raincole

When interacting with LLMs it feels a lot more like I'm just receiving knowledge passively

It's... almost an oxymoron, and very different from my experience.

? In this very particular case, yeah, the original commenter was holding it wrong, as they inspected the wrong thing. People sometimes hold things wrong and it's okay.

So what's the best real time alternative to SSAO?

Except cloud services won over local-first apps.

Put all of these trends together, and I think that in 10-15 years, we are going to have consumer PCs (and phones!) doing

I'm not even sure in 10-15 years whether we're still going to have consumer PCs, or PCs at all.

I suppose that's the main reason as Houdini's real customers are those who can afford huge render farms. Even before Karma, Houdini's renderer (Mantra) wasn't known for its speed.

That's quite a bummer for me, an indie user. With the foundation Houdini has established, they probably could've released a real time viewport composition before Blender. But at the end of the day it doesn't make a lot of business sense for SideFX to care about users like me.

It's a weird way to frame the event. Yeah, Andrew (Zig)'s responses were astonishingly immature, but it doesn't matter. Even if Andrew literally ate baby seals for breakfast and kicked puppies as a hobby, it wouldn't make Bun look better. You'd notice that the parent comment didn't even mention zig at all.

It (3d asset creation) practially hasn't changed much since ~1995.

I disagree. In the past ~20 years, more and more 3D assets have been created as very high-poly models and high-res textures first (via sculpting, photogrammetry and baking) and the optimization process was pushed to much later stages. I'm quite sure it was not the case in 1995.

It's just like programming. We do have better tools and abstraction layers. But Jevons paradox got us.

Or perhaps I don't understand how things were done in 1995 as much as I believe?

As someone who spent quite some time (>1000hr) in Houdini, I really wonder what's the secret behind Blender's rendering engines and why SideFX failed to catch up. Cycles are so much faster than Karma on my machine. Eevee can get impressive results in real time while Houdini's real time viewport is bare bone.

This is the correct approach (I dare to say the only correct approach) when porting to another language. You can examining the architecture later.

I literally use vim. But if I can customize something system-wise instead of app-wise I'd always do that.

Only slightly related, but what increased my code typing speed the most is this (AutoHotKey 2):

; CapsLock + J/K -> ( )

CapsLock & j::SendText "("

CapsLock & k::SendText ")"

; CapsLock + U/I -> [ ]

CapsLock & u::SendText "["

CapsLock & i::SendText "]"

; CapsLock + L/: -> { }

CapsLock & l::SendText "{"

; SC027 = physical ;/: key on a US keyboard

CapsLock & SC027::SendText "}"

; CapsLock + O/P/M/E -> _ + - =

CapsLock & o::SendText "_"

CapsLock & p::SendText "+"

CapsLock & m::SendText "-"

CapsLock & e::SendText "="

I was about 50% sure this whole piece is parody when I read through the LLM part, and after I read the whole thing now I'm 99% sure.

the weights of the model—which you have to sell many million times over before you turn a profit—easily fits on contemporary pocket-sized storage devices.

Which model is this author talking about? Which pocket-sized devices? Where can I get them? No one is using Gemma 4 to find cybersecurity issues.

Edit: there are a lot of sentences that I can't distinguish from sarcasm in this article. I guess I read it too seriously.

It's such a weird train of thoughts lol. You're using the fact that

- Claude isn't doing that

as evidence to support the assumption that

- it's a marketing trick

Which is obviously non sequitur, as if it were a marketing trick, Anthropic could do it too. Anthropic isn't known for not spending on marketing.

Honestly, nowadays I question human's reasoning ability more than I question AI's.

It genuinely scares me that some people's first reaction to this news is banning LLM.

When I realized that I'd spent more time on explaining that my question wasn't the same as another existing one than asking the question itself, I stopped using SO all together.

To compete against America. If your country has something like DeepSeek you really can't afford to let it fall as it's your best leverage if the US government decides to ban companies in your country from accessing American LLMs. And this is why there will never be a "DeepSeek of the US."

So what's your unique contribution to this world - is it the prompt you once wrote

Why not? You're implying writing good prompts doesn't require effort or thought, which is false to me.

Are you going to apply that knowledge of physics or computer science yourself, or will you just prompt the LLM again?

To write better prompt is a good reason to learn new things. It's so obvious that more knowledge and experience you have, the better results you get from LLMs.

There was an almost exact example of this: City Skyline 2. It's (was) a poorly optimized game and they said Unity's DOTS didn't match their expectation, making Unity look bad, while some issues (like they didn't use LoD where they should have) weren't Unity's fault.

And even in this case, it'd be extremely weird if Unity published a blog post about how City Skyline's studio is a "Just a total shit show, from an employment perspective" (quoting Andrew word by word here).

I talked to those who interviewed for a job at Oven. I talked to people who worked there. Those people talked to each other. Everybody talked to everybody. The grapevine was large and healthy and full of juicy grapes, and all those grapes contained the juice of the same message: Jarred was a stinky manager. Poor communication, unrealistic expectations, low empathy, no experience. Just a total shit show, from an employment perspective

Yeah, exact words you would expect from someone who is happy about a win-win situation.

Past college you're not going to learn physics by trying to self-study an entire university course

And why not? Because humans' brains stop working after they graduate from colleges?

Most college students don't work that hard.

It reads as "I don't use Claude Code."

Yeah, exactly. It's weird that Zig even responded to that. Imagining that your studio switched from Unity to Unreal and Unity proceeded to release a hit piece attacking your codebase quality and workplace environment.

So not just free work, but you expect Anthropic to pay to maintain a Zig version...?

Opportunity cost isn't just a real cost. It's THE real cost.

Or, you know, you can use Postgres. It's right there for you.

Reminder: the "expected" result is replaced all programmers and white collar employees. Even they don't necessarily state it explicitly, that's what is in their minds.

Anything less than that is slower than expected.

It's unironically a good practice when you port from an unsafe language (C/Zig) to Rust. Porting isn't refactoring. One should keep the logic mapping one-to-one as much as possible.

The high number of unsafe blocks is a good sign.

"Sometimes I feel like I'm an athlete who trained all day for years, then before my first professional match, I retired and went off to teach PE."

It might be true in some very special cases, but if a programmer can't make games like Crab Champions (the example used in the grandparent comment) performant with Unity/UE, they should be nowhere near to a custom engine.