CompactFlax

Well… if it’s in a cloud provider with 3 copies locally and 3 more geo-redundant why would we need that?

You’re right, but it is an easy fallacy to fall into.

It’s still pretty much 2x. I managed to get ram bought before it went silly but I’m really short on storage.

None of the chips built for datacentres are useful for gaming. None of the RAM in these chips is useful for normal computing - even servers - much less home gaming.

AI compute believes it’s still in a “race” so they won’t want to be using previous gen chips unless forced to in some way.

This isn’t the “dark fibre” of the AI bubble. It’s just a future pile of trash.

To build chips that are used as collateral for datacentres that have yet to be built, so the chips can’t be used, and they’ll just age out of usefulness.

pay for all the required transmission infrastructure that the project will use exclusively.

Emphasis mine. If the AI buildout can claim that the line will serve as a secondary entry point for a neighbourhood they’re not going to pay.

I skipped the 17 which was supposed to be my partner’s upgrade. I am anticipating regretting that decision.

It’s crazy and the chips that are ostensibly taking up all the dram are sitting on shelves waiting for data centres to be built so that they can be racked.

The thing that interests me is that a local business person would see this law and say “fine, no porn”. Remember, when Yahoo bought tumblr, they sank it by banning porn!

So to me this suggests they can’t figure out how to make their LLM reliably not produce pornography.

Apple’s AI strategy is leaving them behind

Apple’s stock drops on failure to meet AI promises

Etc.

Now whose stock is dropping?

There’s a bunch of legislation and case law saying that searches are OK within x (50?) miles of a border crossing which covers something like 90% of the population because they count international airports.

So it might get sticky for the person.

SpaceX will be worth more than Earth if we achieve our goals," stating without evidence that his extremely far-fetched prediction was “obviously true.”

Musk has been spending too much time with a sycophantic ego stroking LLM.

Without googling the specs, are those two on a different channel? ie the chipset has 6 and then another chip handles 2 or vice versa

For mobile

I initially thought this was an argument for electron/PWA bullshit. “Why is <app> eating 2GB of RAM and has no locally loaded content?”

When companies are pushing apps as hard as they usually are, I assume there’s a benefit to them and not to me.