Introduction

About a month ago Matrox announced the G550, their latest chip that boasted some very interesting features, especially the new HeadCasting technology. We promptly did a preview of the new chip and technology. However, the new G550 chip does not focus on raw 3D performance, and NVIDIA's GeForce3 is still the current choice for the 3D gaming enthusiast.

With the company seemingly turning away from the high performance 3D gaming market and focusing their efforts on other areas such as 2D performance and DualHead display, many people had questions and concerns. We decided to gather some of these questions and get an interview together to see if we could get some more insight into the inner workings of Matrox.

We got a chance to speak with Kamran Ahmed, senior manager of strategic marketing and product management for Matrox Graphics. We would like to thank Kamran for putting up with our questions, as well as Patricia Jreige, public relations specialist for Matrox Graphics, for helping setup the interview. Enough chit chat, on to the interview shall we?

The Interview

Thank you for accepting our interview and taking the time to answer our questions. My first question concerns the new HeadCasting technology. Matrox has always been pushing new and innovative technologies such as EMBM and the new HeadCasting technology. When and how did the new HeadCasting technology come into the picture. Why did Matrox choose this technology instead of something else?

Well, Matrox has always believed in the need to introduce features that appeal to a large segment of the market. Most graphics chip companies are focused only on developing technologies that push the envelope to make games run faster. In this quest, some companies even compromise on basics such as improving 2D image quality and other features that are important to non-gamers. While Matrox understands the need to provide better hardware for gamers, it also believes in introducing features that provide new benefits to every user's overall graphics experience. In the case of HeadCasting, we recognized the trend of online communication that has taken the industry by storm and developed a way to add a visual aspect to this otherwise text-based experience. We also identified that using good quality two-way video to communicate is far from reality due to the bandwidth limitations today, even over average broadband connections. HeadCasting was designed to use the latest in 3D technology to provide a compelling online communication experience to the masses.

With the Matrox G550, the target was not the high-end enthusiast/gaming market unlike with the Matrox G400 and Matrox G200, when they were first launched. The goal with this chip was to add the visual online communication capabilities to the best corporate and 2D workstation chip available today at a low incremental cost.

Matrox was the first company to push EMBM, and succeeded in getting support in a large number of games, even at a time when other cards weren't utilizing the technology. Now Matrox has HeadCasting technologies. In what way do you expect to see Matroxs' innovations impact the industry, despite producing chips that aren't intended to be competitive in the gaming market?

EMBM had a real impact on the gamer's visual experience and we were able to do a pretty good job at convincing developers to support the feature. To date, I believe there are over 40 different titles shipping with EMBM support - HMMmmm, I wonder if that's even more than the number of T&L titles shipping today?

The technologies that Matrox brings to the market have a pretty significant impact in the industry - mainly due to the fact that these technologies benefit a wide variety of users for whom these features become extremely useful in their everyday user experience. With the first Millennium, back in 1995, Matrox was the first company to introduce a 64-bit graphics chip that would bring high-resolution desktops to the new Window 95 user interface. Additionally, things like being able to switch resolutions and color depths on the fly, contributed to a much better user experience for everyone.

Matrox was also the first company to focus on improving image quality. With the Matrox G200, we introduced VCQ Rendering with 32 bit internal rendering and UltraSharp DAC technology, bringing the importance of having high 2D and 3D image quality to the attention of an industry, which was otherwise focused mainly on performance.

DualHead Display technology was the next big innovation. Another first from Matrox, it found its way intro all types of application areas and has once again improved the way every user interacts visually with their desktop. The impact of this technology on users can best be judged by seeing some of the customer testimonials at . And the impact of this technology on the industry can best be judged by the fact that competitor's feel compelled to imitate it.

All in all, I believe that Matrox is the only graphics company that has consistently innovated features that truly enhance almost every user's visual interaction with their computer. Indeed, there should be graphics chips that push the envelope on 3D technology and provide the ultimate gaming experience for the gaming enthusiast, just like there are graphics cards that are designed to provide the most high-end solution for 3D workstation users - but, we shouldn't forget about innovating for the masses.

The concept of 3D communication is appealing but with the increased availability of webcams and broadband connections, how would the use of the HeadCasting engine be a better alternative? In other words, why would you use this technology instead of just a regular webcam?

First of all, I have to point out, that 3D communication has more facets than the two applications we have chosen to introduce this concept with. Application areas like Visual CRM (Customer Relationship Management) and Virtual Web Hosts, employ virtual 3D models to communicate with people. Web cams and video certainly does not address these areas. Virtual Presentations and training with the Matrox Virtual Presenter application also cannot be replaced with a web cam. As for "3D video conferencing " with HeadFone, which is what I guess you were specifically referring to, there are a lot of reasons as to why communicating using this 3D method would be a better alternative than video conferencing using web-cams. Bandwidth is the main issue. Remembering that this technology is for the masses, who are still stuck with 56 K connections (especially in Europe), the quality of video when video conferencing is unacceptable. Yahoo's recently announced Messenger Service with video conferencing boast a frame rate of ONE frame per second - imagine if a gamer heard that ;)

A lot of people think of broadband as being sufficient enough to transmit high quality video over the web. Sitting at work using a T1 connection, I usually have trouble streaming video over the web - and I'm talking about low to medium quality video.. Moreover, this is a one-way transmission only. Video conferencing is real-time two-way transmission of video and can therefore be unacceptable even over broadband. Some of these bandwidth issues can also be attributed to server traffic and reliability of service, which do not get overcome by broadband access.

