Nvidia GPU Technology Conference, Final Day 3
The third and final day of Nvidia’s GPU technology conference was last Friday, October 3, 2009. It was another beautiful day in San Jose and the conference started even earlier, at 8:30 AM. Even though it was early, the keynote address was jam-packed and standing room only as the attendees eagerly awaited Richard Kerris of Lucas Films to give his one hour presentation. It was worth getting up early for as it was spectacular and we will cover it briefly in this Part 3, Day 3 of our short series covering Nvidia’s GPU technology conference.
The first installment of this short series covering Nvidia’s GTC, Day One was published here on Thursday, the morning of the second day of the conference. Day Two coverage was published yesterday here. If you want to check out the jam-packed schedule, or read anything and everything to do with the GTC or the just-released Fermi GPU architecture, Nvidia’s own site has a wealth of information available from the conference.
Nvidia’s three-in-one GPU technology conference used an entire floor at the impressive luxury Fairmont San Jose Hotel in beautiful downtown San Jose, California for 3 days – September 30 through October 2. We are finishing up for you our impressions of this event for day 3. For the final part of this series, you can expect a more polished summary of Fermi architecture when we review all of our notes, transcribe all of our audio and the forty-plus gigabytes of full HD-HR (1080i) video that we shot, and finally make sense of everything that we observed. We will also post our own video and links to full discussions of what we covered at Nvidia’s GTC. In this part, we are also including a few impressions of some of the exhibits. In addition, ABT had two separate semi-private press conferences in the 20th floor penthouse with Nvidia officials that we will cover for you today.
The main things that we carried away from the first two days of Nvidia’s GTC – besides realizing that new architecture and a new much more powerful “Fermi” GPU will be shipping this year – is that this conference is about software, not hardware. We saw incredible breakthroughs using the GPU in high-performance computing that are changing the world. We saw an announcement being made of Nvidia’s commitment to their new “ecosystem” of support for the computing and programming communities for their “Tesla” General Purpose Units that is very much like the incredible support they have been giving for years to developers in the gaming community with their the way it is meant to be played program (twiimtbp).
According to Nvidia’s expanded vision, it’s not just about graphics anymore. They want to announce a new platform in its own right – GPU computing. According to them, the PC platform is largely unchanged for 25 years and parallel computing is now becoming disruptive to the old platform. Nvidia wants to be known as a company that makes general processing chips that also have amazing graphics.
We saw Nvidia announce new support for traditional programming languages like C++ and Fortran that ride on top of CUDA to make it much easier for high performance users to write in their own domain languages. We also saw Nvidia release new GPU/CPU debugging tools and even make Microsoft Visual Studio completely GPU friendly. Of course, they released their new 3billion transistor chip Fermi which promises major performance increases in high performance computing – up to 8 times faster in peak double precision performance! Of course, ECC (error checking, so necessary to high-performance computing) will only be featured in Tesla’s (supercomputing; no output to a display) Fermi chips as there would be a small performance hit for Quadro workstations or especially for GeForce gaming. So, Nvidia is very flexible with their modular-designed new Fermi architecture. Well, today we will look at some of their new tools and a quick glance at Visual Studio. However, first, we want to report on the morning keynote of the third and final day of the GTC.
Opening Keynote, Day 3 – Richard Kerris of Lucas Films
Richard Kerris showed his clips from the “Star Wars” and “Star Trek” as well as from current and even unreleased movies (that we cannot show you) which demonstrated how much progress has been made at Lucasfilm due to the increased power of GPUs. It has changed the very way they do production and has speeded it up greatly.
One of the most impressive and difficult effects is to render fire properly. Without the GPU, it would not be possible to do it in such detail and with such realism (or surrealistic “fantastic” effects). Here is what we meant when we said “Nvidia killed Dumbledore”; their GPUs made this fire rendering possible in “Harry Potter and the Half Blood Prince.”
