I repeatedly keep seeing praise for Ilyas achievements as a scientist and engineer, but until ChatGPT OpenAI was in the shadow of DeepMind, and to my knowledge (I might be wrong) he has not been that much involved with ChatGPT?
the whole LLM race seems deaccelerate, and all the hard problems about LLMs seems not do have had that much progress the last couple of years (?)
In my naaive view I think a guy like David Silver the creator/co-lead of Alpha-Zero deserves more praise, atleast as a leader/scientist.
He even have lectures about Deep RL after doing AlphaGo: https://www.davidsilver.uk/teaching/
He has no LinkedIn and came straight from the game-dev industry before learning about RL.
I’m not optimistic about AGI, but it’s important to give credit where credit is due.
Even assuming the public breakthroughs are the only ones that happened, the fact that openai was able to make an llm pipeline from data to training to production at their scale before anyone else is a feat of research and engineering (and loads of cash)
I agree. Quanta is the sort of material I like to read during my morning coffee ritual before I start a day of screen time.
Are there any quality print magazines like it? I used to get Physics World a long time ago and have been looking for something in print to scratch that same itch.
I think it is remarkable how important ABI is to C++. I think 'making it easier to enable them' i.e. enable the sane defaults went down the drain once they locked in on not breaking ABI.
I am sure that Herb wrote this with good intentions, but no concrete measures was proposed but trying to standardize compiler tools, and to some extent downplay Rust.
I agree there is a misconception around programming safety, but this reads the same as when big industry say the want to focus more on climate change.
I have not found a better 'true' IDE experience than JetBrains. Once it has done it's thing, it works pretty good. If your projects are big like a C++ codebase +2mil LOC then well the latency is going to suffer.
If you count VSCode as an IDE then I would suggest neovim with an LSP and a fuzzy finder. It is the fastest/best dev setup I have experienced for most things.
If you are very latency sensitive I am afraid an IDE/'smart'-editor may not be for you. Then I again would consider some lightweight editor that has fuzzy search and no intellisense.
this makes me believe you never were in a quant shop, sorry.
At least if your goal is making money.
I think it can be a fun project to make an engine, and try to make models, together with some form of ingestion pipeline and finally solve and calibrate the models. Try some simple products, and see how close you can get to real world prices.
Making pricing/risk engines are in my opinion somewhat close to game engines even through that the domain is so different.
Somewhat related: In 2020 gcc bumped the requirement for bootstrapping to be a C++11 compiler [0]. Would have been fun to see the kernel finally adopt C++14 as the author suggested.
I don't think that Linus will allow this since he just commented that he will allow rust in drivers and major subsystems [1].
would have hoped see more answers or see something in here from actual kernel developers.
I don't think this is true, but I don't know this for sure.
From what I have read it seems that 'only' Bloomberg, Meta and Microsoft that uses c++ with exceptions.
And since both microsoft and meta are adopting rust in their services it seems to me that they are looking for another language than C++. (why else adopt a new language?)
LLVM has been for quite some time driven by Apple and Google.
WebKit, another Apple child.
Qt, it has to support environments where exceptions are not allowed, otherwise they would be losing customers, specially since Qt is older than C++98.
gcc, was initially written in C, and for quite long time had a mixed code base with minimal C++.
The companies adopting Rust aren't doing so because of lack of exceptions, they would still adopt Rust if the language had exceptions support (which panic and std::ops::Try kind of are), rather due to the type safety that C and C++ aren't able to provide.
You would be surprised how many games actually do support exceptions.
> - Ukraine will be forced to start negotiations on giving up land to Russia.
Unfortunately, you probably right. Very bright seen in Ukraine, Korean scenario.
Most, West could do at the moment, to convince Russia, to not include Ukrainian territories into their official papers, but territories will remain under occupation.
One thing that would have helped me when I started learning C++ was learning the C++ compilation model i.e. translation units.
It is surprising how many people look surprised (that also claim they know the language) when you tell them that code in .cpp does not get inlined into other .cpp files no matter what doing compilation. (yes the linker can (and should) do that with LTO doing linking)
Compilation units are mainly about symbol visibility and as you already realized at the end of your comment don't have anything to do with inlining as far as the C++ standard is concerned. Neither does the inline keyword btw.
The linker doesn't inline anything, LTO/LTCG is about running (part of) the compile process at link time. But that's really no concern for the C++ code but an implementation detail of the toolchain - you could just as well not have a separate link step and instead have the compiler process all compilation units in one step.
yes, LTO itself does not inline but from a user point of view it does, and that is exactly my point. If you are a beginner and you do not come from C, then you might not think about these things (maybe you do I did not).
I get that it is just an implementation detail, but so are many things in C++ and you usually care about these things otherwise you would not be using the language in the first place.
Many projects also disable exceptions that is also an 'implementation detail' (or at least something that is definitely not required by the standard I would imagine), but now you are technically not writing C++ anymore.
I guess this is also one important thing when learning C++. The community is very fragmented, and there seems to be disconnect between C++ users and committee.
This also helps understanding why template code completely wrecks compiletimes and ram usage since the compiler cant share template instantiations. This becomes very relevant if template metaprogramming is used in bigger projects.
That's not why. Template processing happens during processing of the translation unit, so it would be expensive even if you only had a single TU in your build. It's true that a template has to be reprocessed for each instantiation, however, that's not merely from one TU to the next, but even inside each TU! For every distinct value of T for std::vector<T> in a single TU, the compiler has to process and generate std::vector entirely
I think one clear difference is that they are just so focused on shipping compared to the others, and they gain many of the benefits that comes with that.
I think it is reflected in both Greg and Sam that they really want to ship, and this have made a positive feedback loop into the team and what talent they have been able to acquire but also build up.
Another point might be that AI chatbots are a first-movers market. Even if Grok turned out to be much better I would still miss some of the UI features that ChatGPT provide along with my chat-history.
In regards to their fast shipping I think it is also reflected in their tech-stack. I suspect from reading their job posts (I might be very wrong here)
that they started just coding everything in Python and the tooling/ecosystem that goes along like FastAPI/Django etc. maybe a bit C++/CUDA for the training.
Then when they needed to scale they migrated from Python to Rust in the more critical areas of the codebase. They clearly also have a monorepo mentioned from [1].
if you look through the their career-page the job description of a software engineer for developer-productivity [1] mentions "Our current environment relies heavily on Python, Rust, and C++" also "Are a proficient Python programmer, with experience using Rust in production" I found an earlier one where they mentioned that their backend was written in Python.
"Proficiency with some backend language (we use Python)" [2]:
the whole LLM race seems deaccelerate, and all the hard problems about LLMs seems not do have had that much progress the last couple of years (?)
In my naaive view I think a guy like David Silver the creator/co-lead of Alpha-Zero deserves more praise, atleast as a leader/scientist. He even have lectures about Deep RL after doing AlphaGo: https://www.davidsilver.uk/teaching/
He has no LinkedIn and came straight from the game-dev industry before learning about RL.
I would put my money on him.