This is crazy insightful, thanks! I’d really love to learn how to get to this level of understanding, but can’t seem to figure out what curriculum I’d follow where I’d end up with this level of technical competence.
In my experience, Macs use up a ridiculous amount of "System" storage for no reason that users can't delete. I've grown tired of family members asking me to help them free up storage that I can't even find. That's the major issue from what I've seen; unless this app prevents apple deliberately eating up 50%+ of the storage space of a machine, this doesn't do much for the people I know.
I haven't, but I have used PythonCall, and it's really rough to use and constantly go back and forth between types. The main reason I love julia is because all julia packages build upon julia arrays, and don't require conversion to special arrays for that specific package.
I don't have top-tier publications, and I haven't gotten any awards. I've seen people get awards for bullshit and farming prestigous publications. I only have one citation for work that is 100% mine. But every time I read something like this, I feel proud that I don't bullshit, and at least try to do real science. I truly believe in all my work and every sentence I write.
That being said, the insane emphasis on venue is what's pushing me out of academia. I can't compete with people like this.
Julia is still such an absolute joy to use. I have a dream that somehow, Torch.jl, which attempts to provide a julia interface to pytorch, will magically take off. Python is ok, but julia is just so fast, fun, and easy to scale.
I've used, and am still using, Julia for my PhD research. It's perfect for parallel/distributed computing, and the neural network primitives are more than enough for my purposes. Anything I write in pure julia runs really, really fast, and has great profiling tools to improve performance further.
Julia also integrates with python, with stuff like PythonCall.jl. I've gotten everything to work so far, but it hasn't been smooth. The python code is always the major bottleneck though, so I try to avoid it.
Overall, julia is a significantly better language in every single aspect except for ecosystem and the occassional environment issue, which you'll get with conda often anyways. It's really a shame that practically nobody actually cares about it compared to python. It supports multi-dimensional arrays as a first-class citizen, which means that each package doesn't have it's own array like torch, numpy, etc, and you don't have to constantly convert between the types.
I didn’t think I could be further shocked by how dangerously incompetent our leadership is, but here we are. It’s mind blowing that someone would propose this without talking to a single person that understands the subject area.
Good point, We'll just have to see, I suppose. I'm really curious how far Intel will get before they fail, if they do indeed fail. Given the current demand for cutting edge chips and TSMC's risky monopoly, I know a lot of people are hoping they somehow pull through
From what I understand, Intel went with TSMC because their own nodes weren't ready yet, which makes sense to me. I imagine once their node is production ready, they will switch back. It's also interesting that for a single generation or two, they were able to switch to TSMC, with the plan to switch back, which, as an outsider, suggests to me that changing between nodes is not as tall of a task as one might think.
Additionally, I feel a lot of people are forgetting their biggest surefire customer--the United States military. If they get anything operational, even at 2x or 3x the cost of TSMC, the US will buy them and give them the resources and experience to serve larger markets, as the US has done with computing in the past.
They have a mountain to climb, for sure, but the path towards recovery still seems possible as long as Uncle Sam wants to buy from them.
I appreciate the in-depth reply. But if the foundry is able to produce the next gen of their core ultra processors, wouldn’t that lead to a sudden flock of renewed interest, especially if the foundry becomes the new core business model? Not doubting that inertia is there, or that management is incompetent, but they’ve bet the entire company on foundry, unlike before, which shows a lot of commitment this time around. Especially if people aren’t able to get access to the cutting edge node in Taiwan due to giants buying out the entire capacity, it seems like intel would be a natural second choice for anything bleeding edge, no?
From what I can tell, none of these problems really matter if their 18A bet pays off, and this article doesn’t talk about their 18A progress, from what I could quickly read.
Been eyeing this project for a while, hesitant to pull the trigger before a 1.0, just because email is a pain to configure, and having to keep up with changes before a 1.0 would have been a pain.
I'd love to see a blog or some post on the roadmap for this project (and the org in general).
Not an insider, but this doesn't seem good. I more or less agreed with every call Intel's made over the past few years, and was bullish on 18A. I liked that Pat was an engineer. His interim replacements don't seem to have that expertise.
Intel wasn't doing great to start, but Pat put it on the path, potentially, towards greatness. Now even that is in major jeopardy.
Thanks, lol. I actually looked up the spelling of the word when I was writing the article (it was spelled correctly in the first line) but then proceeded to misspell it for the title (:
Interesting, I thought 'Overman' was just the english translation. The context for the time was definitely a shift away from pessimistic Chrstian values, but as a non-christian, it still strikes home with me in the modern era.
My GitHub Pro plan just expired, and the repo was hosted privately, so it became inaccessbile. Just made the repo public, should be viewable now. Thanks for letting me know!
Lina Khan has been fantastic imo, even for tech. I think she forces companies to compete where we actually want competition, and not let us rely on insane levels of lock in
One of the main things I like about Julia is that all the libraries are just built on their core Array type. In Python, there’s torch tensors, numpy arrays, and I think Jax has their own array type too. Also, Julia has a really beautiful distributed computing paradigm, whereas python needs to use libraries like Ray, which have their own quirks and documentation and community
Julia’s been stable for my work, from what I understand, things have gotten better over the years on that front. The biggest issue Julia gives me is packaging, just because two libraries that you import can use different versions of the same dependency (among other things)