279 karma · joined June 24, 2021
The cost? You have to research a machine that will run Linux, not everyone can do this. I'm quite happy with my hp zbook is it a mac, no bit it's as close as I can go.
with that said I was blown away by the steamdecks support for AMD's APUs so I was looking for a laptop with the same generation APU to ensure support.
I was looking at framework(too big wants 13-14") dell xps (I don't like the new design)
Then HP released the ZBook Ultra G1a 14" and I brought it on the spot because it lined up before ram went to hell (128GB unified RAM).
They do have an ubuntu image but it's heavily patched, so I was running bleeding edge mesa for a few months and beta kernels for 6 months.
On Gentoo this isn't too bad but it wasn't a turn-key experience.
Regulate them! that is the only way.
Their should be a path for an AOSP build to be just as privileged as a google signed build.
The driver situation is a better but for cheap hardware it's so hit and miss, and when you mention this they shame the user for not knowing better and that isn't their fault.
At least if they buy a mac they know it's going to work otherwise you need to buy a computer that's two years old.
1. government gets new income source.
2. government is invested in making sure they are not getting short changed.
3. failure to pay me is fraud.
4. When you pay me now I know your crap is spying.
5. Government taxes almost never go down, with this system it going down will immediately be visible to consumers who will balk.
Problems: ad revenue is the new shrink-flation without it prices will rise. companies will fight this tooth and nail, but voters get "free" money.
1. This chip for an 8B model even if it was done at 5nm would still be twice the size of a conventional CPU die so what are the yields for this going to be like for even a 30B model?
2. They say 2 months but llama 3.1 was released 2024, ~2 years which is normal lead time for silicon, I suspect this would take longer if the architecture is not llama?
3. Can google do the same thing in house with their Gemma 4 series (two year lead time puts Gemma 4 on silicon April 2028) ?
I can see the benefit for hyper scalers but at the rate of model turn over does this type of investment make sense?
I don't think the government themselves are prepared for the amount of identity theft that is going to follow mass collection of government documents.
That being the case my understanding is splitting this project up is not only a technical challenge but a legal one.
You could make a common API for driver code, this would make the main driving force of linux much weaker.
Nvidia tried to stay out of tree for years and was dragged kicking and screaming upstream because they needed the reliability.
I'm saying the very thing you are calling out as a problem is currently what is forcing everyone contribute to the mainline kernel.
Without centralized leader ship the facturing of the linux kernel would be a disaster.
You can always use XWayland if you feel that way.
In that light I’m not happy about it, but the code always was just a means to an end.
Then they just validate on the front end, and we validate on the backend.