There’s no question that with the Nvidia GeForce 9400M Intel finally has a serious competitor on the integrated graphics front. Appearing first in the Apple MacBook, the 9400M claims up to 5X the performance of Intel’s X4500HD integrated graphics, which is part of the Centrino 2 platform. In our tests we’ve seen upwards of a 2.5X increase in performance, which is still impressive. More important, the 9400M is quite efficient. In our tests the MacBook lasted nearly 5 hours on our battery life test. This is what Intel had to say about missing out on the new MacBook: “We believe our products offer significant performance and value to our customer and we will work hard to win back the business.” The problem is that Nvidia is working equally hard to win more market share away from Intel. Yesterday Nick Stam, Director, Technical Marketing, told us that:
But that’s not all Stam had to say. We talked about the mainstream appeal of better integrated grahpics, whether the 9400M could migrate to netbooks, the fallout from the recent defects, and more. Editors’ Note: Portions of this interview have been amended or corrected due to factual errors. Will consumers who don’t play games or use Photoshop care about getting a 5X boost in graphics performance? Do they use Google Earth? Which, of course, is a cool new app; people love to surf and look around the world and fly into cities. Google Earth chokes on integrated graphics but does just fine on this solution, so that type of application we see from some people. Microsoft Live Maps and Virtual Earth 3D is another great usage for these graphics. Then there is Second Life and virtual world type of things. But a typical person who is using the laptop to surf the Web, check e-mail and do that type of thing won’t see a big benefit. But those also may be the same people who are encoding a bit of video and going into Piscasa to zoom and move around in your images. For those later tasks they will see those types of reactions and processes respond quicker. Is an important next step for Nvidia to engage with the Web 2.0 community to optimize their Web tools for graphics? Today you can have a lot more Web 2.0 functions and that is exactly where the world is going; much more graphical and interactive sorts of things in the browser and at Web sites. We see the world going this way and we are working with a number of vendors on GPU enablement so its part of our CUDA (Compute Unified Device Architecture) plan, and there will be a lot of video enhancement technology. For instance, with a YouTube video even if you hit the High Quality button but they are still not great but we have a number of technologies that will look ahead a number of frames and behind a number of frames to combine to make the frame you are viewing much clearer. This is to create the image in a much clearer way, much like upsampling DVD players. So the 9400M could enhance Hulu.com and other video sites? You should see an improvement in fluidity of play. It depends on what file formats they are using. It is unclear today if it would make much of a difference, but when you have more GPU-accelerated video decode and enhancement functions implemented in software-based video players it can make the image look that much better. Is there an ETA for enabling this kind of performance boost? We are working with 7 or 8 companies that are making video enhancement technology. There are a bunch of people that are coming out with stuff that integrates with GPUs. Apple’s Open CL is very similar to NVIDIA’s CUDA technology – both taking advantage of the GPU to accelerate applications. CUDA is a general, cross-platform C environment, while OpenCL is more aligned to Apple and OS X. The OpenCL API uses portions of the CUDA driver stack, so OpenCL and CUDA code ultimately look the same to the GPU hardware. It is one of the technologies Apple is excited about with Snow Leopard. Open CL can be thought about in comparable terms with CUDA. What percentage of the integrated graphics space do you capture? I personally think that we can be displacing 30 percent of the existing integrated market. You aren’t going to capture it all and Intel is out there in a big way, but if we can get a number of the big vendors that could help our growth on the sales side. We know that five of the major notebook vendors are planning to ship systems with [the 9400M] chip. Apple has taken some credit for how the 9400M was implemented in its new MacBooks. How instrumental were they in leveraging this technology? They worked very closely with us to be the first out on the block with the chip and integrated with their system architecture, and we worked with them on the thermals. It has been a long collaboration with them on the MacBooks. We really worked collaboratively for a long time. Do you feel that given the intense collaboration with Apple that their implementation might be better than what might be coming from other partners? We really can’t say that. There are other capable partners out there who can build great and nice designs. We know Apple is known for their great designs and they do build an amazing product. What I like is that they do know how to do that. On notebooks with switchable graphics from Intel you don’t have to log out and log back in order to switch modes. Why is that the case with Nvidia’s graphics on the Apple MacBook Pro? The hardware is capable of switching between integrated and discrete on the fly and it’s available today on the Window platform. It is not there today with Apple, but in some of our upcoming designs and in our hybrid SLI designs you can have both chips work together. One may not be the same power as the other but we can manage the SLI so it gives you a net boost, what we call a GeForce boost. That isn’t here now but that may be coming in the future. How have you been dealing with chip flaw issues and voluntary recall? There are different manufacturers, be it Dell or HP or Apple, that were all exposed. But it has to be a bad thermal environment and lots of power cycling. A lot of different scenarios have to come together to cause the problem. We set a lot of money aside, but we really haven’t had to service a lot of that. All of a sudden when you say something like that the whole world blows up and thinks it’s every notebook. Do you think the 9400M could scale down to netbooks? With what we have just done with the MacBook Air and making the reduced factor is an example of smaller forms. I am not exactly sure of their thermals and power requirements, but we can absolutely scale these down in terms of clocks and we have a reduced function version. These are 65 nanometer devices and you will see them going down to 55, 45, or 40 over time. Has there been any interest from netbook makers? Yes there has been interest, but I cannot tell you who or what. How does your Tegra platfrom tie into this? Tegra is very low power and you can see high-end Tegra in small netbooks, depending on how big the netbooks are. This is kind of where they cross over. Tegra is going to be very big next year. We have high hopes for it.