Imagine Principal Investigator: William J. Dally

Overview The focus of the Imagine project is to develop a programmable architecture that achieves the performance of special purpose hardware on graphics and image/signal processing. This is accomplished by exploiting stream-based computation at the application, compiler, and architectural level. At the application level, we have cast several complex media applications such as polygon rendering, stereo depth extraction, and video encoding into streams and kernels. At the compiler-level, we have developed programming languages for writing stream-based programs and have developed software tools that optimize their execution on stream hardware. Finally, at the architectural level, we have developed the Imagine stream processor, a novel architecture that executes stream-based programs and is able to sustain over tens of GFLOPs over a range of media applications with a power dissipation of less than 10 Watts. More details on this work is provided in the project pages.

Research Contributions Stream Architecture The Imagine Stream Architecture is a novel architecture that executes stream-based programs. It provides high performance with 48 floating-point arithmetic units and a area- and power-efficient register organization. A streaming memory system loads and stores streams from memory. A stream register file provides a large amount of on-chip intermediate storage for streams. Eight VLIW arithmetic clusters perform SIMD operations on streams during kernel execution. Kernel execution is sequenced by a micro-controller. A network interface is used to support multi-Imagine systems and I/O transfers. Finally, a stream controller manages the operation of all of these units.   Stream Programming Model Applications for Imagine are programmed using the stream programming model. This model consists of streams and kernels. Streams are sequences of similar data records. Kernels are small programs which operate on a set of input streams and produce a set of output streams.   Software Tools Imagine is programmed with a set of languages and software tools that implement the stream programming model. Applications are programmed in StreamC and KernelC. A stream scheduler maps StreamC to stream instructions for Imagine and a kernel scheduler maps KernelC to VLIW kernel instructions for Imagine. Imagine applications have been tested using a cycle accurate simulator, named ISim, and are currently being tested on a prototype board.    Programmable Graphics and Real-time Media Applications The Imagine stream processor combines full programmability with high performance. This has enabled research into new real-time media applications such as programmable graphics pipelines.   VLSI Prototype A prototype Imagine processor was design and fabricated in conjunction with Texas Instruments.and received by Stanford on April 9, 2002. Imagine contains 21 million transistors and has a die size of 16mm x 16mm in a 0.15 micron standard cell technology.   Stream Processor Development Platform A prototype development board was designed and fabricated in conjunction with ISI-East Dynamic Systems Division. This board has enabled experimental measurements of the prototype Imagine processor, experiments on performance of multi-Imagine systems, and additional application and software tool development.

Acknowledgements This research was supported by the Defense Advanced Research Projects Agency under ARPA order E254 and monitored by the Army Intelligence Center under contract DABT63-96-C0037, by ARPA order L172 monitored by the Department of the Air Force under contract F29601-00-2-0085, by Intel Corporation, by Texas Instruments, and by the Interconnect Focus Center Program for Gigascale Integration under DARPA Grant MDA972-99-1-0002. In addition, several current and past students have been sponsored by National Science Foundation Graduate Fellowships, Stanford University Graduate Fellowships, and an Intel Foundation Fellowship.

Translations Ukrainian translation of this page courtesy of Zonderpump

gajh@cva.stanford.edu