1000 Core “Proof of Concept” By Intel Stirs The Imagination
It’s a thing that the imagination can hardly conceive and the most avid gamer could only wish for. Intel has brought us a step closer to having the unicorn-like 1,000-core processor with a “proof of concept” that can perform nearly 20X faster than “modern computers.”
A team of scientists lead by Dr. Wim Vanderbauwhede have used Field Programmable Gate Arrays (FPGA) technology to create the processor. Intel has been playing around with the super-processor concept for a few years now and shipped samples of its 48-core processor earlier this year.
Intel’s Terascale Computing Research Program is working through the issues of scaling processors at this level. Last month, engineer Timothy Mattson made some waves in the tech world when he said that 1000 cores was not only possible but really only faced a handful of roadblocks.
Mattson laid out those challenges when he asked, “The challenge this presents to those of us in parallel computing at Intel is, if our fabs could build a 1,000-core chip, do we have an architecture in hand that could scale that far? And if built, could that chip be effectively programmed?”
The possibilities are apparently endless, but that isn’t stopping speculation as the tech world buzzes and attempts to grasp what this really means for computing.
1000 CISC core or 1000 RISC core processor? will be a major question as dual core CISC CPU can do the same job in less instructions then a dual core RISC CPU can but the RISC CPUs designed by ARM can work independently of each other so that is another question which the article did not address. The probability of all 1000 cores working in tandem or independently of each other which seems like it would be a better design then having to wait on each other to complete the operation to continue on since there are so many cores. The ability to work independently of each other will probably be a need for a CPU when you get to a very high core count as if 999 cores have to wait for 1 core to finish a longer operation so they can continue then they are not being that effective compared to other designs of less cores for the same job sent to the abstraction layer to divide into threads to be fed to the hardware CPU(s) physical and logical. A CISC CPU with a high core count and with the ability to work independently of each other will work much faster then one with same CISC CPU speed high core count processor without the ability to work independently of each other thus making that one ability a must for higher core counts for CPU intenive jobs on severs and other machines which actually use all the cores.