Looking at this issue from the perspective of time spent solving the problem versus the speed increase has been perfectly valid in the past and even now. We are, however, starting to enter the era where other concerns emerge and begin to dominate. The energy spent annually into processing data in all its forms is enormous. The sum totals we see now are already comparable to national energy budgets and will certainly increase in the future. This is a significant source of CO2 emissions, too.
http://arnoudm.files.wordpress.com/2009/01/co2datacenters.jpg
How large part of this energy is currently spent processing unnecessary code from unnecessarily large frameworks, using unsuitable algorithms to solve well understood problems or wasting CPU cycles in otherwise inefficient ways? As software - like frameworks - bloat, this is only getting worse and worse. Advances in computer architecture and power efficiency are not even close to keeping up with the pace in which new needs for data processing are being uncovered. This is not so much of an issue in personal computers, while the idle power consumption is still relatively high. As these computers are slowly moving towards being able to sleep when not in active use, it's becoming more important to write software that doesn't keep computers artificially active. This is also a domain where significant energy and cost savings await, should the software developers be aware of the benefit of efficiency.
Programming effectively in any language is something that should be appreciated in equal merit because of the efficiency gains and not just because of the advantages it brings to the user or the development process. This is something that I believe will be more widely recognized over the next decade.
Slashdot Mirror
Looking at this issue from the perspective of time spent solving the problem versus the speed increase has been perfectly valid in the past and even now. We are, however, starting to enter the era where other concerns emerge and begin to dominate. The energy spent annually into processing data in all its forms is enormous. The sum totals we see now are already comparable to national energy budgets and will certainly increase in the future. This is a significant source of CO2 emissions, too. http://arnoudm.files.wordpress.com/2009/01/co2datacenters.jpg How large part of this energy is currently spent processing unnecessary code from unnecessarily large frameworks, using unsuitable algorithms to solve well understood problems or wasting CPU cycles in otherwise inefficient ways? As software - like frameworks - bloat, this is only getting worse and worse. Advances in computer architecture and power efficiency are not even close to keeping up with the pace in which new needs for data processing are being uncovered. This is not so much of an issue in personal computers, while the idle power consumption is still relatively high. As these computers are slowly moving towards being able to sleep when not in active use, it's becoming more important to write software that doesn't keep computers artificially active. This is also a domain where significant energy and cost savings await, should the software developers be aware of the benefit of efficiency. Programming effectively in any language is something that should be appreciated in equal merit because of the efficiency gains and not just because of the advantages it brings to the user or the development process. This is something that I believe will be more widely recognized over the next decade.