October 08, 2013
Silicon photonics is in the spotlight again, being pitched by researchers at the University of Colorado Boulder, the Massachusetts Institute of Technology and Micron Technology Inc. as a potential Moore's Law extender.
The technology of silicon photonics refers to using light, instead of electrical wires, to enable silicon-based transistors to communicate on a single chip. The technique could lay the groundwork for computers that are remarkably fast, cost-effective and energy-efficient.
The project is headed up by CU-Boulder researcher Milos Popovic, an assistant professor of electrical, computer and energy engineering. Popovic and his team developed the technique, which employs two different optical modulators, structures that detect electrical signals and translate them into optical waves. The major benefit to this particular method is that it can be fabricated with standard CMOS processes already in use by the industry. An article describing the research appears in a recent issue of the journal Optics Letters.
The approach addresses the main stumbling blocks to current transistor designs, energy and heat. Because it takes a lot of electricity to turn transistors on and off, there is excessive heat buildup. The heat buildup means that further electricity must be expended to cool the device. Furthermore, as transistor sizes shrink, the number of wires occupying such a small area of space leads to "cross-talk." The multicore/manycore design was essentially a workaround to this problem but this technique is limited by communication between microprocessor cores, which is also energy-intensive.
Optical communication circuits, aka photonics, is dramatically more energy efficient than electrical wires. A single fiber-optic strand can carry a thousand different wavelengths of light at the same time. This allows multiple communications to take place simultaneously in a small space with no cross talk. The Internet and the majority of phone lines already rely on optical communications technology, but in order to be economically feasible for microprocessors, vendors need to be able to use the same fabrication process and foundries that produce the current generation of microprocessors. This integration of photonics and electronics is what's necessary to get buy-in from the microprocessor industry, according to Popovic.
"In order to convince the semiconductor industry to incorporate photonics into microelectronics you need to make it so that the billions of dollars of existing infrastructure does not need to be wiped out and redone," he added. Popovic and researchers at MIT have demonstrated that this is indeed possible.
Two papers published in Optics Letters describe an optical modulator that is compatible with a current manufacturing process known as Silicon-on-Insulator Complementary Metal-Oxide-Semiconductor, or SOI CMOS. This is the same process used to manufacture cutting-edge multicore microprocessors such as the IBM Power7 and Cell, which is used in the Sony PlayStation 3.
The research team also describe a second optical modulator that could be created with another popular chip-manufacturing process, called bulk CMOS, currently used for memory chips and high-end microprocessors.
10/30/2013 | Cray, DDN, Mellanox, NetApp, ScaleMP, Supermicro, Xyratex | Creating data is easy… the challenge is getting it to the right place to make use of it. This paper discusses fresh solutions that can directly increase I/O efficiency, and the applications of these solutions to current, and new technology infrastructures.
10/01/2013 | IBM | A new trend is developing in the HPC space that is also affecting enterprise computing productivity with the arrival of “ultra-dense” hyper-scale servers.
Ken Claffey, SVP and General Manager at Xyratex, presents ClusterStor at the Vendor Showdown at ISC13 in Leipzig, Germany.
Join HPCwire Editor Nicole Hemsoth and Dr. David Bader from Georgia Tech as they take center stage on opening night at Atlanta's first Big Data Kick Off Week, filmed in front of a live audience. Nicole and David look at the evolution of HPC, today's big data challenges, discuss real world solutions, and reveal their predictions. Exactly what does the future holds for HPC?