1 min read
Are Carbon Offsets for Open Source software doable?
Instead of planting a tree or sequestering CO2 why shouldn't we spend that money making open source software more energy efficient?
"Keep the coal fired power plant -- we'll just optimize openssl instead!"
Heh, of course this is like cap-and-trade, where scripting languages are the polluters :)
Does this mean we can tax the crap out of Oracle for Java!
Actually Java is not bad in terms of CPU use, it can be a memory hog, and due to trying to be OS agnostic, it is bad at using modern system APIs, but in terms of being power efficient it is not bad.
( I have actually been thinking alot about this topic over the last year. )
Just think of all the power used swapping pages to disk :-)
Russell Cattelan -carbon- computing tax vs cap-and-trade :)
And JVMs are not soooo bad. Jonathan's correct that the memory overhead is a bit ugly (48+ bytes overhead for a String in openjdk comes to mind).
And if you want energy efficiency Dalvik does okay.
Me? I'd just like to find more ways of funding grungy optimization work on the open stack that powers most of the computing grid...
Paul Lindner I wonder how hard it would be to find funders for some proposed Apache/python/php/ect improvement if you could could show a watt per mega request gain or something?
I know it is not the short term answer but I really think we need new language that has high-level primitives but lets you get down to the OS/bare metal and control memory layout of your data. Rust seems like a possible candidate but it is hard to tell at this stage how it will play out.
Rust looks interesting, and Go is also a contender here. Will have to see if I can get someone interested in megawatts/changelist...
I suppose one could use Alexa or what ever to figure out how much power could be saved across the net for some change.
at least for http servers
I am with you. Ruby, Python, etc programmers should be buying offsets. How many mega wats a lost to bad code.
Jonathan Moore , Aug 29 2012 on known.dev.inuus.com