The growth of the web has given rise to more distributed ways to harness computer power - usually of the humble desktop type.
These projects typically take advantage of the fact that the processor inside most desktops spends most of its time idle. Rather than just let it sit doing nothing, it is getting increasingly easy to get it doing real scientific work.
One of the first to use that idle computer time was the Seti@Home project, which harnessed processors to look through reams of radio signals seeking signs of alien life.
Now, many more projects are keen to tap that processing power.
"Some scientists have single jobs that they need to have done very fast like simulating an explosion or something like that," said Dr David Anderson, director and developer of the Boinc distributed computing project.
"Other scientists have many, many jobs and want to complete the whole group of them relatively fast," he said.
This is where Boinc's volunteer computing project comes in. Those that download the Boinc software can offer their PC to help with one of many different scientific projects.
"Volunteer computing is by and large doing the same sort of computation that people do in supercomputer centres or other conventional hardware," said Dr Anderson.
Get enough PCs together and you can amass a significant amount of computer power, he said.
"There is in the order of 350,000 people running Boinc on their computer," he said. "We hope and expect that number will increase."
"Currently the largest supercomputers are 1-2 petaflops machines," he said. "We have 3 or 4 times as much power on tap."
The separate Folding@Home project, which is studying the ways that proteins fold inside cells, can call on about eight petaflops of computer power. Other distributed computing projects are using games consoles such as the PlayStation 3 and Xbox as sources of compute power.
"The volunteer computing approach works for a surprisingly large fraction of all scientific tasks, about 95%," said Dr Anderson. "Those that it does not work well for are ones that require individual jobs to be done as fast as possible or require large amounts of memory."
More . . .