|The Leading Source for Global News and Information Covering the Ecosystem of High Productivity Computing / May 11, 2007|
Dr. Burton J. Smith, Microsoft Technical Fellow, former chief scientist at Cray/Tera Computer, and a world-renowned expert in high performance computer architectures and programming languages for parallel computers, will present the opening keynote address at the 2007 International Supercomputing Conference (ISC'07) on June 27. His talk, "Reinventing Computing," will look at the ubiquitous deployment of parallel processors and what it could mean for HPC and the larger computing and IT industries. Recently, Smith took some time to answer questions related to his upcoming talk and the computing world of the future.
To hear Smith's keynote in person, register for the ISC'07 at www.isc07.org/registration. The conference, conducted in English, will be held June 26-29 in Dresden, Germany, and advance registration is open through May 26.
Q: The abstract to your upcoming talk at the International Supercomputing Conference says that we are embarking on a new computing age -- the age of massive parallelism. Can you give us a more detailed vision of this new world and what it will mean for those in computing fields and those who rely on computing?
Smith: Moore's Law will continue to improve transistor cost and speed, but single-processor performance will no longer keep pace. There are two possible future scenarios: either computers get a lot cheaper but not much faster, or we use parallel computing to sustain continued performance improvement. In the first case, computing becomes a "mature" industry, and hardware and software become commodities. In the second, consumers will continue to enjoy the benefits of performance improvements, but successful software and hardware providers will have to embrace parallelism to differentiate themselves and compete.
Q: How ubiquitous will multicore processing become? Will everyone have multiple parallel computers at their disposal every day?
Smith: Yes. Even mobile devices will exploit multicore processors, not only for better performance but also to extend battery life by replacing the relatively power-hungry serial processors used today.
Q: As parallel processing devices become the norm in everything from smart phones to GPS systems to laptops, how can (or must) the HPC community take advantage of this new environment?
Smith: The HPC community has been successfully using mainstream hardware for quite a while, but mainstream software has not been adapted for HPC in the same way. However, as parallelism becomes increasingly ubiquitous, HPC should be able to exploit and extend mainstream programming languages, operating systems, development tools, libraries, and even applications intended for smaller scale systems. Also, as more developers successfully exploit parallel computing, the available HPC talent pool will expand.
Q: What should be the plan for vendors in the HPC arena to take advantage of this reconfigured world? Does Microsoft have a strategy for competing in this arena?
Smith: Microsoft has the advantage of being a software supplier to both HPC and the mainstream, which is obviously helpful to us. Other HPC vendors' strategies will be up to them.
Q: What will the workstation of tomorrow look like? Will heterogeneous, multicore processing be the norm? Will laptop and desktop machines be overtaken by parallel processing PDAs?
Smith: Tomorrow's workstation will be a PC equipped with a "system on a chip" that contains CPUs and GPUs of several types, intelligent I/O controllers, and every other device that might be useful and reasonably popular in such a system. Smartphones will assume roles as PDAs, MP3 players, etc. as processing power per watt improves. But smartphones will not replace the laptops and desktops with their full-sized screens and other amenities. Desktops with their more weakly constrained power budgets will become the "personal supercomputers" of the mainstream.
Q: What will be the parallel programming model of the future? For example, are there prospects for global address space (GAS) languages?
Smith: Whatever the mainstream programming model of the future is, it must enable free interoperability of shared memory and message passing. Shared memory is needed to enable high-bandwidth fine-grain communication and automatic load balancing, but the increasing importance of Web-based services demands frequent message-based communication between clients and servers via the Internet. HPC has had limited success in mixing the two paradigms; we will have to fix that. As for GAS languages, I doubt they will be important for PCs any time soon, but they could be useful for some mainstream server applications. We'll see.
Q: Do you foresee a new language, a new paradigm for computing? Where will we get the people to work with these machines? Who will write the software, develop optimization strategies, etc? Do we need to undertake a major retraining effort? Refocus our undergraduate and graduate curricula?
Smith: Our field must reinvent not only computing but also the computing profession. The universities will do much of the work, but companies like Microsoft will have to help, for example by educating the developer community about this new way of thinking. We will also make it much easier for developers to write parallel software.
Q: What are the consequences for HPC and the larger computing world if we can't change our standard mode of operation to meet the challenge of manycore computing?
Smith: The computing world at large will suffer the commoditization scenario I mentioned before, with tremendous negative consequences for anyone currently in that business. As for HPC, it will be forced to use chips originally intended for games or other applications for which parallelism is available and understood.
Q: You are well known for your pioneering work in designing multithreading architectures. Do you see yourself continuing in this direction? To put it another way, what's next for Burton Smith?
Smith: My multithreading work is continuing to have some impact on the field, but it's up to commercial hardware companies to take it forward now. As for me, I have always enjoyed using what I have learned to extend my circle of colleagues and to help me learn new things. My work at Microsoft is just such an experience, and it's definitely going to be my "what's next" for the foreseeable future.
For more on ISC'07 keynote addresses and the conference program, please see the conference/programs section of the ISC'07 website (www.isc07.org).
Source: Karen Green for ISC'07