Interview: Daryl Strauss, Precision Insight
Want to find out what's happening with 3dfx graphics hardware and the port to Glide? Read on.
The last time we talked to Daryl Strauss, he was working for Digital Domain, helping them generate graphics for the movie Titanic--that was February 1998. Daryl has since moved on to work for Precision Insight, where he has continued to work in the area of 3-D graphics. Steven Pritchard talked to Daryl about his current projects at the end of last year.
Steven: First of all, for those who don't know you, can you give a little bit of background on yourself? Specifically, how did you get involved in porting Glide to Linux, how long ago was that, and what exactly is your job description now?
Daryl: I started working on the 3dfx hardware just about three years ago. When the original Voodoo graphics came out, I looked at the specs and noticed that the performance numbers were very close to the original SGI reality engine. Since I knew the reality engine was a box that you could do real work on, I knew the Voodoo Graphics would be a very capable card. It literally changed the way the PC industry did 3-D graphics.
I ran out and bought an Orchid Righteous 3-D for $300 when they first became available. I wasn't disappointed. 3dfx has made the developer kit available to anyone from the very beginning. I downloaded it and noticed they had mentioned Linux in the documentation as being TBD. There were also a few #ifdefs in the headers. So, I knew someone there understood Linux. I went ahead and applied to their developer program to port the libraries to Linux.
It took about six months to work out all the details and about two weeks to do the first port. I think they were sort of surprised it happened so quickly, because there was another delay before the first version got posted on a web site.
Since then, I've continued to support the rest of their boards. This was always done as a volunteer effort in my spare time. Sometimes my real job got in the way. The work on the Voodoo Banshee took me a long time to get done, because it required writing an X driver (which I had never done before) and I was very busy creating the special effects for Titanic while that was happening.
Back in April, I realized that Linux was ready to have a large impact on the 3-D graphics industry. I left my job as software manager at Digital Domain and joined Precision Insight. My title there is Visual Mason and Evangelist. In June, Precision Insight signed a contract with 3dfx to develop a DRI implementation for their Voodoo Banshee and Voodoo 3, 4 and 5 hardware.
Steven: What exactly was just released today? What cards are supported? When do you anticipate a “production” release?
Daryl: This is a pre-release of an implementation of the OpenGL API for the Voodoo Banshee and Voodoo3 based on Precision Insight's Direct Rendering Infrastructure. That's quite a mouthful. What that means is users can run applications written for OpenGL in a window on a Linux system with a Voodoo Banshee or Voodoo3.
Another key feature of this release is the entire implementation is open source. Now that the basic capabilities are running, we're going to set it up as a public project so that anyone who wants may contribute.
It is still a pre-release, which means we know it still has bugs. We're getting it out there so that people can start trying to run their applications and see how they behave. We want to identify any issues. The final release will be in 2000.
Steven: Do you know if 3dfx takes Linux seriously, or if they just consider it a niche market?
Daryl: I can't speak for 3dfx. They were the first, and one of the few companies, to put money into a fully open-source solution for Linux. That's a big level of commitment. Some companies have released some code and some have released some specs, but few companies are making code and specs available and putting money into the development.
With that said, Linux is somewhat of a niche market today. Linux systems represent a very small number of sales. 3dfx realizes this is changing quickly. They want to get involved early so they can catch Linux as it grows.
Steven: How much do you know (or rather, how much can you tell me) about the new chip set and boards that 3dfx just announced? Are Linux drivers being developed? If so, will they be ready when the boards ship?
Daryl: 3dfx has published a lot of material on the new Voodoo4 and Voodoo5 chip sets. The key component is what they call the Voodoo Scalable Architecture. The first chip is the VSA-100. You can put between one and thirty-two of these chips together on a board. The fill rate on each chip (the number of pixels that can be drawn on the screen in a second) is a very impressive 333 megapixels. They've also improved the capabilities of the chip from that of the Voodoo3; it does 32bpp, stencils, 2048x2048 textures, texture compression, more color-blend modes, and a few powerful features called the T-buffer which allows full-scene hardware anti-aliasing. These boards will be very compatible with the OpenGL API.
Precision Insight will be working with 3dfx to support the Voodoo 4 and Voodoo5 boards. At this point, we haven't announced a release date for that software.
Steven: The specs on the Voodoo5 6000 look fairly impressive. Do you know if 3dfx is still just targeting gamers, or are they planning on targeting more “professional” 3-D users?
Daryl: The distinction between “gamers” and “professionals” is sort of artificial. What really matters is the sort of application being run. There are professionals who develop simulations (such as flight simulators or arcade systems) for whom these will be excellent boards. Quantum3D is a partner with 3dfx and will sell high-end configurations of the VSA-100 chip for this sort of user.
