March 1st 2006 Open Graphics Project Newsletter

• March 1st 2006 Open Graphics Project Newsletter
• ANNOUNCEMENTS
  • OGD1 Revision B
• Licensing
• OGD Alternative Usages: the wild, the crazy and the interesting
• Progress and developments
  • OGD1, the beginning
  • Geda Schematics
  • OGP and Traversal
  • Web Presence
  • Qualifying as OGP Developers
  • Pricing and Sales
  • Website Sales
  • Licensing
  • Drivers
• Hardware Selection and Revision
  • Hardware Clarifications
• Who is Who..
• OGP in the News

ANNOUNCEMENTS

OGD1 Revision B

There is a revB of the OGD1 board

http://natsuki.kinali.ch/OGD1_Schematic_RevB.pdf
http://natsuki.kinali.ch/OGD1_RevA_BOM.txt
These have been checked into the subversion tree

P. Brett is maintaining documentation for the Bill of Materials (BOM)
http://www.peter-b.co.uk/downloads/opengraphics/OGD1_RevA.2_BOM.ods

Licensing

T. Miller

• The IP being licensed here refers to schematics, artwork, and other information necessary to reproduce the OGD1 blank printed circuit board. Additional IP that may be necessary for manufacturing a fully functional device (such as bit files and firmware) is not covered here.
• Traversal retains copyright of this design and has the right to privately and publicly license it any way they see fit, as well as produce derivative works without restriction, license others to have similar rights, etc.
• This design is being released to the community under the GNU General Public License (GPL).
• By default, Traversal reserves the right to adopt into their privately licensable IP any changes or modifications offered by the community to the OGP, Traversal, or an agent of either. All such changes will also be released under GPL immediately. Trademarks associated with Traversal Technology and the Open Graphics Project are barred from any other use of this IP.

OGD Alternative Usages: the wild, the crazy and the interesting

• It's a PCI card with FPGA and DVI connection with future good software support
• accelerating encryption, running arm code, maybe a hardware-accelerated jvm
• an Open Gate - Array Project
• an Ethernet - to - video bridge
• With an ADC daughter board - a 60-channel audio board Video Capture and/or TV in
• a high-performance motion controller

Progress and developments

OGD1, the beginning

OGD1 is the name of the first board to be made. T.Miller clarified its purpose: We're trying to sell it as an FPGA development board. And some of them will make their way into the hands of OGP developers for a much lower cost. ... It's simply a step in the process of DESIGNING a graphics card that happens to be a good thing to sell as a product.

Geda Schematics

P.Brett is working on a geda version and asked for volunteers to assist. He gave an update of his progress and reported that he had completed just over half of the symbols. P. Sartain offered his help and more are welcome. He then itemised some things where assistance is wanted:
1.component verification
2.providing raw text files to make components out of and manually editting of those components for consistency & aesthetic purposes.
email me if you need help setting up & working gschem for some of these manual labour tasks =)

OGP and Traversal

Miller was asked about the relationship between OGP and Traversal:Does that mean that it's supposed to be like the Mozilla Foundation to Mozilla, or more like what Netscape is to Mozilla? Or both?

That's a good question. I'm not sure. I suspect it's more like the Netscape - Mozilla relationship. The difference is that Netscape existed first. In this case, OGP existed first and will therefore have more control over what Traversal does.

Here's something funny about it. To protect the interests of the OGP, Traversal will release chip IP under the GPL. But Traversal will also have the ability to license that same IP under other terms, because Traversal holds the copyright. Of course, since OGP controls Traversal, OGP has no less control over that IP. It's a bit of hand-waving put in place so that Traversal looks like a regular company to other companies that we'd want to do business with.

(When I say that OGP controls Traversal, I mean in a community/democratic sense. I don't mean that we'll do any random thing that any random person suggests on the list.) Corporately, I want to position Traversal as the company that implements stuff that the FOSS community needs. Hypothetically, we could build proprietary products that are unrelated, but tending to the FOSS community is a better use of resources, in part because the market research is free.

So, when OGA/OGC (...graphci boards are...) selling, one thing we'll do is start work on the next generation. How will we find out what to PUT into that new device? We'll let the community decide. If we can grow as a company, we can add more product lines, ones decided by the needs of the community. When we hire people, the first people we'll consider are ones who have been volunteers with the community. Traversal will be built from the FOSS community.

