mratsim

43 ms·

451.
by mratsim 7y ago
and introduce new flamewars about "Not-Invented-Here", why don't you contribute to a common codebase, ffmpeg vs libav/avconv debacle.
452.
by mratsim 7y ago
I'm not advocating for or against accepting the patch. I'm saying that the tone of the refusal is inappropriate.
453.
by mratsim 7y ago
- I wouldn't conflate blockchain and cryptocurrency. Cryptocurrencies are one (infamous) use-case of blockchain technology. - I agree that cryptocurrency turf fights are toxic - censorship resistance does not equate privacy. The main b
454.
by mratsim 7y ago
The maintainer still shouldn't dismiss a patch + test case with "it's boring".
455.
by mratsim 7y ago
Thank you very much. Are there some deleted comments that might have been missed? If not the conversation was civil until the maintainer refused a patch for a "boring" reason.
456.
by mratsim 7y ago
Probably related to the Reddit job ads when you read /r/rust. That said, blockchain companies using Rust take security very seriously. It also funds cryptographic research which has a huge need of memory safety, performance and th
457.
by mratsim 7y ago
I didn't think I would see a "I will pay you with exposure" argument.
458.
by mratsim 7y ago
They owe people who write PR basic courtesy, not refusing patches on ground that it is "boring".
459.
by mratsim 7y ago
But then you get drama about Not-Invented-Here, why are you creating yet-another-X-framework instead of improving the existing ones. Or you get drama like ffmpeg vs libav/avconv.
460.
by mratsim 7y ago
The PR contributor was not paid either to investigate the bug, reproduce it, write a patch and test it. However instead of raising a "don't work, please fix" bug s/he took the time to do all that. Anyone that went to suc
461.
by mratsim 7y ago
RHEL and CentOS are not targeting the same users as Arch. Distros targeting the same desktop needs are Ubuntu, Mint, Debian-flavors, Fedora, Gentoo, ... In corporate environment for critical software you would get a contract with an on-call
462.
by mratsim 7y ago
Uncertainty is absolutely key everywhere decision making has an impact on human lives. Detecting if we see an expresso or a cappucino, sure fine. But uncertainty on a CT scan would be extremely valuable just to avoid people (even trained pr
463.
by mratsim 7y ago
While I have a lot of critics for Academia in particular regarding code publication or reproducibility or the chase for the next conference paper, this is completely uncalled for. Their real world constraint is publish or perish, and they d
464.
by mratsim 7y ago
If you only had a domain you're likely not affected (metadata). If you also have a website hosted there (data), you may be.
465.
by mratsim 7y ago
No that's not true. As soon as you use sequences or strings or async you depend on the GC. You can however compile with gc:destructors or gc:arc so that those are managed by RAII.
466.
by mratsim 7y ago
I guess, convincing Elon Musk that we need tractors on Mars could stimulate some competition.
467.
by mratsim 7y ago
See http://www.1024cores.net/home/lock-free-algorithms/tricks/fi... on ucontext And for longjmp, Apple source for example https://opensource.apple.com/source/Libc/Libc-262/i386&
468.
by mratsim 7y ago
I don't agree. For many domains, being able to implement a domain-specific language with a set of expressive rules for the domain is an incredible productivity boost and also prevents many mistakes because you cannot represent them. No
469.
by mratsim 7y ago
C is portable assembly until it isn't: - no access to carry/overflow flag in registers making it a chore to write bigint libraries - no way to guarantee emitting add with carry or sub with borrow even though those are simple instr
470.
by mratsim 7y ago
You call malloc/free, or if working with GPU cudaMalloc/free. You can write your own memory pool or object pools, you can use destructors, and even implement your own reference counting scheme. This is what I use for my own multit
471.
by mratsim 7y ago
Apparently it's fantastic so Clojurescript is here to stay, the future stack will probably be Clojurescript in the frontend and Nim in the backend. See our Clojure repos: https://github.com/status-im?utf8=%E2%9C%93&
472.
by mratsim 7y ago
We did cover part of it here: https://our.status.im/status-partners-with-the-team-behind-t... But to give more detailed explanations: - We are a blockchain company and close partner with the Ethereum Foundation. Ethereum re
473.
by mratsim 7y ago
Status dev here (i.e. colleague with the benchmarker) and prolific high-performance Nim libraries author. Note that Nim standard library focuses on maintenance cost as the Nim team is small. When raw performance is needed, Nim gives us the
474.
by mratsim 7y ago
Nim compiles to C, there is no cost in calling C.
475.
Show HN: Zero-Functional 1.0, Zero-Cost Functional Programming in Nim(github.com)
1 points by mratsim 7y ago|1 comments
476.
by mratsim 7y ago
In the mean time, public transportation the SNCF railway company is accumulating debts at an amount that would raise brows in a private company. Though I don't have any stance on the private vs public for utilities/healthcare and
477.
by mratsim 7y ago
Isn't Twitter ad revenue actually super low compared to the infrastructure costs?
478.
by mratsim 7y ago
Good question. I can't find a single SPIR-V AXPY (a*X+y) example which is the Hello World of linear algebra in Vulkan/SPIR-V so I'm not optimistic on ease of use. You can probably do texture hacks like what people did in 2006
479.
by mratsim 7y ago
Instead of having to reimplement a series of linear algebra transformation as coarse-grained transformation on ndarrays/tensors, you create a domain-specific language that still give you the mathematical feel. This produce a computatio
480.
by mratsim 7y ago
The follow-up paper should include the deep learning and linear algebra compilers: - Halide - Tiramisu - Ngraph - Weld - PyTorch JIT & PyTorch Glow and Tensor Comprehensions - TVM - Futhark - A toy MLIR language build on the linalg dial

More ›