The entire scene in the fire/water cave except for the actors was purely CGI and incredibly complex. It is only made possible by the GPU. In fact, they pointed out that the parallel processing of GPUs actually mimics how our brains work. It is this programming that is responsible for the continually increasing quality of animation. To that end, Lucasfilm has begun to construct a custom GPU render farm to better harness the power of the GPU. In fact, they brought over fifty people to this conference.
The presentation ended with scenes from several movies including next year’s “The Last Airbender” from director M. Night Shyamalan that we were asked not to photograph. It was a very impressive demonstration of the current state of the art. Now, it’s off in just a couple of minutes to the Peddie/Jensen “Fireside chat”
ECS Summit – Peddie/Jensen
This “fireside chat” with Jon Peddie interviewing Jen-Hsun Huang (Jensen), was billed as “Jensen unplugged”- but I believe they really mean “unleashed”. Jensen likened himself to the Chris Rock of CEOs in that he “actually says what is on his mind”. This interview was billed as unscripted and it consisted mostly of Jensen responding to Peddie’s questions and taking what ever direction he wanted. At the very end, a couple of members of the audience got to ask Jensen questions.
In answer to a Peddie question, Jensen started out by talking about radically different corporate philosophies of different companies in a recessions. The “safe” way is to just stay alive but he says he never picks that way nor does he panic. He told the emerging companies to first ask if what they are attempting to do is even possible – if you can afford to do what you are doing. Then he said an emerging company in a recession needs to ask themselves if they really believe in what they are doing. During this, I thought about our own emerging company and our own vision and silently answered ‘yes’ for all of us. He said, “believe in what you believe in and keep going”. Indeed.
Jensen acknowledged that Nvidia “delivered badly last year”. He said it was “OK because we did the right thing for the right reasons.” He said he is not going to slow down but he did examine Nvidia’s priorities regarding parallel computing; they are not going to stop investing and he renewed his commitment to creating and improving developer tools for it.
One of the things Jensen stressed was his disdain for the extreme emphases placed on “financial performance” which he likened to getting grades in school; except now you were being graded by your employees, customers and shareholders. He said, it is far more important to do what you do “for the right reason” and to make the experience worthwhile; one should not go to school just to get good grades. Rather, good grades come from the hard work and effort that you put into learning; if you do it right, you get good results. Jensen reminded his listeners, “Keep the game separate from the score’”. It is very good advice and it was delivered in the most enthusiastic manner possible. You can sum up much of what he says by this: “If you do amazing things, the world notices”. Well, I have to agree with him again as that is my own belief.
The discussion continued after Peddie commented about the low attrition rate at Nvidia compared with other companies. Jensen said he wants to keep one, so that only the people who really fit in well, will work there; in so many words, they both should be happy. It is clearly important for Jensen that his employees believe in their company. He then emphasized how this conference was all about software and about Nvidia providing tools for a currently very tiny market because it is the right thing to do at the right time.
He talked about the few acquisitions that Nvidia has made each year and how important it is for the companies to be matched to each other, as in a marriage. Jensen said he deplored acquiring technology for the sole purpose of keeping it from the market and from the competition. He mentioned he was proud of Nvidia’s promoting the little startup Keyhole Corporation that became google earth. Nvidia is always looking for a worthy company to promote with no limit to their budget for it; depending on need, up to the company’s entire 1.4 Billion dollars if necessary, he added for emphasis.
Jensen said that one should see the world with optimism and to “do good”. He told the emerging companies, “if you don’t build it, they cannot come”. So there will always be some uncertainty which means there are risks to be taken. However, he said it is obtuse to think of “taking” something at the expense of others. Rather one should take advantage of what is already there to do bigger and better things. However, I am not sure how he relates it to gaining marketshare.
As if to answer him, Peddie asked Jensen about the current “squeeze play” by Intel on Nvidia and what they intended to do about it. Jensen answered, “by building amazing things”. He then went on to list the criteria Nvidia uses to determine if they should help an emerging company: 1) it has never been done before, 2) it is incredibly difficult to do, and 3) that company is uniquely qualified to do it – and I thought of our own vision as I mentally checked off ‘yes’ to each one of his requirements.
