DOOM's Vulkan update is now live

I noticed a similar stuttering as well on my system. Hopefully the next Nvidia driver addresses this.
NVIDIA addressed the issue as being real and that they found the issue. As of yet no time is set for when it will be fixed. The issue being latency and that it affects some users not all.
 
Vulkan woke up the Nano in 4K! With 30% powertune added. With Vulkan I am playing over 40fps easily up into the low 50's all FreeSync on Ultra settings TSSAA8. Previously it was mid 20's to high 30's. Very good experience. This has to be the highest performance patch I've ever seen in any game.
 
NVIDIA addressed the issue as being real and that they found the issue. As of yet no time is set for when it will be fixed. The issue being latency and that it affects some users not all.

That's good to hear. According to this article Nvidia is releasing a driver update later this week to fix some VR related issues. Maybe it will contain some fixes for Vulkan Doom as well?

Nvidia GeForce GTX 1080 Doesn’t Boost In VR With The Latest Game Ready Driver (Update)
 
Just tested in this room, using PresentMon
pL9lnkw.jpg

Here are my results

OGL:
AVG, MIN, MAX
99.7913288314, 76.5520936998, 148.6325802616

VK:
AVG, MIN, MAX
105.4843487499, 79.1514959633, 190.6577693041


1440p max settings (nightmare), 5820k@ 4.3, 980ti @ 1450/7400
 
I can't remember which ones but seem some maxwells had worse issues as well but again sporadic. Seems to be a new topic so may take a day or 2 to get a clearer picture.
 
I played briefly yesterday with my 1080 at 4k resolution and I saw a drop. At one point I went from 45fps in an area to 39...
 
Stupid question. Is there an option in the settings tab to change back and forth between Vulcan and opengl? Haven't gotten home yet from work. Like to see if this helps me out any
 
Stupid question. Is there an option in the settings tab to change back and forth between Vulcan and opengl? Haven't gotten home yet from work. Like to see if this helps me out any

Yes

Or change r_renderAPI in the console.

It requires a game restart though.
 
Just tested in this room, using PresentMon
pL9lnkw.jpg

Here are my results

OGL:
AVG, MIN, MAX
99.7913288314, 76.5520936998, 148.6325802616

VK:
AVG, MIN, MAX
105.4843487499, 79.1514959633, 190.6577693041


1440p max settings (nightmare), 5820k@ 4.3, 980ti @ 1450/7400

Can you try different shadow setting and AA please.
Was that with TSSAA or other?
Anyway worth testing with different AA and shadow just to see how it boosts.
Thanks
 
I wonder what's different?

I seem to only get a solid performance boost, no stuttering.
What is your setting in terms of shadow detail and also AA?
MIght be worth the others to try that as well.

Regarding hardware, mind saying what Nvidia drivers you are using and what your CPU is.
Also Win10 or earlier?
Thanks
 
Yeah 8xTsaa

Heres the first proper doom vulkan review, look at that FX go

Doom mit Vulkan: Benchmarks mit AMD und Nvidia
Thanks,
so mind trying with one of the other AA and also maybe shadow setting.
Sorry to be a pain but might be worth it :)

Problem is that some are seeing noteable improvements with Maxwell 2 and Pascal, and then other reviews like theirs and Guru3d have next to zero changes for Nvidia..
So either something is influencing this in terms of settings, CPU,and possibly exacerbating it Windows environment and whether Win10/8/7.
Cheers
 
Thanks,
so mind trying with one of the other AA and also maybe shadow setting.
Sorry to be a pain but might be worth it :)

Problem is that some are seeing noteable improvements with Maxwell 2 and Pascal, and then other reviews like theirs and Guru3d have next to zero changes for Nvidia..
So either something is influencing this in terms of settings, CPU,and possibly exacerbating it Windows environment and whether Win10/8/7.
Cheers

So you want me to lower shadow settings? Should I use smaa? I'll do another run later tonight no problem
 
Now I look at those results again, 1440p fury x result seems off
 
Yeah 8xTsaa

Heres the first proper doom vulkan review, look at that FX go

Doom mit Vulkan: Benchmarks mit AMD und Nvidia

More proof that AMD's drivers are terrible :(

I mean, the scaling falloff clearly points to partially being CPU-limited (ignoring the Fury X being ram-limited at 4k). The majority improvement still comes form async though.

But there is potential here for more DX12/Vulkan games, so we'll see how developers take to these new APIs over the next year.
 
More proof that AMD's drivers are terrible :(

I mean, the scaling falloff clearly points to partially being CPU-limited (ignoring the Fury X being ram-limited at 4k). The majority improvement still comes form async though.

But there is potential here for more DX12/Vulkan games, so we'll see how developers take to these new APIs over the next year.
How do you know the gain comes from async?

I'd say the gain comes from not relying on their shoddy ogl driver mainly.
 
So you want me to lower shadow settings? Should I use smaa? I'll do another run later tonight no problem
Yeah keep shadow as is to begin with and use SMAA and measure.
Then change the shadow setting to one lower.
Thanks
 
I want to preface by saying that I am running Windows Insider Preview just because I wanted to see how the new Microsoft Edge handles extensions. So my results might be different from everyone else's.

