PVS by my understanding will only ever over-count visible polys.
It is essentially a set of all polygons that are visible from any point inside a fixed volume, but the camera only exists at a single point inside that volume so there will probably be some polys that the camera has no LOS (though I suspect these would still be 'rendered') to and a bunch that are out of the view frustum which will not be rendered.
edit: To observe this you can also load any HL1 engine game, run `r_speeds 1` in the console, then it will show you how many world polys are currently being drawn in the corner of your screen, which is probably the count referenced by John Romero here.
The most recent integration I've seen is is OpenMW, which is an open source re-implementation of the Morrowind game engine. Basically it is built on the assumption that people are going to make mods that do a ridiculous amount of number-crunching in lua so any small improvement to performance is welcome.
At 4 bits that model won't fit into 128GB so you're spilling over into swap which kills performance. I've gotten great results out of glm-4.5-air which is 4.5 distilled down to 110B params which can fit nicely at 8 bits or maybe 6 if you want a little more ram left over.
GPT-oss-120B was also completely failing for me, until someone on reddit pointed out that you need to pass back in the reasoning tokens when generating a response. One way to do this is described here:
Once I did that it started functioning extremely well, and it's the main model I use for my homemade agents.
Many LLM libraries/services/frontends don't pass these reasoning tokens back to the model correctly, which is why people complain about this model so much. It also highlights the importance of rolling these things yourself and understanding what's going on under the hood, because there's so many broken implementations floating around.
It starts with the fundamentals of how backpropagation works then advances to building a few simple models and ends with building a GPT-2 clone. It won't taech you everything about AI models but it gives you a solid foundation for branching out.
AIUI they intend to retire support for x86 macOS apps in a few years, but Rosetta will remain as a low-level component so that things like Crossover and Parallels can continue to work. Maybe not forever, but there's no immediate threat of it being EOL'ed.
> Rosetta was designed to make the transition to Apple silicon easier, and we plan to make it available for the next two major macOS releases – through macOS 27 – as a general-purpose tool for Intel apps to help developers complete the migration of their apps. Beyond this timeframe, we will keep a subset of Rosetta functionality aimed at supporting older unmaintained gaming titles, that rely on Intel-based frameworks.
You guys remember when you bought a computer and could run the software you wanted, independent of political motives? In perpetuity? Reading excuses like this makes me feel validated for cutting macOS out of my professional workflow. The concept of paying Apple to provide high-quality long term support only works if Apple does better than the free offerings. Free offerings that still run 32-bit libraries, run CUDA drivers and other things Apple arbitrarily flipped the switch on.
I'm not sure what you are referring to, but I remember way less cross-platform software than we have now, and way worse working WINE. No, there was never time when we could run whatever software we want on a machine of our choice.
> I remember way less cross-platform software than we have now
Really? Outside Electron apps and PWAs, I'm seeing fewer apps than ever support macOS as a native target. Additionally, cross-platform packaging feels much more fragile than it used to, especially if you're using Brew over Nix. And cross-platform games... just forget about it.
Modern macOS simply feels abandoned by cross-platform efforts. Upstream Wine runs worse than it did in 2010, depreciated 32-bit libraries annihilated my Mac-native Steam catalog and AU plugins, Vulkan is ignored and CUDA compute drivers work but Apple refuses to sign them. The professional experience that I attributed to macOS is gone in the new releases. All Apple can innovate in is petty politicking.
i’m not sure how end-of-life it will actually be because rosetta is used in apple/container and seems to be a large part of the virtualization stuff apple’s built in the last few years
I'm replying to you from Windows 11 IoT Enterprise LTSC courtesy of massgravel (or massgrave... not sure wth it's actually called now!) and it's activated until 2038.
The only thing it didn't have out of the box that I wanted was Microsoft Store (so that I could install Winget and Terminal) but you install it from an elevated powershell command with "wsreset -i" and that's it done.
It also has the original version of Notepad, not that abomination with the tabs and Copilot!
Oh, no Copilot whatsoever in fact.
All the instructions for IoT (including where to get it... legitimately) are on the massgrave github page and website.
And before I am accused of sailing the high seas... I'm not! The activation script just activates complicated processes built-in to Windows: it doesn't "hack" it or anything!
I moved all my home LAN Windows machines to LTSC IoT in February; cost me about 90 euros for each license. You can buy individual licenses from online stores that will connect to MS and validate correctly. You'll have to install the MS app store from GitHub (!), and there are some other issues, but at least you're years away from what hit everyone else this October.
You can find some licenses sold online; it costs about 3x the price of Home. But I am not sure if it's legal; I have already bought some and then realized it's just keygenerated.
Normal, reputable websites never sell single LTSC licenses. So go figure
It shows nothing. Normal users dont even get the option. They probably dont give a fuck, based on a ton of other things, but there is no option to even choose the no bloat option.
Yes the security of every Windows computer was much better then, any software that automatically updates itself without user consent is obviously a massive security risk because the user is no longer in control of what software they run.
I would think it is pretty silly if I needed some sort of verification to drive people I personally know around because other people were getting their car hijacked after choosing to pick up strangers they found on the highway.
No need if you use the IoT version and the massgrave activation script. It uses the built-in activation mechanisms in Windows to activate until 2038 or something.
I'm using the Windows 11 Enterprise IoT LTSC with activation until 2038 right now.
The https://massgrave.dev activation scripts have support for various versions and also for enabling Extended Updates (ESU), but this is a bit off-topic and HN mods might come after us ;-)
I'm not sure it is off-topic. The parent to this one was talking about, presumably, dodgy Windows keys (that shouldn't be allowed) but from my understanding, the massgrave scripts just use internal Windows mechanisms to activate.
It's not actively usurping Windows security.
In fact, I've read more than once that Microsoft tech support have been known to use massgrave scripts to help with activation-related issues with clients: Although I should caveat that with saying that it may have been Reddit I found that info so pinch of salt and all that...
So, my take on this is if the massgrave scripts allow activation without breaking any laws then sobeit. I'm talking about doing stuff that, while it appears dodgy, actually just manipulates the ultra-complicated processes under the hood that Microsoft has already built into the OS.
It's like publicising the workarounds for the now-mandatory Microsoft account when installing Windows 11. These involve things like reg hacks and commands: they're already in Windows so publicise them all you want imo.
