Laser-powered CPUs
CPUs are way behind GPUs. It was nothing like this when dedicated GPUs appeared in the mid 1990s but now its more than a fact. That’s why ATI and Nvidia are aiming at general purpose GPUs (GPGPUs) with their next-gen series. And that’s why Intel is preparing its Larabee CPU-derived GPU that it’s supposed to leave ATI and NVIDIA with a fat lip. ATI also prepares the Fusion array of products which will combine CPUs and GPUs in a single chip. OK, I digress, but for a solid reason: CPUs are not fast enough as they’re supposed to. It’s because the CPU manufacturers still use some old technologies.
Nowadays, the CPU components are interconnected with nano-scale wires or pipelines and this is one of the obsolete technologies I was talking about earlier. Just think about simulating the Big Bang or create artificial intelligence for instance. With present day technology, you need a cluster of parallel-supercomputers to initiate large-scale simulations and AI computations. That’s beyond energy efficiency. The idea is simple: replace the nano-scale wires with something that would transport information faster. Replace them with lasers.
In this sense, Sun Microsystems has just received a $44 million contract from the Pentagon to play around with lasers. They're to work on a way of connecting silicon chips via lasers, which, if successful, will increase chips speeds by a factor of thousands.
If all works according to pla, it would be possible to create more-compact machines that are a thousand times faster than today's computers. Each chip would be interconnected directly with every other chip in the array via a beam of laser light that could carry tens of billions of bits of data a second. However, Sun claims that they are only expecting a 50% success rate, so we might not see any commercial implementation until they figure out how to up the success rate.
If you enjoyed this post, please consider to leave a comment or subscribe to the feed and get future articles delivered to your feed reader by FeedBurner.
Comments
Those are all terrible points. CPUs aren’t really behind GPUs. You act like the procesors in GPUs are already based on some more advanced technology. Its the fundamental difference between a Single Purpose CPU and a General Purpose CPU. That point isn’t even related to the use of optical processors.
And the whole point of a GPGPU is to offload common graphical based calculations to the GFX card so it doesn’t have to wait for the CPU. So physics calculations etc.. could be partially ofloaded.
And optical processors have been a dream for several years. Theres still a long long way before we see them in any proper use.
Cambridge have been working on this for years already WTF