gmrrrk.bsky.social
he/him, 24, software engineer, c++, CUDA, rust, compression & lowlevel stuff enthusiast
18 posts
13 followers
91 following
Getting Started
Conversation Starter
comment in response to
post
there is a really good book about this behavior among white ppl: White Fragility. It's rather short but it explains well why we see this sort of reactions
comment in response to
post
yeahhh, I guess you have all of them 😅
comment in response to
post
you can't malloc inside function that is meant to be asynchronous (for obvious reasons)
comment in response to
post
some async C API I worked with:
If a function needed some heap memory for calculations then we had a function that returned the required size, user allocated one blob of memory and then passed that to async fn in which was used reinterpret cast to split one pointer to big heap space to multiple ptrs
comment in response to
post
it is not a sin to want something you like without an animal getting hurt
comment in response to
post
yeah, funningly enough in the OBS source code it is only used in twitch default settings. This is a tricky feature. Basically you need to synchronize two independent media streams and when streamer is dropping lots of frames then it really gets complicated
comment in response to
post
Here is another approach en.m.wikipedia.org/wiki/Pyramid... It is a method mainly used in signal quantization but also gives a nice point in sphere distribution
comment in response to
post
this but its the DeepSeek whale
comment in response to
post
pass ball by reference/value
comment in response to
post
wow that Babel game looks great
comment in response to
post
that's so deep
comment in response to
post
for now it is offline. My main comparison target was just jpeg for now and check it manually to see which looks better/worse. MSE or other metrics don't really resemble the human eye difference error. I heard about Simulacra benchmarks but was curious what was your approach to this problem.
comment in response to
post
So cool!! I'm rolling out my own codec that I plan to do also on GPU. It will be a part of my masters thesis and I looked at ways I can compare with other solutions :) Do you have some hints related to the spark report? Like which codecs did you compare, which image sets worked best?
comment in response to
post
I went from Manjaro to NixOS last year and have had 0 issues so far. I love having my whole config on GH so I can clone it on other devices if needed 😇 And if something goes wrong I can just pick the last derivation from GRUB menu.
comment in response to
post
my elbows hurt like hell when I try to do that longer than 5 minutes
comment in response to
post
👋
comment in response to
post
woohoo, way easier solution works. time to clean this and make it into actual GL_QUADS