[jrmg]

It’s counter intuitive to me that this is a net carbon win.

- Are these phone processors really as compute-pet-watt efficient as a regular data center processor?

- There’s so little embodied carbon in a phone motherboard - and presumably some embodied carbon in whatever custom racking hardware up is being used to house these. Is that really compute-power-per-embodied-carbon-footprint efficient than making a new server?

[Aachen]

I don't have time for an elaborate sourced answer but in short, yes, the efficiency gains of modern devices is almost negligible compared to the CO2e of producing a new device at current device lifespans. I can't remember if you would have to use a device for 15 or 25 years before upgrading is better than continuing to use it, but I thought it was 25

Edit: yes, 25. Found my go-to reference for this quickly after all

> The report about the cost of planned obsolescence by the European Environmental Bureau [7] makes the scale of the problem very clear. For laptops and similar computers, manufacturing, distribution and disposal account for 52% of their Global Warming Potential (i.e. the amount of CO₂-equivalent emissions caused). For mobile phones, this is 72%. The report calculates that the lifetime of these devices should be at least 25 years to limit their Global Warming Potential. —https://wimvanderbauwhede.codeberg.page/articles/frugal-comp...

[jrmg]

The carbon cost to produce or use a new phone isn’t relevant here.

The important question is whether repurposing an already-existing phone to act as a server (including physically adapting the phone and producing whatever extra hardware is required to do that) is less carbon-costly than producing and using a new designed-for-the-purpose server.

[Jaxan]

Producing phones emits a lot of CO2 (or equivalents). It takes more energy to produce, than the amount of energy a phone uses in its typical life.

But yeah, these are great questions which are not obvious at all and should be answered when proposing such a system.

[Flere-Imsaho]

> Are these phone processors really as compute-pet-watt efficient as a regular data center processor?

I have no numbers to back up my argument, but smart phones are very power efficient by their nature are they not? I can play a 3d game, with impressive graphics, on a tiny device powered by a battery... With very little heat generated.

If your goal is to run LLM inference on a gpu in a power efficient manner, I bet a smart phone is a good place to start.