Peddie then touched on Nvidia’s Fellowship program which provides support for people in the form of scholarships. He did not mention Jensen’s own personal contribution of $30 million dollars last month to build an expanded engineering school at Stanford. The Jen-Hsun Huang School of Engineering Center (as in the rendering below) will feature a four-story building housing ultimately a “bookless” library, conference center and a café.
What makes Nvidia unique is their enthusiasm; their money is not unique – their people are, according to Jensen. And again, he picked the words “exquisite” and “amazing” to describe the level of work and the products he expected his company to produce in striving for perfection. When asked about the next ten years, he responded that “the computing industry will be totally different. Hardware will be less magical”. He said he pictured the day when a “personal computer will come free with a service” and the ones that you buy will be truly amazing.
Jensen went on to say that when you increase performance by 1,000 times, everything changes. He also likened these times to 1995 – a metaphysical 15 year cycle of innovation – when Win95 was released and everything since then has been a refinement of it. He says PC hardware has made great progress but now it is time to advance software. Jensen said the key to future hardware is “efficiency” and that “more of everything is a bad idea”. In other words, he believes hardware power requirements will not go up but remain flat due to innovation. Jensen finally concluded his fireside chat with taking some questions from the audience and he urged people to be enthusiast and optimistic.
After that was a panel discussion dealing with raising capital in difficult financial times but this editor was off to the exhibits and then to learn about CUDA and Nvidia’s new debugging tools as well as some hands on experience with Nexus now integrated with Visual Studio. It was lunch time also.
New Tools for CUDA including Nexus within Visual Studio
The GTC featured one entire segment of the conference devoted to programming and developing for the GPU. This editor has not touched any programming whatsoever since the early 1980s and was amazed to see the new tools. CUDA is not going away. It is Nvidia’s proprietary GPU programming language that is the equivalent of x86 instructions for the CPU, but for the GPU. However, most researchers do not want to learn a new language and would much prefer to program and debug in their own domain language. Nvidia has released tools to do this with full support for Fortran, C++, C, OpenCL, DirectCompute, Java, and Python, and to top it off, Nexus, the world’s first fully integrated computing application development environment within Microsoft Studio.
This editor attended a couple of programming hand-on demonstrations. Here is a summary chart of the tools that are currently available and what is coming soon.
We then saw a demonstration of how to actually do some debugging.
These tools are intuitive and are just what is needed, and the instructor went step-by step on how to effectively do this.
We also got a preview of what’s coming:
Nexus is an amazing tool that has just debuted that allows the programmer to look at and debug the GPU in the same manner as the CPU. These are tools that they are already familiar with but now targeted for the GPU. The will increase the efficiency of the people who work with GPU programming.
Again, everything was laid out step-by-step although I was way too slow to follow along properly.
Finally, off to the Virtual Studio Lab where I got completely lost attempting to follow along. However, I do realize that the premier application that Microsoft has created for its purpose is now completely GPU-friendly with Nexus. Again, Nvidia targets the inceased productivity of the programmers by giving them access to tools that will allow them to produce better applications much the same as they work with developers in their the Way it is meant to be Played Program. I expect they will get the same results with this new program and they will grow a community that comes to depend on Nvidia to make their work more productive. It is brilliant planning ahead that is applied and adapted for a new industry. We can see that Nvidia intends to be the industry leader in the GPU high performance computing future as they are pioneers now.
However, I did get hands on with two 30 inch monitors and realized what an important innovation Nvidia has made that will be appreciated by developers and programmers the world over. They will do better work and create better applications because Nvidia provides assistance for them.
We see an impressive commitment by Nvidia to their ecosystem of developers and programmers with these great tools. As a practical result, in the months to come we can expect to see better and better applications come out of this. This is industry shaping and transformation in action, at its near-beginning.
Meeting with Nvidia Officials
ABT was fortunate to have over an hour to sit down with Nvidia’s Drew Henry, the general manager of the GeForce business unit and Jason Paul, GeForce product manager. These men are sincerely enthusiastic and attentive – and you have to consider that they have been doing these same exact kinds of interviews with the press basically non-stop for 2 days! This is very rare to find in most other companies and yet we encountered it over-and-over. Clearly these people do what they love and love what they do and it is noted as being one of the keys to personal success, never mind for success as a company.
