RANT: Gloabal Warming and Software Development.

[DavidS]

There are people that advocate the concept of just use a very powerfull system to build your software, it saves power in the long run, and is only a one time cost. They think this is the way to reduce the overall effect of global warming from the energy used and heat disapated from the thousands of devices that will run the software. I feel this statement is greatly flawed, and I am going to provide my reasoning.

1. The amount of energy it takes to produce that one time cost system is greater than the amount of energy it takes to produce those few thousand SBC's. This is verifiable, you can look up all the data to verify with a few simple DuckDuckGo searches.
2. That "One time purchase" is more like a one time every 6 months to a year purchase. Even a single DIMM takes more energy to make than a Raspberry Pi 3B, so the argument of upgrading for a while does not fly. Otherwise they would still be able to keep up with the build system they purchased in 1984 (probably an Apple Lisa back then ).
3. The amount of energy that this big build system consumes is exponentially greater than the increase in raw performance, and equal to 100's of the target systems (10 watts for a RPi 3B, or 2400 watts for a high end 8 socket 4 cores per socket, 64GB RAM modern current workstation [I do IT for a station that uses these for stream editing, and real time rendering]). Is that really worth it?
4. When compared to building with a faster compiler, the performance difference once done is not enough to save as much power in a year as it took to build the system once by this over do it method, and that is if you have hundreds of thousand of devices using the product. There may be one or two exceptions to this rule, though very few.
5. It still takes longer to build a large project with the high level of optimization than it does to build it with a lower level of optimization and hand optimize the VERY FEW portions that really benifit from optimization. Even if the lower level of optimization is done on an SBC it is saving time. Now this saves a lot of power compared to the BASH IT HARDER way of doing the same thing as is the use a more powerful build system meathod.

So the people that believe that using a more powerful build system is helping save the plannet by providing better optimization, I would say are likely killing the plannet. The little extra optimization they get may save 1000 clock cycles per day per device, though the amount of power saved in those 1000 clock cycles would take millions of devices running for years to just touch the power used for a single compile on the high end build box.

[DavidS]

Remember the old saying "Just use a bigger hammer, it will fit" does not always work, and sometimes is completely the wrong solution.

[DavidS]

I have always been for saving power, worldwide. I would like developers to think not only about the couple of hundreds of watts that may be saved over all of there end users combined (assuming low energy targets of course), versus how many thousands of watts they used to compile the program to save that couple hundred watts worldwide.

People still waste energy (in feuls and electricity) doing things like over using lighting, heating, cooling. You do not need to light an area to be brighter than sunny day outside, you just enough to keep the eyes comfortable and see what you are doing. You do not need to keep the room temperature at exactly 24C, just cool to 28C in the summer, and heat to 20C in the winter, that is going to save a lot of energy. Then there is the use of refrigeration in inefficient ways, there are many top opening freezers that take a lot less power than half of a super energy efficient fridge, so using two of these one set to a temperature for a fridge and one to sub freezing will save a lot over using a single stand up energy efficient fridge.

For that matter if it is consistently below freezing outside for a period of time, then you are wasting energy by using power to cool a fridge or freezer at all.

Then there is the wasted energy in the majority of automobiles, usually in the form of fuel consumed per work performed. In most cases automobiles consume more than three times as much fuel as they need to in order to produce a given amount of work (usually in the form of accelerating, resisting forces to maintain speed, etc). This is one that is more of a challenge to fix, though one to keep in mind.

Then there is the lost energy of transmission on the power grid. An issue that is getting worse as more power is being pushed through the same lines, and they are changing the rating scale so they do not have to upgrade the lines. This is probably the single biggest power waste in the modern world, and a very good case to go off grid (the 25% conversion losses you will end up with are a lot less than the conversion, sync, and transmission losses of the utility company), and a large part of the reason I am off grid for Power (and water).

And what about the amount of power being used to keep the line pressure up in the water mains, to get water to your house? When you can (in most areas, including some deserts) harvest rain water, purify it, and use a gravity feed system for pressure, all with 0 power usage outside of gravity an the potential energy of the water.

Then there is the over use of water heaters. How many people do you know using a tank style water heater? On demand are more energy efficient by a long shot, and direct radiation solar water heaters are even more efficient in overall power usage (and there power comes from the sun, not fuel or electric).

[mahjongg]

David,

Does your rant has any raspberry PI related content, as otherwise its way too off-topic.

I got a few complaints asking me to remove this thread.

[DavidS]

David,

Does your rant has any raspberry PI related content, as otherwise its way too off-topic.

I got a few complaints asking me to remove this thread.

The RPi related content is that it is developing for the RPi and statements about such that sparked the rant to begin with.

If you wish to lock it that is ok, just please leave it if you do lock the thread.

[Heater]

DavidS,

As my colleague used to say to me: "I admire your thinking, unfortunately your conclusions are wrong". I think he was just being polite.

I think I'll leave it to others to explain why.

[jamesh]

Whatever.