No open-access location is known for this record yet.
No open-access location is known for this record yet.
Background I run my bioinformatics tasks on two machines. The first one is a Tru64 DS20 AlphaServer, bought in 1999. This has two processors running at 512 MHz with 2 Gb of memory. The second is a custom-built Linux box, purchased in early 2005, which has 8 processors running at 2.7 GHz and 12 Gb of memory. Although performance speed does not quite scale linearly against processor speed, this represents just over a 5-fold increase in computing power over a period of 6 years. This kind of thing has been happening since the 1960s, when Intel-founder Gordon Moore observed that available processing power doubles every 2 years or so. Indeed, my "fast" Linux box is already quite pedestrian compared to the 3.8 GHz processors that are now routinely available. Under these circumstances, it is easy to become complacent about handling awkward jobs. Large, viral-genome-scale ClustalW jobs that used to run for several days on my DS20 are now finished overnight. Within the horizon of a typical 3-year scientific project, I could conceivably be able to buy a machine that shortens the time to a couple of hours or so. But just because bioinformatics tasks are now becoming increasingly trivial in terms of computer time, does that mean we can expect similar gains in systems biology problems? There is a whole industry of popular science books which attempt to persuade us that "Moore's Law" will be the basis of a future world in which anything is computable almost instantly, and we will, even within our lifetimes, know "everything". How seriously should we take such claims and what relevance do they have to research in the here-and-now?
Loading…
Imported from OpenAlex (CC0) — nothing here is generated.