FPGA on a smaller scale, we can move on to the extremely expensive step of making an ASIC. By that point, I want commercial partners and possibly grant funding, because this is what we need to get real energy numbers for research purposes. And reliability numbers for doing things like running at near-threshold voltage. Indeed, the first real hardware should include lots of other circuits for testing things like systematic process variation, canary circuits, voltage and temperature sensors, etc. So, let's break "1." down further (and by extension also "2.", which has the same components): a. Design shader engine (stream processor) (we have already specified this and have simulator code) b. Design stream multiprocessor and local memory arbitration c. Design multi-SM GPU compute array and NoC d. Design host interface e. Design memory system These are the normal steps to go through. One challenge is that we need the participation of chip designers. So far, you have me, and I don't scale very well. But even before that, we need people to just do some software for the simulator. I'm trying to get help from students at Binghamton, but let's not sit back and rely on that alone. This is a FOSS project with FOSS goals, so we should get FOSS enthusiasts to work on it. On Fri, Nov 2, 2012 at 3:37 PM, Troy Benjegerdes <[email protected]> wrote: > ... replying because I typoed the oc-team at opencores email, and > I would very much like to hear their response as well... > > On Fri, Nov 02, 2012 at 02:31:00PM -0500, Troy Benjegerdes wrote: > > So let's suppose we target tapeout of the chip by July, 2013. > > > > What would it cost (in USD) to contract someone to get the OpenShader > > GPU in verilog fully validated by then? > > > > I'm also very interested in what it will cost to get all the pieces > > in place to have HDMI and a full gigabit PHY (with open-source of > > whatever magic needs to happen to make that work) for the rest of > > the SOC design. > > > > If we are talking about SoC's, are we including power sequencing, > > or would be be better to us something like: > > http://www.ti.com/product/twl6032 > > > > and then work on an open-source power controller chip as a separate > > project? > > > > If we have some (strawman) cost numbers, and a timeframe, this gets > > to be a lot better plan for a kickstarter project that I think could > > deliver real working hardware AND get lots of people really excited. > > > > On Fri, Nov 02, 2012 at 01:04:09PM -0400, Timothy Normand Miller wrote: > > > Well, we have enough knowledge and defined architecture that we could > > > probably code up a reasonable GPU in Verilog already. It's a matter of > > > allocating time and people to it. > > > > > > > > > On Fri, Nov 2, 2012 at 9:30 AM, Troy Benjegerdes <[email protected]> > wrote: > > > > > > > http://opencores.com/donation > > > > > > > > Seems like they are missing a GPU component. > > > > > > > > Can someone here comment on the project and what it would take > > > > to get the OpenShader GPU included in the opencores asic? > > > > _______________________________________________ > > > > 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) > -- 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)
