Rarebox
Joined 13 years ago
Karma
139
Submissions
43
Member since
Apr 6
Recent submissions
Comment · 8 months ago
If one lidar hits another, it will result in at most one bad reading (perhaps a bad column?). This can likely be filtered, or a bad scan (360deg) can be altogether rejected and the data predicted using models based on past sensor readings.
Comment · 11 months ago
That's a pretty good example. The summary is actually useful, yet it still annoys me.
But I'm not usually reading the comments to learn, it's just entertainment (=distraction). And similar to images or videos, I find human-created content more entertaining.
One thing to make such posts more palatable could be if the poster added some contribution of their own. In particular, they could state whether the AI summary is accurate according to their understanding.
Comment · 1 year ago
Interesting example! I've been learning AVX512 by using it to optimize Huffman coding. I found _mm512_permutexvar_epi8 and used it to do byte-indexed lookups, but _mm512_permutex2var_epi8 means I can get by with 2 shuffles and 1 comparison instead of 4 shuffles and 3 comparisons. In the end, on my CPU (AMD 9950x), changing to _mm512_permutex2var_epi8 only sped up compression by ~2%.
Compared to Huff0[1] (used by Zstd), my AVX512 code is currently ~40% faster at both compression and decompression. This requires using 32 datastreams instead of 4 used by Huff0.
Comment · 1 year ago
Also just stopped reading at that point. The idea seemed quite clever.
Comment · 1 year ago
The peak of human civilization, before robots took over.
Comment · 2 years ago
I could be wrong but I think many times the researchers don't care about the exact function. It could be something like 1/log(log(n)) .
Comment · 2 years ago
https://yuri.is/not-julia/ is a good write-up of one person's opinion on the problems of Julia. I'm much less experienced with Julia but I somewhat agree. There's too much focus on "magic" instead of correctness for me to try building serious software using Julia. An amazing language in many aspects though.
Comment · 3 years ago
He says he'll port his performance optimizations to the original game once he's done with his game / romhack. Otherwise he'd have to always update two codebases when he finds a new optimization.
Comment · 3 years ago
Back then I remember switching to Chrome for these reasons:
1. Feeling faster than firefox (still does)
2. Process separation for tabs. Firefox used to crash the whole browser, whereas chrome would crash just a single tab.
3. Shipped with support for Flash and PDFs
Comment · 3 years ago
Exactly. It could even be turtles all the way down, with new building blocks of physics becoming relevant as we go smaller and smaller (and back in time).
Comment · 3 years ago
Would that be morally close to torturing an equally intelligent real worm?
Comment · 3 years ago
It's jarring initially, but becomes natural very quickly. Writing loops like "for i in 1:length(arr) ... end" is pretty neat compared to C++ or even python. Plus in math sequences typically start at index 1.
Comment · 3 years ago
It is nitpicky, but the fact that these classes deal with decision problems and not general problems is something which feels like should be mentioned more often.
Of course this ends up not mattering in practice since we can convert a problem with arbitrary (bitstring) output into a decision problem. To do this we just include a index n in the input and make the decision problem about whether the bit at given index n is 1.
Comment · 3 years ago
This would make sense right? You probably can't run all cores at max turbo frequency before hitting power and thermal limits. So dividing chiplets based on binning would result in good performance for 1-8 -threaded tasks while reducing chip lottery.
(But it's possible there are other reasons for the discrepancy)
Comment · 4 years ago
I thought google did this versioning thing for libraries before, but it was stopped for reasonable reasons (g3 components).
Basically if you could pin lib versions everyone would be stuck on old versions for a long time, causing difficult debugging work for each user of the library. You'd then also have all sorts of diamond problems: what if you want newest absl but older bigtable client?
It's a difficult problem no matter which way you go.