mratsim

60 ms·

271.
by mratsim 6y ago
Did you see that "differential power analysis attack" was underlined with a link to Paul Kocher paper? https://www.paulkocher.com/doc/DifferentialPowerAnalysis.pdf
272.
by mratsim 6y ago
Switzerland is very welcoming
273.
by mratsim 6y ago
Reinforcement learning uses Monte Carlo a lot but traditional machine learning or deep learning don't.
274.
by mratsim 6y ago
Reencoding will cumulate H264 artifacts with H265 artifacts (and psychovisual optimization for one with psychovisual optimization for the other). Unless you reencode from source don't do that.
275.
by mratsim 6y ago
And then we have a nuclear incident because "everything we did worked for 30 years so we didn't do maintenance".
276.
by mratsim 6y ago
I would expect that it's actually a requirement and that you could get fined for that in an inspection. 146 people died in club in France due to inwards doors in 1970: https://en.wikipedia.org/wiki/Club_Cinq-Sept_f
277.
by mratsim 6y ago
In big C++ project, your number one problem to solve is likely C++ build times and how it eats your dev productivity.
278.
by mratsim 6y ago
Are you aware of D libraries in that vein? I'm especially interested in the first 2, since those are something I'm actively working on at the moment.
279.
by mratsim 6y ago
Everyone stands on the shoulder of giants. Some stands on the shoulders of TPUs that do 1 century worth of compute in a day. Other stands on the shoulders of C or Python libraries. Also, is engineering "just a glorified wrapper around
280.
by mratsim 6y ago
It depends. On current CPUs the bottlenecks are in order: - IO processing - Memory - CPU In particular, any language that doesn't have any escape hatch to do manual memory management will have troubles to optimize memory bound workload
281.
by mratsim 6y ago
> Syntax wise Nim it is a step back. It's hard to read and understand while any C/C++ dev will have next to no effort reading through D code. And of course while D is C ABI compliant it interoperates with C++ well too. On the o
282.
by mratsim 6y ago
The biggest sponsor of Nim Status: - https://github.com/status-im?q=&type=&language=nim It's used by Beamdog for tooling for Neverwinter Enhanced Editions: - https://github.com/niv/neverwin
283.
by mratsim 6y ago
The CUDA kernels are stored here: - https://github.com/mratsim/Arraymancer/blob/master/src/tenso... With the higher order function: - https://github.com/mratsim/Arraymancer&#x2F
284.
by mratsim 6y ago
kotiert seems surprisingly like "cotée" for "société cotée en Bourse" in French so I wouldn't be surprised if it were the Swiss that used that term.
285.
by mratsim 6y ago
I think it was a jibe at Kubernetes and "cloud" in general.
286.
by mratsim 6y ago
checksum can tell you if there is a difference but not what caused it.
287.
by mratsim 6y ago
Zoom is free so you're the product.
288.
by mratsim 6y ago
Even if you pay, software are packed jammed with telemetry that reports on every single thing you do (Windows 10, Amazon Alexa, Siri ...) under the pretext of "convenience" and "improving experience".
289.
by mratsim 6y ago
Germany's Financial Authority is having a bad week.
290.
by mratsim 6y ago
It's even worse than that. Assume you want to train an AI to recognize shops or buildings for example for a Google car. Well if you do it in the US with skyscraper, in Europe, in Africa, Middle-East or Asia, you will get completely dif
291.
by mratsim 6y ago
For proving distributed computing protocol soundness, I'm a big fan of formal verification and model checking, in particular TLA+[1] by Leslie Lamport who is behind the Lamport clocks referenced in the paper and curiously Latex, and th
292.
by mratsim 6y ago
2-Phase-Commit protocol i.e. a scheme with "prepare-commit" and eventual rollback.
293.
by mratsim 6y ago
Looks very interesting but ... AVX512 is already problematic cooling wise, this seems even worse.
294.
by mratsim 6y ago
BLAS libraries, oneDNN, OpenCV, Eigen, Tensorflow, PyTorch, LLVM MLIR, ... AMD usually implements them but with a couple of years of delay. For example AVX512 is not implemented, and we had to wait for Ryzen 3 to have the same AVX capabilit
295.
by mratsim 6y ago
Author here, This is a guest blog post on exploring Nim features: - Speed - Correctness via modelling physics units, tracing side-effects and mutability - C/C++ interop and how they can be applied to a domain with a critical need for a
296.
Ray Tracing in Nim(nim-lang.org)
6 points by mratsim 6y ago|1 comments
297.
by mratsim 6y ago
Not having a package manager for your language means you're back to CMake and the "FindBoost", "FindCuda", "FindMKL", scripts that always fail for one reason or the other. It encourages reinventing the whe
298.
by mratsim 6y ago
Seems like they learned from Trump and the change in visa/immigration to distract the public.
299.
by mratsim 6y ago
If you replace platform by country, this logic doesn't work though.
300.
by mratsim 6y ago
You sparked my curiosity, do you have examples you can share?

More ›