The second meeting was with Andy Keane, general manager of the GPU computing business unit and with Andrew Humber, Tesla & CUDA Enterprise Products. Before the meeting we chatted about many things including moving to a new house – which one of them had to do over the weekend, something I hate doing. Each of these meetings went over whitepapers and answered questions that we had. Mostly it was clarifying what was said in the conference and emphasizing what Nvidia wanted us to take away from the conference; that they were committed to this new direction in parallel computing, while not forgetting the rest of their ecosystem and customers.
At this meeting was clarified the naming of the new architecture and we got to look at the new Fermi GPU. They said they were announcing Fermi early to prepare the industry for it. We could not help but think to ourselves that the timeliness of their announcement would also prepare their fans to ignore the competition’s latest DX11 offerings and to wait for the Fermi GeForce GTX.
They insisted that Fermi would also be a formidable gaming GPU and it would be the single fastest GPU when it is shipped this year . They said that Fermi’s 384-bit bus is very well-balanced with GDDR5’s increased bandwidth, although lesser models are set up to also use GDDR3. They pointed out their philosophical differences with ATi and said that within 18 months everyone would see the clear divergence. ATi is mostly committed to graphics chips while Nvidia wants to elevate the GPU to a true co-processor with the CPU. They also pointed out that PhysX was making the biggest difference in the gaming experience along with 3D Vision.
They pointed out that Ion was not an Atom exclusive and that Nvidia would build it for any CPU. However, they were not going to build DMI for Intel nor were they going to build new motherboards for AMD as their high-end CPU business had mostly collapsed; so, no new core logic for AMD in the near future is the prognosis.
They also talked about their licensing of SLi to motherboard manufacturers and how it “just works” because they have one of the largest compatibility testing labs in the world where they test all conceivable hardware combinations. They are proud of their “team testing” also, where every game setting is tested individually . They also said to look forward to a much improved hybrid SLi for extreme power savings in notebooks.
We also discussed the improvements in the Fermi architecture over Tesla’s besides the obvious doubling of the cores to 512. They pointed out the new GigaThread technology that allows for each solver to be a kernel to execute instructions is much faster than waiting in a queue for the previous instructions to complete in the old serial fashion. Thus there is greater efficiency when all 512 cores are used instead of having idle waiting time.
They were very proud of their new debugging tools which would allow a much higher quality of application to be created more quickly and we met CUDA’s creator, Ian Buck, in our second session. We asked if the philosophical differences between their chief scientists have made any changes yet. Bill Dally brings extraordinary parallel programming experience to Nvidia but Fermi was started about four years ago; long before he came to Nvidia.
The Exhibits:
Next we saw over 60 exhibits in one large room. Nvidia regrets that they were not able to allow more companies to present and promises that next year, there will be more room. Of course, there are many exhibits we did not cover, but here are a few interesting highlights of what is available and what is coming.
Here is Scalable Technologies:
The conference must have been very successful for Scalable Display Technologies, a leading provider of software for multi-projector display systems. Today, October 6, they announced record sales for the third quarter, driven by increasing demand for high-resolution displays in a variety of markets.
Scalable Display’s patented software, EasyBlend™, simplifies and automates the creation of seamless, ultra high-resolution displays. EasyBlend enables users to automatically align, edge-blend and color-match multi-projector displays on screens of any size or geometry, with just a few clicks of a mouse.
Three leading integrators placed volume orders for Scalable Display’s EasyBlend during the quarter, including Immersive Display Solutions (IDSI), a provider of turnkey immersive simulators for military and commercial applications; BOI Solutions, a provider of visualization and meeting collaboration services and products; and VDC Display Systems, which specializes in the design and manufacture of projector systems for the simulation and training community. Video Display Corporation’s (NASDAQ:VIDE order represented the largest channel order for the military market in Scalable Display’s history.
