igpu Running Doom Eternal on 2gb iGPU on Ultra Nightmare.

MINEJHAZZ · July 21, 2022

So, just like what the title says, I've apparently managed to run Doom Eternal on Ultra Nightmare graphics at 1080p while only having a Ryzen 5 5600g 2gb VRAM while even streaming it to some friends on discord at 720p 30fps

Also before you ask and here's the thing that's making me more confused is that my fps is at 35-45 fps

As I've mentioned, I've allocated 2gb of my ram(2gb/24gb) as a vram for my iGPU and doom apparently thinks that I have 13gb(12,991gb to be exact) of vram installed on my system.

So, can I ask any experts or tech savvy people out there and explain to me what's happening here? is it a visual bug on doom or did doom eternal decided that it can just use my extra ram as my vram instead?

My PC Specs if you need to know

MB: B450 Aorus elite v2

CPU: Ryzen 5 5600g

RAM: 3x 8gb Kingston HyperX Fury 3600Mhz XMP

PSU: Corsair CV550 - 550w 80+ bronze

Cooling:

Zoomflow 240x AIO Cooler(CPU Cooler)[the 2x 240mm fans are set to Exhaust)

1x cheap 120mm fan(Exhaust)

3x Rakk Maris Pro 120mm 2,000RPM Fan(Intake)

OC:

CPU: 1.35v, Clock: 3,900

IGPU GFX: normal, Clock: 2,300

~~don't question it on why I'm overclocking my clocks on an APU xD~~

Levent · July 21, 2022

36 minutes ago, MINEJHAZZ said:

can I ask any experts or tech savvy people out there and explain to me what's happening here?

Simple, windows is just using system RAM to allocate more than available dedicated vram to system ram. This happens on every single system with or without dGPU.

Spoiler

Microsoft makes it easy to monitor it. This was introduced with Vista (IIRC) and makes a big difference even though people dont notice it.

Red is vram space the GPU has (In your case, your GPU has 2G of RAM allocated to it even when it doesnt use it all).

Green is what windows can spare from system memory (I believe there is a ratio here but I am not sure) to allocate to GPU in case VRAM is not sufficient. Purple is the total size of RAM and VRAM your system can allocate to GPU at best.