AboutOpenGraphics

Back to Open Graphics
fr:Sur le projet OpenGraphics

Who are we?

This project was started by Timothy Miller, an employee of TechSource, who produce a variety of specialized graphics hardware and are friendly to this open graphics project. The hardware side of the Open Graphics Project will be undertaken by Traversal Technology, a separate company created by Open Graphics Project members Mr. Miller and others.

Many others participate in the project, see ProjectMembers.

Mission

Availability of a graphics card with fully published specs and drivers that are free software.

Vision

There is a market for graphics hardware with good support for free software and free operating systems (there may or may not be a market for open graphics hardware also, but that is beyond the scope of this project). Such a graphics card would benefit from lower software development cost and mindshare in order to be commercially viable. Free software could benefit from the active cooperation of the manufacturer of such a card to create better drivers and to get a card that better meets the requirements of free software.

Currently, the market for such cards is not served very well. NVIDIA has no offering in this market, ATI's older cards have very limited support, while their new ones have none, and Matrox has no offering in this market either. XGI are off to a good start but still no 3D code yet.

In order to get manufacturers to make such hardware, we have to show that it will be economically viable to do so.


The Project volunteers and the associated business have complementary goals. We cooperate so that both can succeed. This inevitably involves some compromises, and we believe we've done some pioneering in working out an effective relationship.

The community needs include:

Openly published documentation for the hardware's logical, electrical, and mechanical interfaces, so that the community can maintain drivers and other software during its full service life while realizing the full performance potential, and so that the hardware can be correctly integrated into larger systems

The ability to provide technical support during the full service life, regardless of whether the manufacturer is able to do so

The freedom to modify and redistribute supporting software as needs evolve

The freedom to re-use all volunteer-contributed parts of the design in fully open designs

The business needs include:

Assistance in product development, to conserve start-up capital and reduce time to market and break-even selling price

Identification of the actual product requirements of potential customers

Assistance in documentation and technical support, to meet the needs of new users more rapidly

In order to be interesting to the free software community and OEM vendors such a card should at a minimum meet the following requirements:

• Programming interface must be fully documented
• No IP encumberment for implementing drivers
• Very good 2d graphics performance
• Full OpenGL implementation with as much hardware acceleration as possible
• Good support for xv (yuv->rgb, scaling) for video playback
• Reasonable price!

The design up to this point

A fair majority of responders pointed out that 3D, at least partial support, is absolutely necessary for modern desktop systems. MacOS has already gone fully 3D with their desktop, and various experimental desktops in the free software community are going that direction as well. In addition, the Composite, Damage, and Render extensions of X.org require some 3D functionality. The number of 3D features necessary is enough that we have decided that a fully 3D engine with 2D emulation would be the best approach.

The design so far is a 3D rasterizer and fixed-function fragment processor that conforms to most of OpenGL 1.3 plus some later features. (We are working from the 2.0 spec but leaving out features deemed "not important for most uses.")

Restrictions

• While the hardware will be open spec, not all of it will be open source at first. This gives Traversal Technology the advantage it needs to recoup its non-recurring engineering (NRE) costs. (Edit by Tim: We have committed to releasing all of the RTL to the chip. Some of it is being developed in the open already. Some of it may be released at the time the ASIC is released. Some of it will be released after a time delay necessary for said competitive advantage.)
• Due to market size it will not be possible to compete on 3d performance with market leaders such as ATI and NVIDIA. This is not an immediate problem because gaming is not what this card is aimed at, but performance should be good enough for scientific simulations and similar.
• An initial version of such a card will likely be limited to being single-headed with a single video-output. TV-out may be an option but may not be usable together with vga out at the same time
• This project is currently limited to an initial product offering with a PCI-based variant with PCI-express as a follow-up after product viability has been secured.

Feature Set

• The feature set is still being refined; please see the FAQ and the mailing list archives for specifics.
• If review of these resources fails to identify a feature that would appeal to the target market, please feel free to bring that feature to the attention of the mailing list.

Target Market

• The target market for this product is currently identified, but not limited to: Enthusiasts, Universities, Linux OEM Integrators, Graphics specialists and other niche markets.
• The target market for this product is NOT to be a direct competitor to the mainstream graphics market. (i.e. Nvidia, ATI, Matrox)

Economics
We must find a technical solution which makes the appropriate tradeoffs so that we can produce a viable product at a reasonable price. The parts costs has to be minimized, and the logic design must fit into the parts. This will be a challenge.

Many hardware vendors may see the free software community as a minority that asks for special attention. Though that attitude may be misguided, the result is that they see Linux users as costlier, at least indirectly. Traversal doesn't see it that way. As the user community, we need to be clever, fair, and realistic about the requirements we place on the hardware that is being designed here. This is an opportunity for the Free Software community to prove that their ideals are right and that capitalism and free society would be enhanced by them.

For more information, see the FAQ .
There is a mailinglist for this project at http://lists.duskglow.com/mailman/listinfo/open-graphics