Outside of the bandwidth issue there are other less major benefits such as not requiring a web-cam, not having the other person see where you are or what you're really doing etc. etc.

Although the HeadCasting technology is primarily targeted towards 3D communications, are other applications possible or in the works? Such as in online games where you would be able to customize your character with your face? Does Matrox plan to expand this technology to other applications?

The technology itself is fully DX compliant and therefore can be used in any 3D graphics environment. With the Matrox G550 product, our focus will be on applications that enhance online communication. Since this product is not heavily targeted towards gamers, we do not plan to actively pursue application areas related to gaming.

With the graphics assets of 3dfx sold to NVIDIA, ATI and Matrox were left as the sole producers of both mainstream 3D graphics chips, and the graphics boards supporting these chips. ATI publicly held steadfast to this business model up until just about the point at which they publicly announced their change. What impact, if any, do you see this playing in the graphics market, and do you believe that this model is losing its advantages as time goes on?

Using 3rd party board manufacturer is more suitable when providing commodity graphics at the lowest price possible. For the mainstream to high-end market, building our own boards does give us a competitive edge - especially in terms of quality control and reliability. Its becoming harder for end users and even SIs and OEMs to rely on board manufacturer's that are here today, and then gone tomorrow. Remember Creative and Diamond.? I think for end users, its important for them to know which board their desired graphics chip is coming on. I'm sure if you bought a board based on a TNT2 chip from different board manufacturers, you'll find varying degrees in output quality, driver quality and sometimes even feature support.

With the economy as bad as it is now and many companies announcing losses in the last two quarters, how has this affected Matrox's over-all business model? How is Matrox adapting to the current state of affairs?

Matrox as a company, has been around for over 25 years. Facing and successfully surviving these kind of downturns in the industry is not new for the company. Apart from the regular cautiousness that needs to be exercised at times like this, its business as usual here at Matrox.

The video card market is constantly getting more competitive with new products from companies such as NVIDIA and others constantly raising the bar in mainstream performance and features. What is Matrox planning to do so it does not get left behind in the next year or two?

Matrox's main reason for success is the fact that we focus on providing products that address the needs of the customers we target. Our development cycle and roadmap are not based on playing catch-up with competitors but more on introducing the right mix of technologies for the right audience/customers at the right time. While competitors may be raising the bar on gaming performance and features, Matrox is raising the bar on other things such as basic desktop image quality, DualHead and DualHead features such as true Multi-Display under Windows 2000 and single-chip Dual-DVI support.

NVIDIAs aggressive 6-month product cycle has made it very difficult for competitors. Was the introduction of this fast-paced production detrimental to Matroxs' ability to remain in the gaming arena?

The formula for success is more than just having a six-month product cycle. The thing that really matters to a gaming enthusiast is which product is the most competitive and best suited to their needs, at time of purchase. While Nvidia's 6-month product cycle has certainly helped them refresh their products in order to stay on top in the enthusiast gaming market, any competitor with a one or even two year product cycles could easily spec their feature-set and performance levels to compete with other products shipping in that time frame. A good example is Kyro. It came out of no where, does not even support DX8 features but has the right price point and competitive performance in the games that matter - and the result is that a lot of gamers are choosing this product over other products enthusiast products.

With Matrox's strong 2D graphics and DualHead technology, 3D performance is all that is missing. Is Matrox still planning to re-enter the 3d performance/enthusiast market at some point?

Matrox has been, and continues to be, a key contributor for advanced graphics specification, be it AGP 8X or DirectX. There are a lot of different factors including market conditions that will determine when (and if) we target the enthusiast gaming market..

Bandwidth has been a serious issue with gaming performance. Matroxs' decision to utilize a 64-bit DDR bus instead of a 128bit DDR bus certainly lowers costs. Bandwidth has been a major bottleneck in increased 3D performance for a few years now. Despite the fact that Matrox has chosen to orient their products towards the home-internet/corporate user, at some point, 3d performance will become more and more important (for example, there are 3d web-browsers in development). What technologies are Matrox considering to deal with this issue?

While memory bandwidth is an issue for enthusiast gaming, there is really no big technical hurdle that needs to be overcome in order to increase the amount of memory bandwidth - if a chip manufacturer chooses to. Memory speeds will continue to increase. 300 MHz DDR could be common next year and then there is DDR-2. Of course, you could add a 512-bit memory interface if you really really wanted to ;)

For other kinds of applications, such as 3D over the internet, the bandwidth issues. internet bandwidth issues that is. are harder to resolve. This is why we dedicated resources in developing HeadCasting. Its effectively a high quality compression algorithm for 3D graphics. Using audio and animation data that is low bandwidth is well suited to the internet.

