Frostedegg is correct.
Swap is the nearly the same as a windows virtual memory. The only difference is that windows virtual memory is limited only to the size of any free space on the hd, while a swap partition is created to only allow a certain amount of the hdd to be used. In the old days like when I ran Red Hat 8.0, it was necessary to run 500 mb of swap space because our computers only had 128 or 256 physical ram on board. Most process in linux don't requires alot of memory, however, as in the server world we all like to have multiple applications opened and the memory is used up. In windows the OS will then retrieve a space on the hdd and deposit data for shorterm storage until the processor is ready to run the cycle. Swap is the same idea. The main problem with swap and virtual memory is that the lower/south side bus is limited in mhz, and will run dramatically slower than the frontside/northside bus. Thus, with standard ram the speed between ram and processor is 333 mhz, or 400 mhz, or with rambus 800 mhz, but the IDE cable and south side bus will only work at 100 or 133 mhz. What about SATA some say? Well SATA bumps that number up to 150 mbs, but it still can not compete with the frontside speeds. Therefore data travels slower from the hdd to the processor which slows the entire cylce rate down. With PCI E and other adaptations in the southside bus the numbers may increase, but that is not the pipes IDE or SATA travel. The best we can hope for are the new hybrid hdd's with flash memory of 256 or 512mbs on board and a 10,000 rpm speed of the platters inside the hdd. Next, they will have to increase the pipes from the IDE or SATA to operate at the same mhz as the frontside bus. Biggest problem is the material to use and the distance between the two. And this is the state of affairs.
Simple solution to swap? Get more physical ram and don't make a swap partition.