Advances in semiconductor manufacturing are making it easier to etch two or more processors onto a single piece of silicon. For the last year or more, IBM, Sun Microsystems, and Hewlett-Packard have been building such “multi-core” chips for use in advanced business servers, and this week Intel announced that it plans to create multi-core chips for consumer desktop PCs. The advantage of multi-core chips is that they can finish certain tasks faster by running them in parallel, and can run multiple operating systems and other software more easily by assigning them to separate processors.
Sounds great, but there’s a catch, according to an interesting article this week at ZDNet: Some software makers are treating dual-core chips as if they were two separate computers, effectively doubling the cost of software licenses.
Intel, Sun, AMD, and Hewlett-Packard are urging software makers to count processors on a per-chip basis, but database firm Oracle continues to charge double for software intended to run on dual-core chips, and others companies charge a similar premium. But that pattern can’t hold forever. Many observers say that as chips with two, four, sixteen, or more processors become more common, software companies will have to rethink the very nature of their licenses–for example, charging by the amount of computing work done rather than by the number of copies of a program that are running.