mxmlnkn

19 ms·

91.
by mxmlnkn 3y ago
There also is this: https://github.com/mxmlnkn/pragzip I did some benchmarks on some really beefy machines with 128 cores and was able to reach almost 20 GB/s decompression bandwidth.
92.
by mxmlnkn 3y ago
From my perspective the reconstruction of lingual thoughts is the new one. There already were reconstruction of thought images several years ago and recently using diffusion models: https://www.biorxiv.org/content/10.11
93.
by mxmlnkn 3y ago
This is basically the same reason why I started with ratarmount ( https://github.com/mxmlnkn/ratarmount ) but the focus was more on runtime performance and random access and as the name suggests it started out with acces
94.
by mxmlnkn 3y ago
https://github.com/rhasspy/piper I was confused as well
95.
by mxmlnkn 3y ago
Slow downloads wouldn't be a problem if shipping actually finished games on Blu-rays were still in vogue. And the disc limitation would make some publishers think twice about shipping 250 GB on 3 or more discs.
96.
by mxmlnkn 3y ago
Might be related to p-norms although it doesn't really fit: https://raw.githubusercontent.com/mxmlnkn/fft-image-experime... Assuming it is a circle under a p-norm, I calculated p to be 3.504, so because of errors
97.
by mxmlnkn 3y ago
Interesting question! You can try it out easily by varying the N parameter in the script. I uploaded the resulting images to the Readme in my fork: https://github.com/mxmlnkn/fft-image-experiments#scaling-the... To sum
98.
by mxmlnkn 3y ago
Many JSON parsers require the whole JSON file to be in memory. Parsers that allow parsing in streaming manner are the exception. I only found jsoncons to work reliable but at the cost of speed compared to something fast like rapidjson. And
99.
by mxmlnkn 4y ago
No, as far as I know, the pigz blocks, not to be confused with deflate blocks, are still compressed by referring to the preceding uncompressed data even if it belongs to another pigz block. Therefore errors would still propagate indefinitel
100.
by mxmlnkn 4y ago
Decompression of arbitrary gzip files can be parallelized with pragzip: https://github.com/mxmlnkn/pragzip
101.
by mxmlnkn 4y ago
Since ChatGPT I've become much more aware of my own thoughts and written text. I'm now often wondering whether I'm just regurgitating the most frequently used next word or phrase or whether it could actually be described as o
102.
by mxmlnkn 4y ago
I observed this phenomenon for myself these days when using igzip. Depending on the file, it can be up to 2-5x faster than gzip for decompressing files even though neither is multi-threaded. I had to compute a checksum on the decompressed f
103.
by mxmlnkn 4y ago
I know the graphs but I still find it amazing to see that this 20+ years old system runs with approximately the same CPU core frequency as today's hardware. The amount of RAM wouldn't suffice today, though.
104.
by mxmlnkn 4y ago
This is a match made in heaven, which makes ChatGPT actually useful for factual data. Or the inverse, it makes Wolfram Alpha even more accessible. That example screenshot of ChatGPT generating three queries to Wolfram Alpha in succession in
105.
by mxmlnkn 4y ago
There are walkthroughs like this on the internet: https://www.ign.com/wikis/pokemon-red-blue-yellow-version/Pa...
106.
by mxmlnkn 4y ago
I love the ReadMe of your project. It is like the other extreme when compared with Geogram. I tried to find out what Geogram was about and what it does better but I lost interest before I could find out. The ReadMe needs some work and the w
107.
by mxmlnkn 4y ago
I see the advantages of bundled websites masquerading as desktop application in regards to platform support. But, is it really worth to have applications that take up 1 GB just for editing text or just for chatting? If it was only one such
108.
by mxmlnkn 4y ago
Any thoughts on the Sensirion SCD41? It is almost half the price. https://sensirion.com/products/catalog/SCD41/ https://www.digikey.com/en/products/detail/sensirion-ag/SCD4
109.
by mxmlnkn 4y ago
That would only apply to repositories. But to train these models, you need hundreds of terabytes of diverse data from the internet. Up until now a relatively straight-forward scraper would yield "pristine" non-AI-generated content
110.
by mxmlnkn 4y ago
That's also the solution I use. Especially because I have multi-rows set up in Firefox and each update breaks this. Plus, when installing the binaries manually, you can also apply some "hidden" settings by creating a distribu
111.
by mxmlnkn 4y ago
You could parse awp1.json and awp2.json to get all links.
112.
by mxmlnkn 4y ago
I would also prefer unsigned indexes but exactly because the compiler may assume that there will be no overflow, signed index access may be a bit faster and therefore preferable.
113.
by mxmlnkn 4y ago
At the very least you are redundantly executing logic without the exception. The check for eof has to be done implicitly anyway inside read because it has to fill the bit buffer with data from the byte buffer or the byte buffer with data fr
114.
by mxmlnkn 4y ago
That particular code came from a bit reader class. So, if you are reading only a few bits per call, then the read call becomes very cheap. It might be as "simple" as this pseudocode if ( offset + requestedBitCount < 64 ) {
115.
by mxmlnkn 4y ago
It is interesting to read how the compiler behaves when using instructions but the number of generated instructions is only one metric. As always, there is nothing better than measuring the runtime of your actual implementation. I'm wr
116.
by mxmlnkn 4y ago
Interesting read. Especially the lookup method based on partitioning. I tried to implement a similar reverse image search based on dHash as explained here https://github.com/Rayraegah/dhash . However, I also had lookup
117.
by mxmlnkn 4y ago
igzip is even faster. See also my answer that contains a quick benchmark in that linked question at the bottom.
118.
by mxmlnkn 4y ago
I recently implemented pragzip for parallel gzip decompression https://github.com/mxmlnkn/pragzip . I would be interested to know how your PR back then worked to parallelize the decompression.
119.
by mxmlnkn 4y ago
Pragzip actually decompress in parallel and also access at random. I did a Show HN here: https://news.ycombinator.com/item?id=32366959 indexed_gzip https://github.com/pauldmccarthy/indexed_gzip can als
120.
by mxmlnkn 4y ago
I have been working on exactly that: pragzip. It decompresses in parallel in a similar manner to the prototype pugz by implementing a two-staged decompression. https://github.com/mxmlnkn/pragzip I also did a Show HN:

More ›