This democratic thinking is why, for instance, I brought up the issue of pricing OGD1. Theoretically, we could figure out a price some other way, but it's better this way. Now that we have out of the way the concern of how OGP developers will get boards (they'll get them one way or another), we can consider how we should price the product for the market. What should WE (OGP and Traversal) sell the product for to THEM (everyone else, mostly chip designers and universities)?

Money is one of those difficult issues, though. Since Howard, Andy, and I are putting up the initial money, it makes sense that we manage the profit from it. But without the community, that money wouldn't have come in. So we use that money very carefully to grow Traversal, OGP, and our relationship so that everyone benefits. As for me, I'd be happy if I made a reasonable salary working for Traversal, but I'd be even happier if lots of other people dedicated to this project could get the same thing.

N. Boulay suggested calling it a foundation to make it more clear, However T. Miller said I want Traversal to have growth in mind. This kind of growth would give us more and more ability to develop a more diverse array of FOSS-friendly products.... My main point is to say that I am not being greedy.

F.Cami suggested to copy from the LML ( http://www.linuxmedialabs.com/ ) ... They state explicitly their goals on their website ( http://www.linuxmedialabs.com/About_LML.php?subject=about ) :"The competitive advantage of our boards is in the use of the GNU General Public License, which significantly reduces software driver development costs while providing the highest software code quality possible." A similar statement should be in the "About Traversal" section of the upcoming website, and I believe this would be enough. Oh, and as stated earlier in the same thread, "the FOSS community has developed a general fear of 'evil big companies'" (my emphasis on big). Traversal is a long way from being NVIDIA or ATI, size-wise.

T.Miller added Howard, Andy, and I are Traversal, ...What I think I would like to have happen for ME, to be perfectly honest with you, is to make a very modest salary from this. I would like to work for Traversal as my full-time job. That would be a nice thing. This may taint what I do. I don't know. But keep in mind that this started as a Tech Source project, and Tech Source management wasn't thinking altruistically about this.

It seems logical to me that to do the kind of business deals and partnering necessary to fund the ASIC development, there needs to be a business. Whenever I make a decision, as part of that business,| I will endeavour to ensure that the OGP is helped by it. But Traversal will do other things that are neutral to this.

T. Miller further qualified the relationship between OGP and Traversal:
Traversal is just a corporate front for the OGP. It's there to centralize "corporate stuff". Naturally, it'll take on a life of its own in some respects, but it'll be a long time before we release any product that isn't based on some sort of community effort. ... I would like to have happen for ME, to be perfectly honest with you, is to make a very modest salary from this. ... Since Howard, Andy, and I are putting up the initial money, it makes sense that we manage the profit from it. But without the community, that money wouldn't have come in. So we use that money very carefully to grow Traversal, OGP, and our relationship so that everyone benefits. As for me, I'd be happy if I made a reasonable salary working for Traversal, but I'd be even happier if lots of other people dedicated to this project could get the same thing.

Web Presence

T. Miller announced the project has been offered server space for a web site and L. Veen provided an example mockup: You can find it at http://nova.student.utwente.nl/~lourens/ogp/template.html In case you were wondering, yet it does pass the W3 validator. Dolphining added some perspective: While the website should absolutely *use* correct HTML, OGP is a graphics hardware project, not a web project, F. van Damme offered to assist with the website.

Qualifying as OGP Developers

Some suggestions about who qualifies as an OGP developer (and a discount) were
R. Heasman I think we want, at the very least:

1. HDL code
2. Drivers for Linux
3. BIOS code
4. Layout/Schematic design
5. Documentation for all of the above
6. Documentation for users to this was added
7. HDL code to stress-test OGD1 boards
8. Software code to stress-test any OGP/Traversal product and
9. Documentation for all of the above
10. Documentation for users

9/10 : Translations in other languages (it was agreed that an automated translation could potentially be much worse than no translation at all.

Pricing and Sales

T. Miller explained that at this stage: The price isn't totally determined yet. We know the cost of the BOM, but the assembly cost isn't settled. Topics dwelt on how the final cost will include assembly and sufficient profits to benefit the next stage of development. ... If Traversal ...dies..., OGP loses its centralized corporate wing. So you get OGC1 (...the first Open Graphics Card...) more cheaply. But what about OGC2?

Dieter made a suggestion: Have you considered a "Dutch auction"?
Several liked this suggestion and T. Miller asked How do we get started on that.

Atilla wrote: Define a target, open up some bank accounts, register a pay pal account and finally set up a web page. BTW: a good starting time for this would be while releasing the schematics. Why? We will have quite a lot of people visiting the web page and they'd see that we have something that's worth some money. And best: no additional announcement needed. Several discussions ensued.