Just playing the game @1080p and I noticed that Fullscreen and Borderless made a difference for me. At the part of the game where I just got a second upgrade for my shotgun where I can switch between the 2 buffs. Vulkan seems to maintain it's frame rate better than OpenGL as it didn't seem to dip. OpenGL would dip sometimes especially for the red tinged screen before you die and the death animation. I didn't notice it under Vulkan. At these frame rates I need to record game play to determine if what is going on.

Borderless:

Under OpenGL my frame rate is in the 110's - 120's. but would dip to 80's for say a death animation.

Under Vulkan my frame rate is in the 110's - 120's all day; everyday it seemed. Need to record game play to see if it dips as I can't tell at 120 fps. Sorry!

Fullscreen:

Under OpenGL my frame rate is 86 fps with dips.

Under Vulkan my frame rate is 86 fps without perceived dips. Again it is hard to tell if you are experiencing a frame dip at 86 fps.


Maybe if you're not getting the frame rates that everyone else is try Borderless?
 
i'm going to test this tonight:

i'm on the latest insider preview of win10
dual e5 2670s - holding back my titan i know
titan x @ 1430mhz
128GB ram
 
So from what I understood the spir-v shader output in this vulkan version includes the use of AMD's shader intrinsics, so I guess those parts don't go through a shader compiler?
 
How do you know the gain comes from async?

I'd say the gain comes from not relying on their shoddy ogl driver mainly.

True, I was just pointing out the cpu throughput (which falls with resolution) versus actual card improvements. We can't boil it down to async unless there's a way to turn async on and off.
 
True, I was just pointing out the cpu throughput (which falls with resolution) versus actual card improvements. We can't boil it down to async unless there's a way to turn async on and off.

There is! Turning AA to SMAA should do it. The perf data from computerbase.de is just strange. Apparently they say at 4K it's not a VRAM issue, and I don't understand why it's 20% faster than a Ti at 1080p, then 30% faster at 1440p. Seems odd.
 
Great to see the developers taking time to optimize for vulkan. Is this game fun to play? I tried the demo and it was pretty cool but thought it was a bit repetitive looking
 
Just tested in this room, using PresentMon
pL9lnkw.jpg

Here are my results

1440p max settings (nightmare), 5820k@ 4.3, 980ti @ 1450/7400

Nightmare settings + TSSAA

OGL:
AVG, MIN, MAX
99.7913288314, 76.5520936998, 148.6325802616

VK:
AVG, MIN, MAX
105.4843487499, 79.1514959633, 190.6577693041

UPDATED VULKAN RUNTIME RESULT FROM HERE ON:

VK(only updated runtime):
AVG, MIN, MAX
111.8795795755 84.7960654626 169.1188905801

VK:(borderless)
AVG, MIN, MAX

110.9528607854 83.8012234979 179.6945193172


Edit:

SMAA Ultra settings

OGL:

AVG, MIN, MAX
105.2419932021 81.9067900729 145.6664238893

VK: (borderless fullscreen)

AVG, MIN, MAX
106.1324120595 80.0897004645 149.7454327643

VK: (fullscreen)

AVG, MIN, MAX
102.6833263994 80.1089481695 142.4501424501


My settings:

 
Last edited:
Great to see the developers taking time to optimize for vulkan. Is this game fun to play? I tried the demo and it was pretty cool but thought it was a bit repetitive looking

The actual game is massively busier than the demo. Think about the room with the gore nest at the start of the demo, double or triple the number of enemies and imagine then they're all different types doing different shit.

The full game is often chaotic as fuck.
 
Great to see the developers taking time to optimize for vulkan. Is this game fun to play? I tried the demo and it was pretty cool but thought it was a bit repetitive looking

If you are at all a fan of the old Doom games the Doom 2016 is definetly a GOTY contender. Ultra Violence difficulty is the only true way to go (outside of Nightmare challenge), so much action packed fun and challenging as hell (no pun intended) after the first few levels.



Anyway I did some testing now. Vulkan is slower than OpenGL with my old GTX770. With the newer Vulkan SDK installed too. Thats a bummer... So it was the faster and more ram + ssd that made the previous difference, not the vulkan which actually slowed things down in my rig.
 
If you are at all a fan of the old Doom games the Doom 2016 is definetly a GOTY contender. Ultra Violence difficulty is the only true way to go (outside of Nightmare challenge), so much action packed fun and challenging as hell (no pun intended) after the first few levels.



Anyway I did some testing now. Vulkan is slower than OpenGL with my old GTX770. With the newer Vulkan SDK installed too. Thats a bummer... So it was the faster and more ram + ssd that made the previous difference, not the vulkan which actually slowed things down in my rig.

Well I don't think Vulkan is supposed to perform better per se (cpu bottlenecks aside)
 
Now I look at those results again, 1440p fury x result seems off

I got the same 50% avg increase on Fury X as computerbase @ 1600p resolution so why it feels off? Heck, it was even 100% more on some situations.

I think this is the first time we see Fury X perform on the level it should if we look at raw tflops value.
 
  • Like
Reactions: noko
like this
Back
Top