To wrap up the positive portion of our annual Winners and Losers list this month, we look at an innovative software platform for the new generation of multicore processors and an embedded memory technology that promises to make future versions of those processors perform faster and cost less.
In "Winner: Cure for the Multicore Blues", Senior Editor Harry Goldstein discusses the advent of the RapidMind Development Platform 2.0 from RapidMind, of Waterloo, Ont., Canada, which enables developers to create amazing applications capable of wringing every drop of speed and complexity from the most advanced multicore processing systems on the market today.
To demonstrate the capabilities of RapidMind, its creator, Michael McCool, a 39-year-old computer science professor developed an artificial intelligence program that controlled the individual behavior of 16 000 virtual chickens, on a farm, to run on IBM's new nine-core Cell microprocessor. The demo wowed them at last year's Game Developers Conference in San Jose.
"I wanted to build something that I could teach in about 10 minutes, that you could use without mental overhead so you can focus on the algorithms, not the details of the particular processor," McCool told Goldstein. With multicore technology, programmers accustomed to writing serial algorithms must now think about parallel algorithms, Goldstein notes, and one of the benefits of working with the RapidMind platform is that users become familiar with a conceptual model of a parallel machine. "It's similar enough to a real parallel machine that you can reason about what is an efficient way to implement an algorithm," McCool said.
In a different computing category, Senior Associate Editor Samuel K. Moore writes in "Winner: Masters of Memory" that a new memory technology from Innovative Silicon, of Lausanne, Switzerland, may change the ground rules for microprocessor design. Called Z-RAM, for zero-capacitor dynamic random access memory, the design requires no new materials or extra processing steps in the fabrication process to attain higher levels of performance—no exotic semiconductors, no oddly structured parts, and no experimental insulators. Each memory cell is just a single transistor.
Moore writes that you can fit as much as 5 megabytes of Z-RAM into the space occupied by a single megabyte of today's conventional embedded memory. That lets you greatly increase the amount of memory on the chip and thus improve its performance, make the chip a lot smaller and cheaper, or do a good deal of both. "The transistor is the most studied device in the world," Serguei Okhonin, Innovative Silicon's chief scientist, told Moore. "To make it work as a memory, we had to find something different."
To that end, Z-RAM's designers found a way to temporarily store a bit as a charge inside the body of a transistor made on a silicon-on-insulator semiconductor wafer, which is gaining ground as the substrate for high-performance processors, such as the above-mentioned Cell and Advanced Micro Devices' Opteron. According to Moore, the Z-RAM has the potential to quintuple the amount of memory incorporated into microprocessor chips and make them both faster and cheaper.
So, for mastering the intricacies of designing a software platform for today's hottest commercial hardware, we designate RapidMind 2.0 a technology winner. Likewise, the Z-RAM memory breakthrough deserves a laurel as a technology winner for the design team from Innovative Silicon. Congratulations to all the winners this year.