There‘s an interactive benchmark further down. They‘re measuring the time to start a container. Honestly something I don‘t cafe for, 200ms more for a container start I do every several days is not something I notice.
And the memory davings is for more than 100 processes. So more than 100 containers? I‘ve got a couple running. Won‘t notice a difference again.
Orbstack is great because the bind mounting is really fast. No talk about that one…
Honestly, if the database is not able to be run with a variety of queries other than the one used by ClickBench then don't submit it to the benchmark. Every other database in their list can run anything you throw at it. Hiding behind messaging that it was an early 0.4 build that can't do everything is ... weird! Just don't submit it to ClickBench if it is not usable by the world yet.
I hope they don't allow pgrust in again until the project matured. Otherwise the benchmark will be useless as it gets flooded by many projects that can run ClickBench fast but have questionable support for everything else.
Yeah, it was a mistake to submit it as early as we did. I was very excited when we got the early version working and overeagerly submitted it to ClickBench
The article says that other models are better at performance or cost. But whaz is not mentioned? Models that are _better_. From my testing, most models are much worse. And if you do an analysis on quality/performance/cost no one is beating Jev. Heck, no one is beating it at performance and cost at the same time.
There‘s a lot of hype with many releasing their decision models the past weeks. But all are worse. So maybe the years they spent in stealth is something you cant replicate in two weeks?
Look up Laya. Also the notion of "best" is flawed because half of that equation depends on the use case. Benchmarks try to answer that by creating artificial use cases and test in isolation but that's not the concrete answer for a given use case either.
And the 'better' ones are slower and cost more for a tiny bit more accuracy. Its hard to sell that as being better when speed and price have been JEVs main selling points.
Its a false premise so often repeated. I am really tired. Compilers are ruleset: If you pass it one statement it'll produce the same output all the time. Its easy to verify it works. And if enough people have used it before and found no problems, there probably aren't many. Spoiler: People always find niche bugs.
But an LLM will generate you two completely different things today and tomorrow. Heck, even if you stash its changes and redo the same prompt.
Even though I disagree with (when I worked on CUDA at nvidia if I ran the TOT compiler on the same shaders one day and the next they could often change), the point is not to compare the two things side by side. It's just to say there are lots of times in life where for whatever reason we can give up old jobs.
When encountering a difficult problem, add an indirection.
That indirection, in the case of LLMs, is formal proof. It can actually turn an LLM into a sort of compiler. Where, if the compiler run completes successfully, you don't need another run, and you are sure it is correct.
Its running for 15 minutes and then must shut down for a long time to cool down. Thats exactly why those "science-fluencers" said its a dumb idea. This is far away from proving that its a viable idea.
_Radiative cooling_ is the only option because it's in vacuum. But you can transfer heat to the radiator either passively (using thermal conductivity), or actively (using coolant). The second option allows to effectively use larger radiators.
BTW, we seem to have Noisesphere instead of Noosphere.
> The cooling system will only be able to operate in brief spurts of about 15 minutes. After that, the TPUs need to shut down to allow the radiators to catch up.
Sam is always shitting on others. For years he has been screaming how bad PostgreSQL is compared to MySQL. Until getting silent about this topic for a few months and then launching PostgreSQL support. Seems he had to realize than the market for PostgreSQL is too big to ignore.
And there is no single company developing. Its a healthy mix of engineers from many companies. A perfect model to prevent this issue here of being aquihired.
So no way it will be as cheap. That is just wishfull thinking.
reply