To me it's very clear that the icons have that "stable diffusion trying to make pixel art" style. I think this needs an extra layer of code that gets the generated image and turns it into actual pixel art
Incredibly fast, on my 5090 with CUDA 13 (& the latest diffusers, xformers, transformers, etc...), 9 samplig steps and the "Tongyi-MAI/Z-Image-Turbo" model I get:
It stays around 26Gb at 512x512. I still haven't profiled the execution or looked much into the details of the architecture but I would assume it trades off memory for speed by creating caches for each inference step
Super interesting project, at first I thought it would be a naive implementation of YOLO but I wasn't aware about retro-reflections. The papers he linked in the GH discuss very interesting ideas
Unfortunately, that last 10% of games are AAA competitive multiplayer that account for a massive user base who are still dependent on windows to play them (battlefield 6, fortnite, any of the call of duty games from the last 8 years, league of legends, GTA online, apex legends, rainbow six siege...)
yeah, and mac too, can't run league or valorant. vanguard is their kernel-level anticheat, and windows is like 95% of their market and the difficulty of implementing it on another kernel i guess isn't worth the <5%.
League works on macOS just fine, I played yesterday. Vanguard is buggy (it occasionally quits the client after I finish a game), but the game generally works and has for at least several years.
Why would it be a "bold move"? Linux gaming population is damn near zero, they do not provide a higher profit margin like mac gamers would, and the documented evidence is that supporting Linux users is obnoxious because they are rude and entitled but not actually that much better at providing feedback.
Epic Games bought out rocket league and turned off a native linux build and faced no repercussions. Instead they made plenty of money.
Not sure that's fair, given most Linux gamers look like Windows gamers to the metrics.
That said, devices like the SteamDeck run games on Linux (and that's without considering that every Android game ever is technically running on Linux too).
Let's face it though, PC gaming is already small enough these vs the consoles, that further splitting the marked isn't going to make sense for a lot of companies.
>Not sure that's fair, given most Linux gamers look like Windows gamers to the metrics.
No. All the articles and testimony of game devs abandoning native Linux versions is from well before Proton was a thing, including Epic Games buying Rocket League and preventing you from playing the Native Linux build they had.
It also was not related to anti-cheat or underlying engine limitations or anything. Developers were clear that the problem was the massive lack of uptake mixed with a weirdly entitled community.
Personally I don't think gamers are entitled. Ultimately games are anywhere between 60 - 120 dollars and often barely work on their target platforms. With kernel level anticheat, you're literally being asked to pay them to rootkit your computer with software you cannot audit.
The last 10 years of AAA gaming have been an absolute shit show. The only people who seem to be even trying are Nintendo. Everyone else releases stuff that's buggy as hell and about as fun as a dental cleaning.
It's bold because it's breaking stuff that already works and will continue to work even if you do nothing.
It's one thing to choose not to develop a new game for Linux. It's another to take a game that already runs on Linux and intentionally break it. You're guaranteed to alienate SOME people who are already fans of the game.
I thought the same but let's not pretend Windows is a holy grail for compatibility anymore. Especially when it comes to older games, this facade / image breaks down fast.
I once tried to play Trackmania Nations (not Forever or United Forever, the ESWC one) because that was the first entry of this series I played. I still have all the files from back then so I thought it would be as simple as installing it and running it. Other games such as Trackmania Sunrise came with the nasty SecureROM DRM that will break your current installs, but ESWC was always free to play and without DRM.
Well, after install, I played a lot in my first sitting. A few days later, my Windows install was broken. I used a restore point before installing Trackmania, everything was back to stable. A few weeks later I tried again, same situation, a day or two after install, my Windows would break.
I thought it was a general system instability, maybe some weird configuration and the game only triggers that specific bug. So I did a full clean reinstall. And installed the game a few days later. Who would've thought, my Windows breaks yet again.
What I'm trying to say is: I've been running Fedora on my main PC for 2 years now and the game has been installed via Proton for 1 year. It never broke, it always just worked.
That's cool but I spent the last week trying to get midi music in dosbox under Mint. It's still not working. Midi.
And Wine works until it suddenly doesn't and searching for solution you get stonewalled with modern day equivalent of rtfms or plain old radio silence.
Thats always the worst part of linux for me. Everyone is always so hostile, I have to say though I have had a little success finding help on lemmy but not much.
To add to this Nathan Briggs does reverse engineering to make old games work on modern Windows. Windows 11 has corrected faults in it's APIs that probably should have been fixed but somehow worked with older versions of Windows that gamedevs built around. He often posts the solution and sends it GOG. Often this involves updating a community maintained wrapper around the DirectX APIs that GOG uses.
Also, really the holy grail of YouTubers — he’s not concerned with clicks or retention. He has a day job, he’s not trying to do this full time or make a career out of it. I’ve been following him for years. Legitimately just an insanely smart person doing projects on his own time he thinks are interesting and that he thinks other people might be interested in too. It’s kinda crazy when it’s as simple as that with no ulterior motives or incentives.
All three (GB10, GB200 and GB300) are part of the Blackwell family, which means they have Compute Capability >= 10.X. You could potentially develop kernels to optimize MoE inference (given the large available unified memory, 128Gb, it makes the most sense to me) with CUDA >= 12.9 then ship the fatbins to the "big boys". As many people have pointed out across the thread, the spark doesn't really has the best perf/$, it's rather a small portable platform for experimentation and development
I wonder how this will affect the large cinema production companies (Disney, WB, Universal, Sony, Paramount, 20th century...). The global film market share was estimated to be 100B in 2023. If the production cost of high FX movies like Avengers Infinity War goes down from 300M$ to just 10K$ in a couple of years, will companies like Disney restrain themselves to just release a few epic movies per year? Or will we be flooded with tons of slop? If this kind of AI content keeps getting better, how will movies sustain our attention and feel 'special'? Will people not care if an actor is AI or real?
