On Wed, Aug 30, 2017 at 9:59 AM, Mick <michaelkintz...@gmail.com> wrote: > On Wednesday, 30 August 2017 15:42:47 BST Ian Zimmerman wrote: >> On 2017-08-30 09:32, Mick wrote: >> > > Unfortunately this isn't a viable strategy because typically you >> > > will, in a few months, if not a single month, spend more in >> > > electricity costs than you would purchasing a new single board >> > > computer. >> > >> > Perhaps in a commercial 24x7x365 high compute cycle application this >> > would hold water, but in the case of a home PC running 14 hours a day >> > at maximum power you might save enough to buy a small spinning SATA
Only recent hardware is actually any good at frequency scaling, or adjusting power consumption when frequency scaling. If the processor saves 10W that's nice but the rest of the computer might still be using 50-100W idle. >> > drive after a year, or a Raspberry Pi without peripherals, but not a >> > new PC. Of course, if: >> > >> > 1. your PC is not running at full speed all the time; >> > 2. it is not a PentiumD dual core (were they the most power hungry?); >> > 3. you're not still running a CRT monitor; >> > 4. you tend to suspend to RAM when not in front of it; >> > 5. a new PC is not at least 50% more efficient; >> > 6. the price of electricity is not exorbitant (I pay approximately >> > £0.13/KWh + £0.29/day) >> > >> > then you will need other reasons to upgrade. When the PC you're using >> > is a laptop, then the case for upgrading on grounds of savings on >> > electricity costs alone is even more tenuous. >> I think it is useful to talk about absolute cost vs. relative cost. The absolute cost is likely low enough you don't care, and if so that's fine, I understand. The relative cost of running a single board computer as opposed to modern desktop is something like 100x cheaper, or two orders of magnitude (~2.5W vs. ~120W). Older desktops fare less favorably and may be 200-500x more power hungry. >> Also: how long is the replacement going to last? Anything with flash as >> the main storage will be back at the recycling station (ideally) within >> a couple of years. This includes all the consumer routers I've ever >> had, including the beloved blue Linksys. > Flash storage lasts far longer than that in practice. How often do you upgrade a router? Even the very inexpensive flash (10k to 100k write cycles) in microcontrollers tends to never reach its useful lifespan even when those parts are used for development; that is, receiving 10-20 updates a day for a few months (worst case ~1% of useful lifespan, more typically ~0.1% of useful lifespan or less). > With consumer grade router/modems I've found the capacitors are of a low > rating and therefore within a few years (or sooner if your area experiences > brown outs and power cuts/surges) they give up the ghost. Replacing the > capacitors in their power supply and sometimes a couple of their internal > capacitors with capacitors of a higher rating for just a few cents, by passes > this built-in obsolescence and extends their useful life for quite a few more > years. I haven't noticed anything similar but I do not recommend Netgear products anymore. The TP-LINK router I purchased has been in use for a few years and has fared favorably. These failures seem to be related to "abusing" the router. If you keep it relatively uncovered in a livable space it will do fine. If you put it outside in a shelter it will not. I do not necessarily agree that computers should be designed to operate at *only* 25C STP (your livingroom) but that is the current state of the market. R0b0t1.
