Many of those older computers are the first computer of someone around the globe still, at least as long as they keep working.
If I am to gauge my app's speed on an anemic hardware I'd prefer running it on a modern Atom or Celeron because (a) indeed they're anemic but (b) at least are more eco-friendly.
On this lane of thought, I'd say better to buy a laptop today (that's much greener than everything before it) that you can hold on to for 10+ years than holding on to another that has long passed its expiration date. But I am aware that this is not always a popular opinion.
I was simply mentioning that the machines get manufactured anyway so I definitely refuse to be guilt-tripped into "it's your fault that those machines exist!" extremist stance. No, it's not my fault. The machines are manufactured regardless of what I do so every 7-8 years I evaluate the market and upgrade.
Reasons are simple -- most modern software gets slower all the time and I still need to work and support my family. I would hold onto my machines for a lifetime but it's not the reality we live in and I wish the downvoters stopped being so tunnel-visioned and were aware of that.
Agreed. A base MacBook Pro isn't nearly powerful enough for all 3 of those at once.
Same goes for power. They're not going to burn one less lump of coal because your laptop uses 100W instead of 300W
Even when we collectively manage to reduce our energy usage they'll likely divert the extra energy to another power substation for storage and redundancy.
So don't think that me and others are defeatist. It's just that the entire grid is well planned for so that any difference we can make, even as 100_000 people, is still not as big. So this whole thing will take a while, likely decades.
In the meantime I was very happy when I replaced a fridge with 43kW monthly expense with a 19kW one, about two years ago.
That's not how it works. On the electric grid, generation must always be precisely matched with consumption, otherwise things can literally burn up. Any excess of generation will cause an increase in voltage and frequency, any lack of generation will cause a drop in voltage and frequency. That's why the generation always follow the consumption; all generators sense changes in consumption (by measuring the system frequency), and adjust their input to match (for instance, by adjusting the fuel intake). When they don't (usually because the change was too fast, like when large blocks of generators or consumers drop suddenly from the grid), protective systems will disconnect the generators and/or consumers until the grid is balanced again. Storage (which is still uncommon) only does a temporal shift of the consumption and/or generation, it doesn't change the amount (other than the inevitable losses).
That is, if you use 200W less, the generators will adjust to generate 200W less power (actually a bit more than 200W, because of losses). If millions of people use 200W less, the generators will adjust to generate hundreds of megawatts less power.
(Also pondering moving my NAS to an ARM SBC as well but not sure it's worth the hassle since the mini Intel i3 PC idles at 10W.)
For work however, a programmer just can't do with those underpowered machines if they want to be productive and not wait 3 minutes for incremental compilation after each change they make (which an LSP server does, and even if it didn't, re-running tests does the same anyway).
I am happy to work on a more ecological machine if somebody optimizes the dev tooling and my programming languages' of choice compilers and linkers. I'd absolutely love it if all my machines idled at 5W and peaked at 35W. But if I am to support my family, that stance isn't easily achievable today. Yet.
And, the amount of electricity that you, the developer, spend testing your code will almost always be eclipsed by that of your users if more than a dozen others use your tool.
So, again: why does it matter how power-efficient the CPU you're testing on is, as long as it's slow (to produce the proper throttling effect)?
I was alluding to replacing the idea of holding on to a Pentium 4 with buying a modern, more eco-friendly, but ultimately just as anemic, CPU.
Not a perfectly representative test, sure, I just felt that it was a good compromise between "test your code on weak machines" and "be environmentally responsible".
So yep, I did conflate weak CPU with low-wattage CPU, you are right.