mratsim

40 ms·

421.
by mratsim 7y ago
I totally agree for things like streams or IO like file handling or authentication. It may be fatal at the IO layer but in a GUI you want a nice error displayed to the user and not a crash. However in some domains you don't want to pol
422.
by mratsim 7y ago
interesting point of view
423.
by mratsim 7y ago
GCC is bad for anything related to big integer arithmetic or cryptography unfortunately. Something like 20 to 40% slower than Clang. Example of GCC awful codegen on add with carry: https://gcc.godbolt.org/z/2h768y
424.
by mratsim 7y ago
I hope one day we will be able to use C to implement efficient multiprecision arithmetic. GCC codegen for add with carry, even with the addcarry_u64 intrinsics is a disgrace[1]. [1]: https://gcc.godbolt.org/z/2h768y
425.
by mratsim 7y ago
They don't want to lose "old" insights if the growth of the group makes it that over 10k messages are sent in a short time (say 2 weeks).
426.
by mratsim 7y ago
You shouldn't use exceptions in cryptography as well. There are constant-time concerns and dumping sensitive data in the heap concerns at the very least.
427.
by mratsim 7y ago
Curious there is no reference to the BLS Short Signature scheme. While pairing-based cryptography is not quantum resistant (yet? Can we mix in isogenies in there?), the signature, secret keys and public keys are really small, for example wi
428.
by mratsim 7y ago
Luxembourg saves on the barriers/checkpoints/controls that people actually paid for their tickets. That said, it's a very small country and so the infrastructure cost are very contained. Also it's balance/budget is
429.
by mratsim 7y ago
You need to compile from source but PyTorch and Tensorflow now support ROCm in their build configuration. Numpy never supported GPUs and the most advanced Numpy-like was Cupy supported by the Chainer team which stopped development in favor
430.
by mratsim 7y ago
Author here: Actually it's 7 [1] But in my future backend revamp I plan to make the default 6 [2] Note that the line can easily be updated to const MAXRANK* {.intdefine.} = 6 as in the planned revamp and then you would be able
431.
by mratsim 7y ago
There were some studies about the impact of the loss of some natural languages. Some represent a complete different ways of handling the world that are worth keeping. And there is no one size fits all.
432.
by mratsim 7y ago
It's easy to call preprocessor macros from Nim, no difference from wrapping C: Example on wrapping a simple benchmark macro: - https://github.com/mratsim/weave/blob/052ae40a/experiments/e... -
433.
by mratsim 7y ago
Just stumbled upon this, $1M to $3M spent each month to acquire software vulnerabilities. Exploiting them must be worth billions.
434.
Zerodium(en.wikipedia.org)
2 points by mratsim 7y ago|1 comments
435.
by mratsim 7y ago
A country like Switzerland or Liechtenstein?
436.
by mratsim 7y ago
Actually Intel is at the forefront of deep learning software as well: - MKL-DNN (rebranded DNNL) is unique and has no equivalent in the ARM space. It even supports OpenCL - OpenCV - MKL - PlaidML and NGraph follow the current trend of deep
437.
by mratsim 7y ago
> Thus, unlike some other source-code management systems, Git does not (conceptually, at least) record "deltas" from one revision to the next. It thus forms a sort of blockchain, with each block containing the state of the repo
438.
by mratsim 7y ago
Nim memory management is tied to the type you use. You either use: - an object. Which is on the stack and is either trivial or uses destructors and is suitable for embedded due to deterministic memory management - a pointer object. Which is
439.
by mratsim 7y ago
It's decorative, it has no meaning. Similarly you have "les choses que l'on voit" and "les choses qu'on voit" with the "l'" being there for prettyness.
440.
by mratsim 7y ago
For me exceptions are for exceptional cases. Any time you have user input, you can expect errors and so should have proper paths to handle file not found, invalid input, ... For those case, exceptions should not be used.
441.
by mratsim 7y ago
Thankfully, Travis, Azure Pipelines and Appveyor are slow as molasses and if you are forced to wait for them before merging a commit you will optimize the speed so that at least CI is reasonably fast.
442.
by mratsim 7y ago
Electron out? But a fast product that doesn't ship will also die.
443.
by mratsim 7y ago
Rust "fearless concurrency" doesn't cover static deadlock prevention or livelock prevention or misuse of C++11 memory model even if you only use safe code.
444.
by mratsim 7y ago
I would be quite interested in an IR/JIT assembler specialized for vector instructions with ARM Neon and x86-64 SSE~AVX512 output. Ideally it handles register allocations and generated function caching as well. The current JIT assemble
445.
by mratsim 7y ago
Wow, I did use Opera for years and also on mobile when in Africa with limited data plans. Time flies, I still remember Opera 10 years ago, they probably had the best tech April Fool's I ever stumbled upon with their "Face gestures
446.
by mratsim 7y ago
This goes both way though. Be nice when people submit patches especially when you reject them. The abusive comment seems to have been a knee-jerk response from someone dissatisfied by the "it's boring" comment about a proof-o
447.
by mratsim 7y ago
It seems like the only "Fossilhub" is this: http://chiselapp.com/ I would guess that it would make it very hard to create a community around the project.
448.
by mratsim 7y ago
"btw I use Arch" is a meme as well
449.
by mratsim 7y ago
Another user provided a patch, the patch got called "boring" by the maintainer, a passerby snapped with offensive personal attacks and the rest is history.
450.
by mratsim 7y ago
He can reject the patch. But the tone is important as well. And that was that "it's boring" reply that triggered an angry response from a passerby about "you shouldn't write libraries" (or something in vein).

More ›