The real question is whether your application is fill-rate limited or triangle-rate limited. Digital content creation systems (CAD, 3-D modelers, etc.) are more likely to be triangle-rate limited. They have to draw many very small triangles. On these boards, all that work is done on the host CPU, and even with the very fast CPUs now available, it is likely that you couldn't push them through to the graphics board fast enough to keep it busy. If your application draws bigger triangles or needs to do a lot of work with multiple-texture maps, it might be able to take advantage of the extra fill rate in these boards.
So, to answer your question, yes, I think these will be reasonable cards for any application, particularly with the new features they've added to the architecture. But, 3dfx is targeting the consumer, and that's where these boards will work the best.
Steven: What brought on the sudden decision to open the source to Glide?
Daryl: You'd have to ask them for a real answer. From my discussions, it is mostly that they want to have their hardware supported on as many platforms as possible, and open sourcing Glide seemed like the best answer.
Steven: What license is 3dfx using for the Glide source? Will it be possible to include it with the Mesa (or X) source?
Daryl: They made their own, but my reading was that it is basically LGPL. That means it can't go into X. It could go into Mesa, but I'm not sure we truly want that. It makes a reasonable stand-alone package.
A different, more ambitious project would be to remove the dependency on Glide from Mesa. That's a big project, but it might be interesting for people to pick up if they want to learn more about writing drivers.
Steven: Are drivers for any of the extra features, like the TV tuner on the V3 3500, being developed?
Daryl: Not by us. I believe they want to release the specs on them. The TV parts are a standard one for which you can get the data sheet. 3dfx will be releasing more specs fairly soon, so that may have the TV details. I'm not sure.
Steven: Was Precision Insight contracted to work on the Voodoo drivers on an ongoing basis or just for the initial implementation?
Daryl: We have a contract that includes work on the V4/V5 and to do some work on future boards. We may see more work after that, but this contract covers us for a reasonable period of time. We're not being contracted to do the end-user support. 3dfx will be doing that themselves.
Steven: What other companies have contracted Precision Insight to write DRI drivers? Are you working on drivers for any other cards?
Daryl: The only companies that have announced anything publicly are 3dfx, ATI and Intel.
Steven: Keep up the good work, and thanks for talking to us.
|Designing Electronics with Linux||May 22, 2013|
|Dynamic DNS—an Object Lesson in Problem Solving||May 21, 2013|
|Using Salt Stack and Vagrant for Drupal Development||May 20, 2013|
|Making Linux and Android Get Along (It's Not as Hard as It Sounds)||May 16, 2013|
|Drupal Is a Framework: Why Everyone Needs to Understand This||May 15, 2013|
|Home, My Backup Data Center||May 13, 2013|
- Designing Electronics with Linux
- New Products
- Linux Systems Administrator
- Making Linux and Android Get Along (It's Not as Hard as It Sounds)
- Senior Perl Developer
- Technical Support Rep
- UX Designer
- Dynamic DNS—an Object Lesson in Problem Solving
- Using Salt Stack and Vagrant for Drupal Development
- Reply to comment | Linux Journal
5 hours 41 min ago
- Dynamic DNS
6 hours 15 min ago
- Reply to comment | Linux Journal
7 hours 14 min ago
- Reply to comment | Linux Journal
8 hours 4 min ago
- Not free anymore
12 hours 6 min ago
15 hours 53 min ago
- Reply to comment | Linux Journal
16 hours 1 min ago
- Understanding the Linux Kernel
18 hours 15 min ago
20 hours 45 min ago
- Kernel Problem
1 day 6 hours ago
Enter to Win an Adafruit Pi Cobbler Breakout Kit for Raspberry Pi
It's Raspberry Pi month at Linux Journal. Each week in May, Adafruit will be giving away a Pi-related prize to a lucky, randomly drawn LJ reader. Winners will be announced weekly.
Fill out the fields below to enter to win this week's prize-- a Pi Cobbler Breakout Kit for Raspberry Pi.
Congratulations to our winners so far:
- 5-8-13, Pi Starter Pack: Jack Davis
- 5-15-13, Pi Model B 512MB RAM: Patrick Dunn
- 5-21-13, Prototyping Pi Plate Kit: Philip Kirby
- Next winner announced on 5-27-13!
Free Webinar: Hadoop
How to Build an Optimal Hadoop Cluster to Store and Maintain Unlimited Amounts of Data Using Microservers
Realizing the promise of Apache® Hadoop® requires the effective deployment of compute, memory, storage and networking to achieve optimal results. With its flexibility and multitude of options, it is easy to over or under provision the server infrastructure, resulting in poor performance and high TCO. Join us for an in depth, technical discussion with industry experts from leading Hadoop and server companies who will provide insights into the key considerations for designing and deploying an optimal Hadoop cluster.
Some of key questions to be discussed are:
- What is the “typical” Hadoop cluster and what should be installed on the different machine types?
- Why should you consider the typical workload patterns when making your hardware decisions?
- Are all microservers created equal for Hadoop deployments?
- How do I plan for expansion if I require more compute, memory, storage or networking?