I think the point of these was for when they read low-precision values out of a printed table and then multiplied them using slide rules. Even though the article doesn't say it, I can't think of any other practical reason to multiply by adding logarithms.
Are the "File 1" and "File 2" links in mail supposed to open something? I got to that point and I think that's how to find the bank account number but I'm not sure.
While CEO, Pinchai's and Cook's companies increased in value by 2.7x and 8.6x respectively, so they must be doing a very high quality job at what they were selected for.
Evangelicals are Protestant so they don't really care about Roman Catholic doctrine. Even though they agree on many issues it doesn't mean the Pope would change their minds.
If you wanted to generate a longer essay by GPT3 but make it really hard to tell, you could just get it started with a 1-2 sentences. Then let it generate N-1 tokens (where N is the window size, I guess 2048), and write a few tokens yourself (and repeat). Then I don't think there's any way to reverse engineer the internal state of the model at any point in the text. It would be like trying to find a string that hashes to a given value.
Apple doesn't directly compete with CUDA, they just want total control of their platforms. Metal does have great performance and tooling. In practice nobody does HPC on macs so there's no demand for linear algebra or graph libraries, which are a big selling point for Nvidia over AMD.
The fact that Apple is trying to kill OpenGL and OpenCL and block Vulkan definitely sucks though for anyone trying to do indie games, or open source ML/HPC.
I've never been to wikinews before, but it looks like exactly what I was talking about. They have a prioritized list of articles that need to be reviewed for bias and accuracy, so that's really good.
They didn't really go that wrong, Cori and Trinity are still useful machines. But I can say for computational science, programming models have gotten way better in the last few years. Now it's as easy to get a new sparse algorithm to high occupancy on a GPU as it is to scale on a manycore CPU. So GPUs just look better now considering cost, power efficiency and software support.
So that means the planes were intended to be at levels 300 and 290. Is it normal to have only 1000 feet of vertical separation? It seems like the higher plane could easily lose 1000 feet in a few seconds before ATC could tell the lower plane to move away.
Metadata from one person isn't worth anything, but the graph of who's contacting who is very useful. Plus, it's computationally feasible to do that graph analysis on everyone, while listening to every call wouldn't be.
To do star star x = f(y), star x and f(y) each need to be evaluated. You'd think the right-hand side would always be evaluated before the left-hand, but that's not always possible (what should "arr[i] = i++" do? I stole this from wikipedia btw). In the above case, f(y) has this kind of side effect that change x, which is undefined.
Rather than changing the macro, this could be fixed easily by putting the assignment into its own function. Then the argument (f(y)) must be fully evaluated before star x.
Edit: I have no idea how to escape asterisk on HN :(
In many situations yes, and C/C++ compilers miss a lot of opportunities for constant folding/propagation. But still, an infinitely smart compiler can’t prove anything about a value that comes from a file, user input or socket at runtime. If control flow depends on those (which is true of every useful real-life program) then you’re out of luck. GMP arbitrary-size integers are amazingly fast but still much slower than native instructions, can’t be vectorized, require heap allocation, etc.
It can absolutely still be useful, and it would make sense for an early stage startup or someone's side project to use it so they're not duplicating work porting it to .NET, Cocoa and Qt or whatever. And sure, having an identical look and feel everywhere is nice. But with all the resources that Slack and Microsoft have, they need to publish a lengthy guide for the users to deal with their performance problems [1]? They take 22 seconds to load a 6 MB XML file [2]? Ehhh... it's not asking that much of a multi-billion dollar company. You deserve better!
I always thought of Electron as a way to kick out a minimum viable product quickly before writing native versions. But nope, I guess give a $16 billion company 4 years and 1600 employees and that's still not feasible.
I would say the biggest challenge with general AI is that our brains are NOT empty when we’re born, and we have no way to map those connections. We don’t have memories yet, but we’re wired to recognize faces, learn languages and use sensory data. If a baby’s brain was empty, we could most likely bootstrap a human on an exaflop supercomputer today by feeding it the things a baby would see/hear.
Yep. That reminds me, there's a major piece of OpenMPI called "schizo" that supports "multiple personalities" of runtime environment. I don't know how they got away with that one...
I just googled "PC professor firing" and found these examples right away. What's especially bad is that the departments publicly denounce the person. In [1] a committee of deans said he was "uncivil, illogical and damaging to rational thought" which is just not true (he anonymously tweeted opinions that can be debated and aren't objectively true or false). This is still going on, [4] happened a few months ago.
In most contexts that's how it is, which is great. But it's going way too far on college campuses, where professors can get crucified on social media and then fired for saying the tiniest "problematic" things.
If anything goes wrong, gravity drains the molten salt into a storage pool that spreads it out, stopping the reaction and letting it cool and solidify. The lining of the underground pool can be made arbitrarily thick and durable. Contamination like at Fukushima is physically impossible unless someone deliberately damages that lining.