“Our dealer network and our sales continue to grow as the demand for high-resolution, multi-projector displays increases,” said Andrew Jamison, CEO of Scalable Display Technologies. “It is clear that the ease of use, quality and reliability of our technology has made it the software of choice for display system providers and integrators.”
We visited Prometech Software that specializes in high performance simulation and high end graphics. Tekken 6 (arcade version) is using their new Octave Engine which allows sand, water and the sky to be combined into one simulation; for example, the sand can be raised and lowered and interact with the water and other objects in real-time.”
Then on to DuxSoft.
And Nova
And of course we visited Visioglobe who released a full 3D representation of Paris that can be used in handheld mobile devices.
Naturally, everyone needs a touch-screen LCD when Windows7 is released. Loilo Touch releases with Win7 this month and features video editing so simple that children can use it.
The GPU servers are impressive. Here are a couple of Tesla servers from SuperMicro:
Check out this Tesla server from Colfax:
How about this interactive computer vision 3D Vision game demo from softkinetic? Jensen shows us his moves as he attempts to fit into the shapes as they flow on the screen – to dance music – on this 3D display:
It looks like he needs a little practice for next year’s event. I hope to see you and him there for GTC 2010.
>
The Summary
This was an amazing experience for this editor to witness the follow-up to a computing revolution that began just last year at Nvision08 and one that has continued unabated with the 2009 GTC. Considering that this is one of the worst recessions in memory, it is an amazing thing that Nvidia is doing with their GPU. It is not longer simply a Graphics Processing Unit, but they have refashioned it into a General Processing Unit – a co-processor with the CPU. To do this, they realized that they need to invest major company resources into not only next generation, brand new-architecture, Fermi; but into the tools and the complete support “ecosystem” for the programmers and developers that work with high-performance computing.
What is most impressive is that Nvidia has been planning this for more than 4 years – for that is how long it has taken to go from design to production for Fermi! Nvidia has created a 3 billion transistor powerhouse that should also be the fastest graphics card when it is released later this year as the GeForce GTX. But Nvidia no longer wants to be known as a company that makes Graphics cards. They want to be known as a company that makes great processors that have amazing graphics. Well, we are indeed looking forward to the Fermi GF100 GTX but realize the importance of what Nvidia has accomplished.
We saw hundreds of enthusiastic attendees – something that cannot be faked – really appreciate the new technology that is coming and what was released here. It is exactly what researchers and scientists need for continuing their important work in almost every field where parallel computing is useful. We saw grateful programmers and developers paying rapt attention to CUDA demonstrations and to new tools that will make their life easier as they turn out better applications.
Conclusion
This editor is very impressed as he is certain that he has glimpsed the future and is now able to see clearly where ABT fits in with our own vision of changing the way humans interact on the Internet. We will need the GPU’s computing power and cloud computing to make what we are attempting real. And we expect to use the services of other companies and programming tools that are being made available by Nvidia to make our 21st century dream solid of us interacting in real-time and in all languages in a 3D forum world.
We expect to be posting more information about Nvidia’s GTC 2009. We still have 40 GB of High definition, high resolution (1080p) video to work on, edit and upload and much more to say about the Fermi architecture as the launch of the new GTX gets closer. Make sure you join us on our ABT forums to discuss anything on your mind.
Nvidia gets a 9/10 for this conference; a solid “A” for what it was and is becoming and I am looking forward to GTC 2010! As a plea to them, make next year’s conference schedule less hectic and definitely make it longer. Kudos for not dumping us into San Jose rush hour traffic at 3 PM, as last year. This editor sees the GPU computing revolution as real and we welcome it!
Mark Poppin
ABT Senior Editor
Please join us in our Forums
Become a Fan on Facebook
Follow us on Twitter
For the latest updates from ABT, please join our RSS News Feed
Join our Distributed Computing teams
- Folding@Home – Team AlienBabelTech – 164304
- SETI@Home – Team AlienBabelTech – 138705
- World Community Grid – Team AlienBabelTech
