hot_gril

38 ms·

721.
by hot_gril 2y ago
Huh, whether these addresses are static or dynamic shouldn't depend on that. I just have a typical router at home with a web portal, is that insufficient?
722.
by hot_gril 2y ago
I'm a home ipv4 admin. What I or someone similar would need is the last part that shows how you set up a basic network with some local addresses I care about and a few I want to expose publicly. It's nice that the guide includes t
723.
by hot_gril 2y ago
(But it seems like an ordered index is fundamentally easier for write performance)
724.
by hot_gril 2y ago
I'm not very familiar with this area, so this is a useful insight. Otherwise I was really confused what's so special about CUDA.
725.
by hot_gril 2y ago
Goes to show how difficult and important execution is. There was also Hangouts vs Zoom.
726.
by hot_gril 2y ago
You and a few others have said that OpenCL is buggy on AMD. Is it buggy on Nvidia too?
727.
by hot_gril 2y ago
They aren't ignoring Microsoft and Google, that's ongoing too. But I agree that they should not be messing with the AI market.
728.
by hot_gril 2y ago
Nvidia and AMD (previously ATI) have had a similar relationship for a long time, before ML was a use case. Nvidia has always been the more expensive option afaik. Kinda like Intel vs AMD.
729.
by hot_gril 2y ago
Did any of those things change? Other than renaming IE to Edge.
730.
by hot_gril 2y ago
I think a century or longer ago, they wouldn't care. It was already pretty easy to fall off a horse and die.
731.
by hot_gril 2y ago
We also won't be ready for the sci-fi-style space age in real life unless there's a use case for it. Imagine Star Wars except there's no life on any reachable planets, what would they even do with their spaceships? And Star T
732.
by hot_gril 2y ago
It's circular the way it was described. Seems like Linux servers are fine with Nvidia GPUs for ML training, but not for graphics, which is more applicable to Linux desktops.
733.
by hot_gril 2y ago
All those Linux servers using Nvidia cards must be getting good enough support from Nvidia for what they need to do, otherwise they wouldn't use it. GNU/Linux desktop is a different story. Yes it's accurate to say that's
734.
by hot_gril 2y ago
They made CUDA though. And they still support the open standards.
735.
by hot_gril 2y ago
I would rather let Nvidia take the win for now since they innovated to make CUDA. This is totally not a played out space yet; it's a kludge in the first place that GPUs are used for ML.
736.
by hot_gril 2y ago
Is Nvidia's advantage literally just CUDA?
737.
by hot_gril 2y ago
Why would Nvidia intentionally use monopoly power to give Linux a hard time? You're also free to use AMD or Intel.
738.
by hot_gril 2y ago
Yeah, the profit doesn't mean anything on its own.
739.
by hot_gril 2y ago
Isn't Dvorak an issue in a lot of places for this reason?
740.
by hot_gril 2y ago
I like it. One thing, it'd be nice if it showed a cheat sheet of the keys. There are 9, and I keep forgetting what does what.
741.
by hot_gril 2y ago
Hm yeah, now I'm wondering how MySQL implements it. Consistent hashing is one way to limit the scope of a remap.
742.
by hot_gril 2y ago
I can see this. Spanner is an example where you don't want this, idk if that's considered common enough. Postgres and MySQL both support hash indexes that are unordered, but the default in both is btree, and Postgres hash indexes
743.
by hot_gril 2y ago
Depends heavily on what kind of database/index. Going with anything other than a random uuid4 adds complexity; for one, do you want to expose time info? I'd rather default to uuid4 as the client-exposed ID* and only change if ther
744.
by hot_gril 2y ago
I don't get why relational DBs are usually taught around natural keys. With MySQL/Postgres you almost always want bigserial PK.
745.
by hot_gril 2y ago
The client doesn't create queries in this situation, you explicitly define them server-side. So optimization is easier. I've seen this time and again with services that aren't GraphQL but tried to provide a semi-freeform quer
746.
by hot_gril 2y ago
I've considered a Chromebook at work where the real coding and builds are being done over SSH on a headless Linux machine anyway, but the fuss involved for just a basic shell and SSH were enough to turn me away. So I can't imagine
747.
by hot_gril 2y ago
Hm. Maybe powers of 4 would be fine.
748.
by hot_gril 2y ago
Yeah, in general it's not this easy. In this case, seems like the worst case is getting x.999 of each chip.
749.
by hot_gril 2y ago
I'm not very experienced at hosting, but this is pretty close to what I'd do. If chip supply isn't a constraint, I'll aim for halving the quantity each step up then redistribute from there, so 12x25c 7x$1 2x$5 in your ex
750.
by hot_gril 2y ago
I can see it, OS doesn't nag the user and things stay static aside from automatic updates, which you could even disable if you wanna be around for them. Problem is even some web things will not work the same in Linux, for example Netfl

More ›