--- crowd-funded eco-conscious hardware: https://www.crowdsupply.com/eoma68
On Mon, Jul 25, 2016 at 11:41 PM, Matt Campbell mattcampbell@pobox.com wrote:
Hello:
I read the white paper referenced on the Crowd Supply campaign page [1], and I was particularly interested in the section on "good enough" computing. The following paragraphs describe some problems that, as far as I recall, aren't addressed in the rest of the paper:
great - ok let's go through it.
This "upgrade treadmill" has bothered me for a while. Yes, with modular hardware like the EOMA68 cards and housings, the environmental impact is lessened because we only have to discard computer cards,
don't do that!!! re-use them in down-stream devices, sell them please or give them away! for example using them in a freedombox, router or low-power NAS device which would normally be served by a struggling sub-500mhz MIPS, such devices could *really do* with a 1.2ghz dual-core boost! don't do it!
not whole laptops. But unless we can stop the upgrade treadmill, we'll still have to discard our old computer cards when they would otherwise still be functional.
$3 / month is the going rate for co-located low-power hosting services with a static IPv4 address. 2gbyte of RAM would be seriously welcome, the hosting company would be delighted to be able to slap more EOMA68-A20 computer cards into a convenient rack, i'm sure they'll take them off your hands.
software development, and listening to music. Now I don't know if X would run at all in 64 MB of RAM.
xf86fb version 3 would be fine - i compiled that up in 2004 using bitbake / openembedded... it worked perfectly, the actual executable was only... 450k.
.. but 64mb of RAM is a total luxury for routers, most openwrt devices have only 32mb, some even only have 8mb.
... looks a bit weird to be running a laptop from that era as a router, but hey.
So what can we do about this? The only idea I've got is that I and other software developers should do all of our work on the most underpowered computer that will let us get buy, rather than the nicest one we can afford. Then maybe, out of necessity, we won't be so wasteful. But then maybe we won't be as productive either, particularly if not being wasteful means we have to write everything in C or C++. And of course, it won't do any good if I'm the only one who chooses to make those sacrifices.
At least with free software, there's always the possibility to fork projects that succumb to the upgrade treadmill. For example, the MATE desktop environment is a fork of GNOME 2,
and Trinity is of KDE 3.5, i actually have that for my clients because they prefer it and are used to KDE 3.5 - i was forced to put XFCE4 in front of them recently, they *really* don't like it.
and one of its explicit goals is to run well on non-compositing graphics hardware. I imagine MATE will run quite well on something like the A20 card.
most likely.
But still, we can't live in a forked, time-warped world.
man i run fvwm2 - i realise i'm odd...
We have to interact with mainstream websites, which means using a mainstream browser or at least one of the major rendering engines. In this regard in particular, I wonder if the upgrade treadmill has already left the A20 behind, particularly since we can't use full GPU acceleration. I can certainly understand why the JZ4775 wasn't chosen, though it checks all the other boxes for ethical computing.
i'd add it... if i had time to, and if i can get past the boot phase. i have the early bootloader up, i just need to get u-boot into memory so i can investigate. but if i add it to this campaign, it will have to be at quite a high per-unit cost.
Anyone else have any thoughts on this? Sorry if this is too much of a rant or off-topic here.
no not at all, discussion is great. interested to hear from other people too
FWIW, I just backed the campaign by ordering an A20 card.
yay! thanks matt
Matt
arm-netbook mailing list arm-netbook@lists.phcomp.co.uk http://lists.phcomp.co.uk/mailman/listinfo/arm-netbook Send large attachments to arm-netbook@files.phcomp.co.uk