Website Sales

FD Cami offered to set up a website for sales using oscommerce

Licensing

There had been some consideration to releasing the PCI core under GPL. T. Miller explained By releasing it under GPL, we can still share it under terms that FOSS people like. But for anyone who doesn't want to infect the rest of their design, they can pay for a proprietary license.

Drivers

One correspondent asked if making a driver would be the same as linking a program against a gpl library. Asking if the license would taint the code, which would prevent the ...making of a... closed source driver. N. Boulay disagreed No it's not. The card is hardware. The GPL is a licence which is based on copyright. Hardware are not manage by copyright law. T. Miller felt an argument could be made that the register spec for OGA is just a list of facts, and you can't copyright that. Anything based on that, whether that be another chip or a driver for our chip, is not affected by the copyright license for the chip... the ...document with the specifications would... be copyrighted. How we license it is unlikely to create any significant restrictions for anyone, although we may find some way to make some small amount of money from it. Perhaps we'll do free downloads but charge money for hard copies or something. But just as when you pay to have a copy of the PCI spec, that copyright protection doesn't apply to anything you build that conforms to that spec.

Hardware Selection and Revision

T. Miller posted: design changes you may notice:

• We have switched our main FPGA from Xilinx to Lattice. Compared to the Xilinx 3S4000, the Lattice ECP2-50 has more logic area, is less expensive, and the synthesis tools are free of charge. Unfortunately, it has about 1/3 fewer usable I/O pins, but we've dealt with that as best we can.
• Primary "user I/O" bus from the main FPGA is 66 pins. Another set comes from pin banks whose reference voltages may or may not make them useful to you. Another set comes from unused pins on the Lattice XP10. Additional ones can be connected to unpopulated pads for optional parts.
• The "local bus" interface between the XP10 and the main FPGA had to be reduced in pin count. The result is that we will need to use a bit more logic to multiplex traffic across this bus at a higher data rate.
  • Some of those signals are connected to pins on the main FPGA where the PCB routing will be suboptimal; as a result, those signals will have to work at a lower clock rate.
• The PCI card edge has been extended to the full 64 bits and can be configured to indicate to the host that it supports PCI-X (133 Mhz). (The RTL code we develop for OGA may or may not support the extended bus width.)
• The Xilinx had an expensive prom, so we were going to use the Lattice to program it from the same PROM that held VGA. The ECP2-50, on the other hand, can program itself from an SPI PROM directly. We now have two SPI proms. One for VGA, and one for the bitfile. There's still reason to have the smaller FPGA, though. The time it takes for the ECP2-50 to program itself is long enough that some systems may not find it on the bus. The smaller one is flash-based and therefore instant-on. Note that you can still reprogram the big one from the small one, in a couple of different ways.

The intention is to allow the following methods to reprogram the SPI PROM on the big FPGA.

1. Via JTAG from ispLEVER software.
2. Via JTAG controlled from the small FPGA.
3. Via register interface accessible from PCI.

We have a header on the board that connects to the main FPGA. That header is designed to connect to a Lattice programming cable that you can use to directly program the FPGA. Due to signal integrity issues we decided not to connect that also directly to the smaller FPGA. BUT, what we've done is put another headers directly parallel with the first one. This way, all you need to do is install a set of jumpers, and the big FPGA's programming interface is connected directly to the small one.

And of course, the small FPGA can reprogram the SPI prom that the big FPGA uses to program itself on startup. Without the jumpers, the header connected to the small FPGA can be used for user I/O.

H. Parkin: You definitely need 7 layers to get all the balls of the big FPGA out, unless you use blind vias, which increases the price of the PCB, in some cases by a factor of 2 or more.

R. Jarquot asked why is there 2 different designs for the analog output ? (one using the TDA8777 and the other using 3 max5886) A. Kinali The one with the TDA8777 will be populated and will be the only VGA output for OGD. The one using 3 MAX5886 has a higher bandwidth (500MHz instead of iirc 300MHz) and could be used for analog HDTV output, but as it is too expensive, it will not be populated.

Hardware Clarifications

MMB7000L refers to MMBD7000L
LatticeECP2 ECP2-50.
latticesemi.com/documents/LatticeECP2_HB.pdf
http://www.pericom.com/pdf/datasheets/PI5C3302.pdf

Who is Who..

R. Miller is caring for the wiki

OGP in the News

• http://osnews.com/comment.php?news_id=13844
• http://www.linuxtoday.com/developer/2006030201526OSKN