DARPA wants to speed up computers by over 1000 times while using less power

DARPA has revealed the research teams selected to drive two efforts to go beyond Moore’s law. DARPA’s $1.5 billion Electronics Resurgence Initiative (ERI) wants to jumpstart innovation in the electronics industry.

Three Dimensional Monolithic System-on-a-Chip (3DSoC) program has groups from the Georgia Institute of Technology, Stanford University, Massachusetts Institute of Technology, and Skywater Technology Foundry were tapped.

DARPA thinks the 3DSoC effort can produce a 50-fold reduction in computation times, while using just a fraction of the power. The 3DSoC designs should support an inter-layer interconnect bandwidth of 50 terabits per second and require no more than 2 picojoules per bit to access memory.

It will address the memory limitations of traditional architectures related to bandwidth, latency, and energy consumption. This is already being done with current 3D and 2.5D stacked memory devices that are integrated with NVIDIA Tesla and AMD Radeon Instinct GPUs, as well as Xeon Phi processors. The 3DSoC designs will be more complex, and have a dozen or more layers, and integrate things like resistive random-access memory (ReRAM), carbon nanotube transistors (CNFET), and regular silicon MOSFET-based processor cores.

Stanford simulations show energy use and execution time will be 323x to 646x better compared to conventional 2D chips at 7nm.

The Foundations Required for Novel Compute (FRANC) program will come from HRL Laboratories; Applied Materials, Inc.; Ferric, Inc.; the University of California, Los Angeles; the University of Minnesota; and the University of Illinois at Urbana-Champaign.

FRANC will fix the memory-logic bottleneck. It will transcend the conventional separation of logic and memory functions in what are known as von Neumann architectures. They will process the data where it’s stored, rather than having to move it to where the computing elements reside, lowering energy consumption and improving throughput. The multi-pronged approach will involve developing novel circuit designs and using new materials and integration techniques to minimize data movement.

FRANC will use new materials and devices to make 10x advances in embedded, non-volatile memories with the speed of static random access memory (SRAM) and the density of storage-class memory.

5 thoughts on “DARPA wants to speed up computers by over 1000 times while using less power”

  1. “That is probably because DARPA has a habit of actually getting what they pay for.”

    Yeah. It seems like increasingly only the millitary can get anything groundbreaking or revolutionary accomplished in this decadent world we live in.

    “I am still upset that we still don’t have the DRACO (double-stranded RNA activated caspase oligomerizer) that was initially partly funded by DARPA. That thing appears to kill just about any virus, potentially curing hundreds of diseases. ”
    ” But that was only found out because we have antibiotics, and people connected the dots. If we had antivirals that could kill virtually any virus, I believe a large number of conditions might be cured.”

    But does it attack all viruses (anything with an RNA), or just specific strains? In the latter case, I don’t see this being developed to attack random viruses in hopes that it will cure something unknown. Not to mention viruses that are not know to be in the human body, or completely undiscovered.

  2. That is probably because DARPA has a habit of actually getting what they pay for. The results are just usually outside the price most people would pay. But things have a way of getting cheaper.
    Not sure how this is going to get cheaper though. It takes time to build layers. This is probably in the hundreds of layers.
    And, the US seems eager to buy expensive hardware for aircraft and such.

    I am still upset that we still don’t have the DRACO (double-stranded RNA activated caspase oligomerizer) that was initially partly funded by DARPA. That thing appears to kill just about any virus, potentially curing hundreds of diseases.

    Some diseases thought to be lifestyle related or psychologically based turned out to be bacteriologically caused (ulcers, and lower back pain, and several others suspected) . But that was only found out because we have antibiotics, and people connected the dots. If we had antivirals that could kill virtually any virus, I believe a large number of conditions might be cured. And, of course, all the ones we already know are virus caused have a chance of being cured.

    There is reasonable evidence that three strains of Adenovirus are responsible for the obesity epidemic. They do in animals, they infect humans, and statically obese people are much more likely to be positive for the one of the three that is more easily detected. The other two have not been tested.

    An number of viruses have been linked to all sorts of psychological diseases. Just statistically I think it is likely at least some are caused by viruses. The ones I strongly suspect are ones that start early in life. If it was all genetic those genes would have been weeded out ages ago. They just don’t make sense from a selection standpoint.

    They keep finding one cancer after another that is actually caused by a virus. And they are not rare viruses: Epstein-Barr virus, human papilloma virus, hepatitis B virus, and human herpes virus-8 all cause cancer. And these viruses are everywhere. We now have vaccines that cover 4 kinds of HPV, but there are over 170 kinds, and vaccines only protect you before you get them. Something like DRACO could possibly cure all 170. We could then proceed to wipe it off the face of the planet. Epstein-Barr has been shown to be associated with all sorts of stuff. That is what causes mononucleosis, also called mono. If you have had that, then you are infected. Also known as human herpesvirus 4 (HHV-4). It is one of those nasty herpesviruses (there are 9 that infect humans) that never leaves and generally hides out in the brain/cerebrospinal fluid from the immune system. We need to wipe them out.

    Well, I am drifting off topic, so I will stop. DARPA, though, had provided funding for some really great stuff over the years.

  3. You add internal heat channels to carry the heat to the outside. From there you can do liquid cooling if necessary. But various process improvements can reduce the amount of heat that’s being generated, so air cooling may still be enough.

Comments are closed.