Matrox made use of the PowerVR PCX2 graphics controller in the m3D. What prompted Matrox to use a chipset developed outside of the company, and what prompted Matrox not to use them anymore?

With the m3D, Matrox's goal was to provide a high performance alternative to the only other 3D upgrade available at that time - the original Voodoo. The 2D/3D graphics cards in those days could not rival the performance ofstandalone 3D cards. The PCX2 controller was best suited to fill this void. What's more, we were able to introduce it at a $99 price-point that was unheard of before. After the m3D, we entered a new era of 2D/3D cards such as the Matrox G200, which offered enthusiast level gaming performance integrated into a full 2D/3D/Video graphics core.

Does Matrox intend to enter the mobile market with 3d capabilities? Is the g550 intended for mobile purposes as well (most businesses have high speed access, so HeadCasting would be good for mobile PC with 56K modems)?

At present, Matrox does not plan to play in the mobile space.

Since Matrox supports AMD's new AMD-760T MP chipset and AMD AthlonT MP processor with Smart MP Technology, how does Matrox plan to realize the full potential of the system? Will there be any new products that specifically take advantage of the multiprocessor architecture?

The specific advantages that can be had with a multi-processor architecture rely more on software development than hardware features. Software enhancements such as multi-threading specifically for MP can benefit significantly from the MP architecture.

Matrox has supported Linux for a while now. What are Matroxs' plans with respect to alternative OS support? Does Matrox plan to provide better support for Linux or other alternative operating systems in the G550 and future products?

Matrox has been supporting 3rd parties for alternative OS driver development for a long time. Additionally, Matrox has by far the most comprehensive feature support under Linux. As you can see from the recently announced PowerDesk support under Linux, Matrox is constantly working on providing better support for Linux and will continue to do so in the future.

Despite support for Linux, the Marvel G450 eTV has no updated drivers for any operating system. Why is this?

The Marvel G450 eTV started shipping in Late Q1 of this year. With this kind of a product, which has a plethora of features, the driver updates typically take longer due to the longer development cycle and an even longer QA cycle. It is usually better to release less updates that are robust with a lot of bug fixes than a lot of frequent updates that resolve less issues and have not been QA'd extensively. Its important to note that the QA time for each release is the same irrespective of the number of issues that the release resolves.

What features do you feel the corporate market finds to be in demand? What about the future?

The corporate market and the 2D workstation market have a long list of requirements. These include high image quality on the desktop, especially at higher resolutions, multi-display support, DVI support, extensive OS support, driver stability, comprehensive resolution/color depth and refresh rates and 2D performance. Matrox has consistently focused on delivering products that have exceeded these requirements making our products the graphics cards of choice for corporate systems by the top 10 OEMs and system integrators. As for the future, there is room for lots of innovation even when it comes to the corporate desktop and you can be sure that Matrox as always will play the leading role in introducing these innovations to the market.

What is Matroxs' position with respect to the high end CAD market?

While Matrox's current product lines provide the best available solution for 2D CAD, they do not target the 3D CAD market. This market requires products that are specifically enhanced for accelerating 3D CAD applications. The market is a lucrative market to play in but we are not targeting it with current products.

Matrox has always been a forerunner with 2d graphics performance; the Ramdac speed plays an important role in this. Matrox has been using a Ramdac of up to 360Mhz on consumer cards for over two years now -- is Matrox considering increasing the Ramdac frequency?

The 360MHz DAC of the Matrox G450 and Matrox G550 is the highest speed DAC available on any card today. It also is the speed at which you can support a refresh rate of 85Hz at 2048x1536@32bbp resolutions - the highest available on monitors shipping today. Increasing the speed of the DAC will not provide any significant benefit for the monitors available today. This may change as monitor manufacturers push the envelop on the capabilities of their CRTs.

It seems 2D graphics performance can only go so far, and we're close to reaching that plateau. What areas are left for improvement in 2d graphics?

With the introduction of multi-display from Matrox, 2D performance has taken a whole new meaning. This is obviously hard to measure since the current benchmarks are not multi-monitor aware. Since doubling the size of the display requires the 2D engine to be faster, 2D performance is still a factor.

With all the new technologies and features that will be introduced in the next year or so, what is the "next big thing" in PC consumer graphics?

I think "3D for the masses" is the next big thing for PC consumer graphics. Improvement in 3D for gaming will continue but that does not dramatically affect the majority of PC users. What is needed is a new genre of 3D applications that bring the benefits of 3D to the user's everyday computing experience. Matrox has taken the lead on this with the HeadCasting technology and applications. But this is only the beginning - the possibilities are endless and are only limited by the imagination of the developer.

We would just like to take a final opportunity to thank both Patricia and Kamran for working with us on this interview. Make sure to check out our G550 preview for more information on the card. If you have any comments or more questions, please post them in this dedicated thread.

Re-Printed From SLCentral