Researchers at North Carolina State University have come up with a way to automate one of the most grueling parts of building a CPU: taking the CPU core architectural designs and implementing them. Usually, the process of going from architectural design to implementation takes hundreds of engineers years, but this new automation process could bring that time down to just months.
Currently, CPU cores are designed to be good at everything. There are ways to make cores really good at just one thing, like retrieving information from memory while processing it, but then it wouldn't excel at parallel processing. Since CPUs in our computers have to handle a wide range of tasks, and since implementation of funky core designs is so difficult, it makes sense to have each core equal to the other and each core able to handle a wide range of computing tasks. With the semi-recent rise in multi-core processors, though, there have been considerations of using different cores all on one CPU and assigning each different core tasks that it excels at. This is usually difficult because it's hard to implement, but with this new automation tool, implementation could be made much easier.
Eric Rotenberg, who is a computer engineering researcher at NC State, said,
So what's this mean for tech-enthusiasts and consumers alike? Well, for one, if this tool is bought/licensed by Intel or AMD, we should see some more interesting core designs from the companies. Another thing we should see is a faster release date between process changes in CPUs. For Intel's Core i series, the first generation was released as early as November of 2008. Sandy Bridge, the second generation, came out in January of 2011. Ivy Bridge just came out in April. If this tool is as the abstract makes it seem, would might be able to see such major processor improvements made all within a year or two.