Const-me

28 ms·

361.
by Const-me 3y ago
> if you live in a city you have to be a masochist to use them I think in the mountains it’s the opposite. I wouldn’t want to drive 1km of elevation down without using the engine to control the speed. Brakes in most cars aren’t designed
362.
by Const-me 3y ago
Interestingly, I have a similar project with C# for the higher-level pieces, and Direct3D 11 GPU API instead of Vulkan: https://github.com/Const-me/Cgml > don't see many other people using Vulkan for GPGPU, and
363.
by Const-me 3y ago
If you just want to run Mistral on Windows, you could try my port: https://github.com/Const-me/Cgml/tree/master/Mistral/Mistral... The setup is relatively easy: install .NET runtime, download 4.5 GB
364.
by Const-me 3y ago
If you’re right and no such evidence exists, seems the original article is an entirely false accusation? On your new topic, I doubt Microsoft would risk their multi-billion/year cloud business complying with illegal requests from the U
365.
by Const-me 3y ago
Microsoft has multiple data centers online within EU: Ireland, Netherlands, France, Germany, Italy, Poland, and Sweden. Is there any evidence of the transfers of personal data outside the EU?
366.
by Const-me 3y ago
> Hybrid fuel-cell vehicles are a great alternative to BEVs, but only if we can get the production, storage and transport of hydrogen solved. There’re fuel cells which oxidize propane instead of hydrogen, here’s an overview https:/
367.
by Const-me 3y ago
Some of these old optimizations are now deprecated. For example, there’s a famous trick for inverse square root: https://en.wikipedia.org/wiki/Fast_inverse_square_root Modern processors have a special instruction for t
368.
by Const-me 3y ago
> in a CPU or GPU, who cares, moving around the data is 100x more expensive than the ALU operation Moving data is indeed expensive, but there’s another reason to not care. Modern CPUs take same time to add or multiply floats. For example
369.
by Const-me 3y ago
> In what way does it look bad? For example, on my 27” 4k display with 150% DPI scaling, I can’t easily distinguish commas from dots. I’ve been using displays like that for a decade now. Aptos is the only font with that issue.
370.
by Const-me 3y ago
I would want an equivalent of cublas optimized for my specific GPU model and implementing the same API. AFAIK cublas and other first-party libraries are hand-optimized by nVidia for different generations of their hardware, with dynamic disp
371.
by Const-me 3y ago
C++ std::mutex provided in Microsoft's STL uses SRWLocks underneath, same size as a pointer. Sadly, SRWLocks have yet another bug, they are very unfair. Write a loop which locks std::mutex inside the body, and no other thread will be a
372.
by Const-me 3y ago
It’s not terribly hard to detect whether a specific page contains any ads. Blacklist any pages with ads, and you remove the incentive to pollute the directory. The internets worked that way for quite a while, before modern big tech invaded.
373.
by Const-me 3y ago
> Obviously the bottom right corner of a rectangle cannot be 10,10 if the top left corner is 0,0 What you said is only true for a mathematician. In mathematical conventions, Y axis is directed upwards. For programmers however, Y axis is
374.
by Const-me 3y ago
Interesting bug, but I don’t like the fix. Most computers are 64 bits for decades now. It takes same resources to add/subtract/compare int64 compared to int32. No reason to use 32-bit integers for time, when I write C++ I use GetT
375.
by Const-me 3y ago
Speaking about random sampling during inference, most ML models are doing it rather inefficiently. Here’s a better way: https://github.com/Const-me/Cgml/blob/master/Readme.md#rando... My HLSL is easily p
376.
by Const-me 3y ago
I agree in general, but note some industries are exceptions from that rule. For example, for Boeing and Airbus these two scales are pretty close because modern airliners are very expensive per unit, and fixed R&D and tooling costs are s
377.
by Const-me 3y ago
Here’s a commercially available high resolution DLP 3D printer people use to make molds for casting gold and silver parts: https://www.junction3d.com/ There are downsides, too: printing is slow, and the build area is small,
378.
by Const-me 3y ago
> error, which is an unobserved, incorrect internal state For example, amount of available space on the system drive is not an internal state. However, once the number reaches zero, failures of all software are very likely to happen. The
379.
by Const-me 3y ago
You could try my Mistral implementation: https://github.com/Const-me/Cgml/blob/master/Mistral/Mistral...
380.
by Const-me 3y ago
I did a few times with Direct3D 11 compute shaders. Here’s an open-source example: https://github.com/Const-me/Cgml Pretty sure Vulkan gonna work equally well, at the very least there’s an open source DXVK project whic
381.
by Const-me 3y ago
> an example of C++ being safer than unsafe Rust? Not really, that’s just my impression reading stuff about unsafe rust, and programming C++. > wouldn’t your comparison need you to take into account unsafe constructs used to implement
382.
by Const-me 3y ago
Sometimes you don’t want libraries, instead you want to implement similar stuff in your own code. And Rayon uses unsafe to workround the compiler limitation of the safe rust I was talking about. Also, it seems Rayon is substantially slower
383.
by Const-me 3y ago
Different elements of the output vector may take very different time to compute. If you do parallelization with split_at_mut API, you won’t be able to saturate all cores because the thread who does the splitting can’t possibly know how much
384.
by Const-me 3y ago
I prefer different strategy. I use C++ for encapsulated cases where it makes sense. I compile that C++ into DLLs, and consume these DLLs from C#. C++ is safer than unsafe Rust. C++ was designed for usability of the unsafe code because the e
385.
by Const-me 3y ago
For example, safe rust forbids code which writes to different elements of the same vector from different CPU cores. C++ compiler has no objections, and doing that is often the best way to parallelize computations.
386.
by Const-me 3y ago
The logical error you just made is called “post hoc fallacy” https://en.wikipedia.org/wiki/Post_hoc_ergo_propter_hoc The reason Yatsenyuk became a prime minister, on February 27, 2014 the parliament of Ukraine made him
387.
by Const-me 3y ago
Are you referring to that leaked call? https://www.bbc.com/news/world-europe-26079957 If the leak is true, the people in the call are doing precisely what they publicly said they were doing — trying to negotiate a peac
388.
by Const-me 3y ago
You’re telling the false version of these events spread by the Russian state propaganda. > US backed a [coup, revolution, insurrection - whatever you want to call it] in Ukraine People of Ukraine did the revolution. The western backing y
389.
by Const-me 3y ago
> GPUs that support multiplane overlay on windows According to Microsoft, MPO is an optional feature: https://learn.microsoft.com/en-us/windows-hardware/drivers/d... According to random people on the inter
390.
by Const-me 3y ago
> Yes, they do. Why are you so certain? > you may be using an older nvidia gpu. Tested AMD Vega 7 inside Ryzen 5 5600U, and nVidia 1080Ti — dxcapsviewer.exe shows the same result, no overlays are supported. Only the D3DCURSORCAPS_COLO

More ›