Dunno why you'd want a program like that to force-ably split CPU resources as XP is quite capable of doing it on it's own. All you are doing is limiting the CPU resources, succinctly demonstrated as you described PERL running half speed. If a program is 'locking up' the computer, then the program has a flaw, or has crashed in some way.

Anyway, this machine I'm on now is a Core2 Duo 3Ghz with only 2 gigs on WinXPproSP3. I currently have at least a dozen websites open simultaneously in Firefox, several PDF files, music, email, newsgroups, and programming. No slow down. In fact, as I write this, task manager is telling me I'm using 979 megs of ram and 3% CPU.

As for games, that gets into new territory. Games are not really CPU intensive, but they are graphics intensive. Even my flight simulator is more GPU than CPU. Thus, you need a good video card. The GPU (graphics processing unit) does all the work, not the CPU.

GPU's are specialized chips that do one task extremely well - render graphics.

CPU's by their nature are general purpose and are therefore an exercise in weighted compromise.

Lately, though, GPU's have become GPGPU's - general purpose GPU's. Software can now use the computational power of GPU's for general purpose computations other than graphics. For example, NVidias CUDA architecture - http://en.wikipedia.org/wiki/CUDA

It is inherent to a video card's job to need to do a lot of math, so this is one of those things it can do very much better than the CPU. GPGPU's can allow for dramatic speed enhancements for tasks needing to do a lot of math. For example, most of the graphics rendering programs I use can also use the GPU in addition to the CPU to process the image. (as an aside, the kind of graphics I generate are different than gaming graphics, so need a lot more computing time)

To give a concrete example of the speed increase, I at times have run the Seti@home project. When they enabled GPU usage, I discovered that my one GPU was capable of processing as much work as my quad core CPU, thus doubling my work output.

There are now even add-on boards that have multiple GPU's with no graphics capability for the express purpose of intensive number crunching. For example, NVidia's TESLA - http://en.wikipedia.org/wiki/Nvidia_Tesla. Pricey, though.

Current supercomputer designs are a mixture of massively parallel multi-core CPU's complimented with multi-GPU add-on boards. For example, the current fastest computer in the world is China's Tianhe-2. It contains 16,000 compute nodes, each with 2 Xeon CPU's and 3 Xeon Phi chips for a total of 3.12 million cores. Each node has 112 gigs of ram - a total of 1.75 terabytes of RAM. It's theoretical top speed is 54.9 petaflops (54.9 million billion floating point operations per second). It consumes 17.6 megawatts of power to run, plus another 24 megawatts to cool it.

http://en.wikipedia.org/wiki/Tianhe-2

Even though many people can be computer savvy, I think I have an additional advantage in that I also understand the electronics behind it all. Went to school for that. I know how a bit gets from a hard drive into memory, used by the CPU, sent to a graphics card, and turned into text on the screen. Although the technology has advanced in the 25 years since I went to school, the fundamentals haven't changed. Only the implementation has improved dramatically.

Brian