mini

3.4K posts

mini banner
mini

mini

@minitbnn

Katılım Ocak 2010
468 Takip Edilen129 Takipçiler
mini
mini@minitbnn·
@stuartseupaul @RefinedLurk @ericdfoley @FrankieIsLost The go to for processing a 5 GB file: 1. Write instructions for CPU using a programming language. 2. Make CPU execute the instructions upon the 5 GB file. If you need AWS for any of this you have lost the plot.
English
1
0
1
34
frankie
frankie@FrankieIsLost·
people who haven’t worked in big tech don’t understand the scale of wasteful spending that happens in these orgs. during the golden age of microservices, you had websites that could have run on a single box and a postgres instance being run across hundreds of services, all massively overprovisioned. it got so bad that companies created entire teams whose job was to harass devs into lowering their cloud spend. token spend can still get much dumber than you think
English
28
28
1.1K
47.9K
Joshua Kibbey
Joshua Kibbey@KibbeyJoshua·
@Jonathan_Blow @chamath Because each rule is tied to a spot on the code. You look at the code and verify it was interpreted correctly
English
4
0
0
1K
Chamath Palihapitiya
Chamath Palihapitiya@chamath·
Here is the first case study of what 8090 can do. Working with CMS, our Enterprise team decoded a 50yr, 18M repo of COBOL and Assembly that governs billions in healthcare payments. We documented 100,000+ business rules in plain English using Software Factory, each traced to its exact source line. The people who own Medicare policy can now read how their systems behave and change as they see fit. Full story in the thread below. We can do this for any large enterprise who have big, legacy codebases.
8090@8090_Factory

The United States Medicare claims system: 18 million lines of Assembly and COBOL, 50 years of policy logic. Working with CMS, our Enterprise team read the code and documented 100,000+ business rules in plain English using Software Factory, each traced to its exact source line. The people who own Medicare policy can now read how their systems behave. Full story in the thread.

