Over the coming decade, memory and inter-core bandwidths must scale by orders of magnitude to support the expected growth in per-socket core performance resulting from increased transistor counts and device performance. Recent developments in nanophotonics can be crucial in providing required bandwidths at acceptable power levels. To investigate the potential bene?ts of nanophotonics on computer systems, HP has been working on an architectural design called Corona.
Corona uses optically connected memories (OCMs) that have been architected for low power and high bandwidth. A set of 64 OCMs can provide 10 TB/s of memory bandwidth through 128 ?bers using dense wavelength division multiplexing.
Once this memory bandwidth comes on chip, the next challenge is getting each byte to the right core out of the hundreds on chip.Coronauses a photonic crossbar with optical arbitration to fully interconnect its cores, providing near uniform latency and 20 TB/s of on-stack bandwidth.
Sluggish transfer is predicated on electric interconnects and the only predicted cure is photonic interconnects. But perfecting the design and manufacturing of tiny lasers and optical channels printed directly on-die is pretty much a hefty task even for the brilliant minds of the 21st century.
Hewlett Packard Comp.’s (HPQ) research wing, HP Labs, revealed information this week on its “Corona” photonic computing push, which aims to produce a 256-core chip with on-die photonic data links between the cores by 2017.
The truth is some of the technology needed to build Corona does not exist yet. However, scientists are having so much of hope in this project and the only thing that comes to mind is the quote by Margret Mead, “A small group of thoughtful people could change the world. Indeed, it’s the only thing that ever has.” What do you think?