HNHacker News
TopNewBestAskShowJobs

Ono-Sendai

839 karma · joined September 6, 2012

my blog: http://www.forwardscattering.org/
submissionscomments
Ono-Sendai··on Fixing the iterative damping interpolation in video games
Proposed solution is too expensive with ln and exp calls.
Ono-Sendai··on Wuffs: Wrangling Untrusted File Formats Safely
Wuffs is great. I use it in Substrata (https://substrata.info/) for loading PNGs. It is both faster and safer than LibPNG. It's something around 2x faster than LibPNG in my tests (depending on the PNG file), see timings here: https://github.com/google/wuffs/issues/13#issuecomment-17325...

So generally Wuffs is great and you should use it to decode your PNGs. There are some downsides: not all of the obscure bit depths and formats that PNG supports are loaded as-is, some are converted to more standard formats.

Also the Wuffs documentation is a bit hard to understand. It's a litle bit of a mission getting PNG decoding working. You can see my code for that here though: https://github.com/glaretechnologies/glare-core/blob/2c7174c...

Ono-Sendai··on It's always TCP_NODELAY
From my blog > 10 years ago but sadly still relevant: "Sockets should have a flushHint() API call.": https://forwardscattering.org/post/3
Ono-Sendai··on Fast, simple, hard real time allocator for Rust
I have a best-fit allocator for this problem (managing GPU resources). It has higher CPU overhead due to using a std::map, but should waste less space due to not using fixed bucket sizes.

https://github.com/glaretechnologies/glare-core/blob/master/... https://github.com/glaretechnologies/glare-core/blob/master/...

Ono-Sendai··on Equinox.space
Nah probably buggy Mac OpenGL drivers
Ono-Sendai··on How Perfectly Can Reality Be Simulated?
Not only that, we can't even simulate a single nucleon
Ono-Sendai··on Chrome Feature: ZSTD Content-Encoding
I implemented this in my server (https://substrata.info/). Works great. Results in smaller files than zlib/deflate. One thing to watch out for - compression levels >= 20 aren't handled by chrome clients.
Ono-Sendai··on LLaMA now goes faster on CPUs
Yeah the code looks like a spinlock. It behaves terribly under contention, resulting in performance falling off a cliff as the number of threads increases. Adding more threads actually slows down the total performance.

I would fix it if I could be bothered. Instead I will just use the Cuda whisper backend which is pretty nice and fast.

Ono-Sendai··on LLaMA now goes faster on CPUs
Multithreading support in llama.cpp is probably still pretty busted, assuming it uses the same underlying NN inference code as whisper.cpp: https://github.com/ggerganov/whisper.cpp/issues/200#issuecom...
Ono-Sendai··on Kolmogorov Complexity and Compression Distance (2023)
Practically speaking, trying to compress a file is a nice way of measuring... something. I was more talking about the theoretical notion of complexity.
Ono-Sendai··on Kolmogorov Complexity and Compression Distance (2023)
Kolmogorov Complexity does not help with giving a universal measure of complexity or randomness: https://forwardscattering.org/page/0
Ono-Sendai··on Airfoil
Why is it wrong?
Ono-Sendai··on Inside the proton, the ‘most complicated thing you could possibly imagine’
For all the simulation argument believers, note that we can't even accurately simulate a single nucleus.
Ono-Sendai··on Orbit: C/C++ Performance Profiler
Sanitizers only sometimes work on Windows too :)
Ono-Sendai··on Orbit: C/C++ Performance Profiler
Superluminal is great as well. I currently prefer Tracy because it gives faster profiling feedback.
Ono-Sendai··on Orbit: C/C++ Performance Profiler
There's a bunch of really good C++ profilers already. My current favourite is Tracy: https://github.com/wolfpld/tracy
Ono-Sendai··on Croquet: Live, network-transparent 3D gaming
I think Substrata delivers on all of those things, although it could be faster to download the initial data with a CDN.
Ono-Sendai··on Croquet: Live, network-transparent 3D gaming
Hi John, Substrata is already open source (well, 99% open source): https://github.com/glaretechnologies/substrata

As far as scaling goes, fancy scaling solutions are currently in the 'future-problems' basket. Currently Substrata just uses a naive client-server solution with O(n^2) messaging where n is the number of connected clients - e.g. each message from a client is sent to all other clients. I'm relying on the performance of C++ to manage despite this. I've tested with a few hundred concurrent clients.

Cheers, Nick

Ono-Sendai··on Croquet: Live, network-transparent 3D gaming
Hi, thanks! I have added a licence file to it (MIT licence).
Ono-Sendai··on Croquet: Live, network-transparent 3D gaming
Have a look at https://github.com/glaretechnologies/substrata/blob/ef48de33... for some (incomplete) notes.
Ono-Sendai··on Croquet: Live, network-transparent 3D gaming
Here's my open source metaverse, which unlike Croquet has an open-source server as well: https://github.com/glaretechnologies/substrata

It's out there live, you can visit the world right now via https://substrata.info/

It can do networked physics, see https://youtu.be/_rECxiwVteY, https://youtu.be/CzGz6voUE_8 etc.

Ono-Sendai··on Intel CEO: 'The entire industry is motivated to eliminate the CUDA market'
OpenCL was a perfectly good alternative to CUDA. Unfortunately due to the short-sightedness of programmers using the proprietary CUDA, here we are with CUDA dominant.
Ono-Sendai··on Unaligned accesses in C/C++: what, why and solutions to do it properly
Have been dealing with this issue recently. memcpy was quite slow in my tests however, it was ~twice as fast to check the alignment and load as floats (was dealing with floats in my case).
Ono-Sendai··on Not Everyone Agreed with Einstein
It is pretty hard to measure length contraction when all your measuring rulers also contract, and it's pretty hard to measure time dilation when all your clocks run slow, yes. That is the conundrum of being embedded in the universe :)
Ono-Sendai··on Not Everyone Agreed with Einstein
Glad you like it.

I haven't gone through any of the maths in detail. I do however have a semi-classical electrodynamics computer simulation that I occasionally work on, so why do the maths when you can just run a simulation :)

When I get back to it maybe I will try and simulate Compton scattering or the photoelectric effect.

Ono-Sendai··on Not Everyone Agreed with Einstein
Oh apparently it was Schrodinger himself that wrote about a semiclassical Compton scattering approach:

http://www.regels.org/Compton-effect.htm

Ono-Sendai··on Not Everyone Agreed with Einstein
Not sure what association you mean. Time dilation? length contraction? Both are well supported by evidence AFAIK.
Ono-Sendai··on Not Everyone Agreed with Einstein
Suprise: i have done first year physics as well. And I am probably more familiar with light-atom interaction than you.
Ono-Sendai··on Not Everyone Agreed with Einstein
You can have vacuum fluctuations without photons.

EDIT: will try and find Compton paper later, going for walk now.

Ono-Sendai··on Not Everyone Agreed with Einstein
Time dilation is a function of your speed through the ether.
← PreviousPage 4 of 22Next →