wewbull

joined 1 year ago
[–] wewbull@feddit.uk 55 points 5 months ago (23 children)

So... Absolutely need to be aware of the impact of what we do in the tech sphere, but there's a few things in the article that give me pause:

Research suggests, for instance, that about 700,000 litres of water could have been used to cool the machines that trained ChatGPT-3 at Microsoft’s data facilities.

  1. "Could". More likely it was closed loop.
  2. Water isn't single use, so even if true how does this big number matter.

What matter is the electrical energy converted to heat. How much was it and where did that heat go?

Moreover, when significant energy resources are allocated to tech-related endeavours, it can lead to energy shortages for essential needs such as residential power supply. Recent data from the UK shows that the country’s outdated electricity network is holding back affordable housing projects.

Can you say non sequitur ?

The outdated network holding back housing is that it doesn't go to the right places with the capacity needed for the houses. Not that OpenAIUK is consuming so much that there's no power left. To use a simily, there's plenty of water but the pipes aren't in place.

This article is well intentioned FUD, but FUD none the less.

[–] wewbull@feddit.uk 46 points 6 months ago (2 children)

Absolutely this. Microsoft is going headlong into the AI abyss. Google should be the company that calls it out and says "No, we value the correctness of our search results too much".

It would obviously be a bullshit statement at this point after a decade of adverts corrupting their value, but that's what they should be about.

[–] wewbull@feddit.uk 3 points 6 months ago

Still worth trying.

The word "support" is used in a very corporate way in the ROCm docs. They mean in terms of support contracts, sending engineers out to customer sites, having people dedicated to specific customers.

The code works on more devices than they list, but they won't enter into a support contract with anyone using anything but their latest and greatest.

[–] wewbull@feddit.uk 4 points 6 months ago (1 children)

"set the magic environment variable because the tool chain will mis-detect the architecture of your unsupported card"

I don't think it's misdetecting it. Rather it detects it correctly, tries to use specific support for that device, but then finds that the support was switched off at compile time. The environment variable forces it to pretend to be a different (very similar) device.

Clunky, yes.

[–] wewbull@feddit.uk 10 points 6 months ago (1 children)

Whereas on Linux I recently upgraded the motherboard on my machine from a B350 to a B550, stripping it down to it's parts and rebuilding. Different network chip, audio chip, WiFi and Bluetooth, etc, etc. 6 SSDs plugged back in in a shuffled order.

Linux booted and worked first time, adjusting which drivers it used automatically, mounting all the drives in their original locations. Similar thing when I upgraded my GPU. Admittedly the old one was AMD, same as the new one, but there was about 4 or 5 generations between them. CPU upgrades too.

I've got a real machine of Theseus here. I think my case and my heatsink is all that's left from the original.

...oh...and the OS.

[–] wewbull@feddit.uk -5 points 6 months ago (2 children)

No. It means the sound energy is dropped by half. Our audio perception is also logarithmic however. It's why we use db.

[–] wewbull@feddit.uk 19 points 6 months ago (3 children)

And so are our ears. That's why we use db. So 12db is not perceived by us to be 94% quieter.

[–] wewbull@feddit.uk 1 points 6 months ago

I keep trying catalog but it doesn't do anything.

[–] wewbull@feddit.uk 1 points 6 months ago

Signal is now "phone number to login to the account, but you can share a username with other people"

Not 100% but better than some.

[–] wewbull@feddit.uk 2 points 6 months ago

Blue bubbles.

(I'm half joking)

[–] wewbull@feddit.uk 3 points 6 months ago

The Accusation: Skewed Benchmarks for Inflated Performance?

Oh no! Say it's not so. /s

We all know not to trust benchmarks from the manufacturer. We know this because we've all been burned by cherry picked results under favourable conditions that can't be reproduced.

[–] wewbull@feddit.uk 5 points 6 months ago

For me the answer is always "snapshots" and normally because of docker.

If you run a docker image store on a BTRFS drive, docker creates snapshots at various times. It never cleans them up; It has no commands that clean them up, and it means that if you delete a file it doesn't free any space because the snapshots keep the file alive.

view more: ‹ prev next ›