YYGML.h patch to prioretize AMD/Nvidia GPUs over iGPU

[inland rampart]

Hi,
I've followed this article ( https://help.yoyogames.com/hc/en-us/articles/360006506718-How-to-make-Windows-YYC-games-prefer-the-AMD-Nvidia-GPU-in-PCs-with-multiple-graphics-chips )
on how to modify the YYGML header in the runtime to prioritize AMD/Nvidia over igpus. First time doing it, not so familiar with the syntax.

copied and pasted the provided code snippet after the #include's as instructed.

I run into build errors on the current runtime (2023.6.0.137) with a cleared cache and restarted system.

Has anyone else run into this, or knows a solution? is the article outdated?

Attaching Build output log excerpt.

[cunning forge]

lemme ask you a different question

why are you implementing this patch?

[inland rampart]

cus my current build is not the most elegantly optimized piece of software ever created and requires more vram than what my laptops igpu can handle.

Short term make the build use a proper gpu for to make it run well while testing on my laptop + on testers machines without changing system settings or adding flags into the exe manually.

Long term, maybe an escape, I'd rather have PC run hot at 90FPS than cool at 10fps. I've got non-pixelart texture atlases and I don't know how much I can cut down VRAM use to realistically make it run as smooth as I'd like at the resolutions I want while using the igpu.

[cunning forge]

have you run into meaningful slowdown for real world users?

[inland rampart]

No large testing group as of yet but I've tested on my own laptop + partners laptop (fairly competent laptop can play last gen 3D titles well).
Since the bottleneck on my laptop is the igpu I assume there are other cases where the same would be the case. I don't want to run into a situation where someone who can run Cyberpunk at 100 gets 15fps cus it runs on the igpu.

[cunning forge]

(you don't need to ping me with every reply)

[inland rampart]

sorry, I'll stop with that.

[cunning forge]

first up i'd report the iGPU issue to YYG

if you've got a project + device that reproduces this issue then they'll wanna know

secondly, i wonder why that's throwing errors ...

GM i think still uses LLVM

it's tripping up on the weak keyword so hmm

[inland rampart]

Yeah even if I optimize the rendering to the point where it runs well on the igpu (which probably is doable) It very much irks me that it doesn't work

[raw oyster]

Remove the __attribute__((weak)) from each line... not sure why that is there

I will get that removed from the FAQ