harrison_clarke

2 bulan yang laludiscuss

i haven't used the openai voice thing

but, if it's trying to respond in a natural way, with interruptions in both directions, it may still be a good idea. if there's a delay between you stopping and it starting talking, it feels weird

(you might be able to fake some of that on the client, but then you need a thicker client)

harrison_clarke

12 bulan yang laludiscuss

there's a psychological bonus for heftier things

maybe if it was larger, thicker, and a more dense material. most of those matter more to the person holding it, rather than an observer, though

harrison_clarke

12 bulan yang laludiscuss

i think it's a mix of conservative/religious lobbying (getting them ire from the government), and chargebacks by embarrassed customers with post-nut clarity being common

harrison_clarke

12 bulan yang laludiscuss

and even if you do want to carry that much cash, surely you'd want a fatter wad with smaller bills, right?

harrison_clarke

tahun laludiscuss

i think the thing that got me was that people wrote articles comparing this thing to a laptop GPU, and i didn't read far enough to notice. normally, people compare desktop stuff to desktop stuff, and laptop stuff to laptop stuff

i haven't bought a laptop gpu since ~2008. from what i remember, they used to at least put "mobility" in the name, or throw and 'm' after the number, or something

harrison_clarke

tahun laludiscuss

well, that's a hostile naming scheme

from the specs, a mobile 4070 looks like its basically a 4060 ti, underclocked to ~60%

harrison_clarke

tahun laludiscuss

a quick search told me that this thing's GPU (8060s) is comparable to a 4070

from what i can tell, a 4070 alone is ~$750. this thing is apparently $799 for the CPU+GPU+RAM+motherboard all soldered together ($300 more for case+PSU+SSD)

if the CPU+RAM+motherboard are costing me $50, i don't really care if i have to throw them out along with the GPU

harrison_clarke

tahun laludiscuss

would have been cool to know about this a month ago

i just built a mini ITX gaming PC for a friend, and this one looks pretty good for quality/$. good enough that i wouldn't be surprised if these get snapped up and re-sold for more than the sticker price

i think it makes more sense to think of it as a high-end console, given that basically everything is soldered, though

harrison_clarke

2 tahun yang laludiscuss

you can avoid resizing and moving the pools if you allocate massive pools up front. most OSs let you overcommit memory, and most programs using memory pools only have a handful of them

(on windows, you'd need to handle this with VirtualAlloc. osx and linux let you overcommit with malloc, and handle it with a copy-on-write page fault handler)

harrison_clarke

2 tahun yang laludiscuss

why would you need to iterate on destruction? you free the whole pool at once, since you allocate it all at once

harrison_clarke

2 tahun yang laludiscuss

biased locking can be faster than spinlocking, in the uncontended case

unbiasing/re-biasing a lock can be quite heavy, though. JVM did it by pausing threads (or waiting for GC to do so)

harrison_clarke

2 tahun yang laludiscuss

game engine architecture, by jason gregory

there's a section on memory allocation patterns

i believe casey muratori talks about allocation patterns in the handmade hero video series, too

edit: ryan fleury has a talk: https://www.rfleury.com/p/enter-the-arena-talk

harrison_clarke

2 tahun yang laludiscuss

vulkan doesn't have global state, and the error handling is better

but it's not batteries-included, and that's often to be a deciding factor at small scales

i think if you're going to dabble in engine dev, you pick which one you want depending on which part of the engine you find interesting. if you want to make a game, you pick up unity or godot or something

harrison_clarke

2 tahun yang laludiscuss

there's a common gamedev practice of allocating a big chunk of memory up front, and then using a bump allocator inside of it

in most games, there are about 3 "lifetimes": - permanent/startup - per-level - per-frame

and they're nested. so, you can use a single stack allocator for all of them. at the end of the frame/level, pop back to where it started

there are more complicated patterns, but this one will get you pretty far. you can use it on the CPU and the GPU

harrison_clarke

2 tahun yang laludiscuss

the biggest part for me is the shader compiler. opengl has one built in, vulkan requires me to pull in yet another dependency

i've heard that vulkan allows bindless textures now, so the descriptor nonsense is a bit less awful that it used to be

vulkan is appealing, but there's a high initial cost that i don't want to pay

comments