Back in November I clocked up one million giganodes on the OGR-25 project at Distributed.net. That took me about 7 years. As predicted (roughly) I have doubled that in just over 8 months thanks to a speedy dual-core processor in my main PC. It would have happened sooner if I had realised that my PC was not running at full speed and had implemented a proper shut-down script so that I did not lose work units each day. I was a little optimistic in how long the project had to run. Current projections give it almost another year, but I think that does not take faster computers into account. I wonder how many people will upgrade in that time. I certainly do not intend to do so unless I suddenly find myself with enough spare cash for a faster CPU. I could get something 25% faster for well under £100, but I doubt I would notice the difference for normal usage. If I were building another PC I would probably look for something that used less power rather than ultimate clock speed. I've not bothered keeping up with what's new in the processor world for a while, so I don't even know what the state of the art is. These days software is more interesting than hardware.
For reference, the earlier slowdown may only have affected apps like dnet with a high nice value as the PowerNow daemon would ignore them and assume that nothing important was running. By setting the '-n' flag in its defaults file I got back to full speed. I used a script from here to make sure the client shuts down cleanly. I could have been losing several hours of work in the past. I've at least learnt some more about the workings of Linux through all this.