English
119
141
2K
362.6K
mini
mini@minitbnn·
@_shadyjoo @muellerberndt Bernhard Mueller has LLM psychosis and believes he has discovered the true nature of the universe.
English
0
0
0
5
Bernhard Mueller
Bernhard Mueller@muellerberndt·
Disclosure day: The shortest proof I could come up with that we are, in fact, inhabiting a simulation. Hadronic contributions are assumed due to lack of supercomputer to run QCD sims on. Ignore what LLMs and community notes say. Use your brain 🤯 github.com/FloatingPragma…
English
128
139
1.5K
9.7M
mini
mini@minitbnn·
@SheriefFYI SSAO, RTAO, SSR and other ray-spawning/denoising often needs normals and ddx/ddy on depth is often not enough.
English
0
0
6
224
Sherief, FYI
Sherief, FYI@SheriefFYI·
@minitbnn what post effects need full g-buffers versus just color, depth, and motion vectors?
English
2
0
1
505
Sherief, FYI
Sherief, FYI@SheriefFYI·
one of my hottest takes is that deferred rendering's benefits flip sharply once you cross a certain resolution threshold as bandwidth consumption begins to choke your performance and games should have explored things like clustered forward rendering as we approach 4K.
English
16
2
106
10.2K
mini
mini@minitbnn·
#Box3D voxel destruction. The API is very nice, impact sounds were easy to implement. I had to lower the sub-tick count to 1 for this scene. Will get better once i topologically optimize the static voxel collision meshes.
English
0
0
2
105
mini
mini@minitbnn·
@samth @lexi_lambda If Andrew is mad about Zig not being memory safe then... why did he design it to be a non-memory safe language? Does he regret that choice? Please tell me more about Andrew's inner thoughts.
English
0
0
0
45
Sam Tobin-Hochstadt
Sam Tobin-Hochstadt@samth·
@lexi_lambda I really don't think the attitude of the Bun people is relevant here. My claim is that Kelley is mad about the idea of Rust being safer than Zig, which is explicitly a reason given in the Bun blog post.
English
2
0
11
869
mini retweetledi
DarknessFX
DarknessFX@DrkFX·
Videos from the High-Performance Graphics 2025 conference are available on their YouTube channel: @HighPerformanceGraphics/videos" target="_blank" rel="nofollow noopener">youtube.com/@HighPerforman…
DarknessFX tweet media
English
1
37
117
11.9K
mini
mini@minitbnn·
@beshr @ngxson Did you miss the "below €150" part? What kind of startup can't spend €150 on prototyping lmfao?
English
0
0
0
44
Beshr
Beshr@beshr·
@ngxson they think startups don't scale in EU because of fragmentation, but in reality it's shit like this that kills any genuine/creative ideas from even starting
English
2
0
14
367
mini
mini@minitbnn·
@Jonathan_Blow Mfw they can't even solve the problem of having to get into a tube with a bunch of demons
mini tweet media
English
0
0
13
1.4K
Jonathan Blow
Jonathan Blow@Jonathan_Blow·
Kind of crazy how passenger air travel was invented by a lost civilization, stopped advancing in the 1970s and the best we can do these days is just kind of copy the Old Ways, but never improve them... ... ...
English
55
17
914
64.2K
mini
mini@minitbnn·
@BlenderDev Genuinely curious why Windows ARM got left out. Not enough testers? Not a single graphics programmer in the world with access to an ARM machine? (Most likely)
English
0
0
1
21
Blender Developers 🔶
Blender Developers 🔶@BlenderDev·
Congratulations to everyone involved. Huge milestone! Dear community, it's your time to test and report. #b3d
Blender Developers 🔶 tweet media
English
2
5
70
3.8K
mini
mini@minitbnn·
@NOTimothyLottes Just yield the thread? I know OS sleeping primitives are horribly inaccurate but surely you can sprinkle in a few until deadline gets close?
English
1
0
0
18
NOTimothyLottes
NOTimothyLottes@NOTimothyLottes·
Delaying work for latency reduction post interrupt typically means utilizing a busy spin that waits until the target time is reached. Which is a horrible place to be at in terms of power.
English
2
0
2
367
NOTimothyLottes
NOTimothyLottes@NOTimothyLottes·
One challenge of optimizing for latency in both GPUs and audio, is the desire to delay work until just before it is needed. And this conflicts with interrupt placement typically being at the point of worst latency.
English
1
0
13
1.4K
mini retweetledi
Dennis Gustafsson
Dennis Gustafsson@voxagonlabs·
Not sure how useful, but you can't argue with the satisfaction
English
75
82
1.8K
138K
mini
mini@minitbnn·
ZXX
0
0
0
52
mini
mini@minitbnn·
@SebAaltonen I have seen 1 renderer in my life that used callbacks for customization, but yes it's bad. If a renderer uses callbacks it's doing too much / is too authorative.
English
1
0
0
635
Sebastian Aaltonen
Sebastian Aaltonen@SebAaltonen·
Many renderers have callbacks. Ours does not. WASM->JS callback would be horribly slow. If you want to run custom code in the rendering, you write a compute shader. Compute shader has access to all the same (interpolated) engine/camera globals as render passes. It can calculate/interpolate data and write to any UBO/SSBO you bind into your materials. Our plan is to be a fully GPU-driven renderer. CPU doesn't even know what is visible. CPU callbacks wouldn't do much good in a modern architecture like this. Custom setup code has to be a compute shader.
English
10
10
253
16.6K
mini
mini@minitbnn·
@munsplit The keyword here is "development". Yes, it makes development easier. That doesn't mean you have to ship the same technique to players. Use RTX to accelerate in-editor real-time lightmap preview, but bake for production. Like Valve did for HL:A.
English
2
0
10
848
munsplit (developing something)
munsplit (developing something)@munsplit·
Non gamedev people have no idea how much easier lumen or rtgi makes development, and i say that as someone who decided not to use it. an average person would fucking crash out if they found out what it takes to make a game with good lighting in old fashioned ways.
Reece “Kiwi Talkz” Reilly@kiwitalkz

If you have been wondering why UE5 games on Switch 2 are still few and far between, it's because devs have been having trouble with the engine on Switch 2 It's largely been alleviated now though, this update will be a big help.

English
45
29
521
45.8K
mini retweetledi
Anshu
Anshu@anshuc·
@github holy shit, how did the attackers find a large enough uptime window to get in?
English
176
694
14.4K
960K

Keşfet

@stuartseupaul @RefinedLurk @ericdfoley @FrankieIsLost @KibbeyJoshua @Jonathan_Blow @chamath @_shadyjoo