DOOM's Vulkan update is now live

GameLifter

Gawd
Joined
Sep 4, 2014
Messages
699
I'll try this out on my 1080 as soon as I can. I'm hoping this patch will also fix the CTDs but they didn't mention it in the patch notes so I doubt it.
 

cageymaru

Fully [H]
Joined
Apr 10, 2003
Messages
20,979
Waiting for revision 2.0 to buy

Get a 3D printer, download the plans for one from Valve for free, and design your own. I think Valve even sells the PCB parts or said that they would. :)

(Of course I have tried none of this and am content with the basic controller as Valve updates it regularly.)
 

cageymaru

Fully [H]
Joined
Apr 10, 2003
Messages
20,979
Won't happen sales are awful for the steam controller and steam link, and frankly this is valve so if they do make a 2.0 expect sometime next decade.

They sold over 500,000 steam machines, controllers, and links combined. I would think most of the sales were for the controller.
 

GameLifter

Gawd
Joined
Sep 4, 2014
Messages
699
Briefly tested it on my 1080 and I didn't notice any improvements. This is at 1440p with Nightmare settings. Not a huge deal since the game already ran great before. Maybe people running it at lower resolutions will notice an improvement? Either that or Nvidia owners will have to wait for Async support to be enabled for them.
 

FrgMstr

Just Plain Mean
Staff member
Joined
May 18, 1997
Messages
52,942
I just did a quick test on my personal box here. IQ turned all the way up, at 4K.

Vulkan was 20% slower than OpenGL in the Foundry level (44fps vs 56fps), just stopping around the map at specific places and taking an FPS reading using the in-game tools. This was with SLI on. With SLI off, both showed to be equal (44fps vs 44fps).
 

MaZa

2[H]4U
Joined
Sep 21, 2008
Messages
3,442
It seems like it runs smoother with my GTX770, but then again I just upgraded yesterday to 16gb 2133mhz from 8gb 1600mhz ram and finally added a SSD drive into my rig too so it could be those affecting the speed and Vulkan is just a placebo.
 

tybert7

2[H]4U
Joined
Aug 23, 2007
Messages
2,745
Numbers are in.

New Patch Brings Vulkan support to Doom + Benchmarks

Solid improvement for people with amd gpus. Nvidia? Go pound sand, no improvement (because pascal still can't do concurrent graphics + compute in the same way as amd).

But take heart, the 1070 is still faster.

rx 480 = 77% the performance of a 1070 in vulkan doom @ 1080p, for half the price.

index.php


index.php




We'll see if future driver updates improve the gap for nvidia. without the added hardware schedulers I think their software engineers have to go in and try to micromanage performance boosts in a way amd no longer needs to when games are designed to utilize the right tools. So nvidia may be able to increase the gap again. With time. This really is a sweet change, amd always used to be the one blasted for taking longer to improve performance over launch day, now with the added man hours that may change with dx12/vulkan when the teams don't do a hatchet job and just bolt it on top of dx11 like tomb raider.
 

jwcalla

2[H]4U
Joined
Jan 19, 2011
Messages
3,628
Solid improvement for people with amd gpus. Nvidia? Go pound sand, no improvement (because pascal still can't do concurrent graphics + compute in the same way as amd).

More like because nvidia has the best OpenGL drivers in the industry.
 

TaintedSquirrel

[H]F Junkie
Joined
Aug 5, 2013
Messages
11,140
More like because nvidia has the best OpenGL drivers in the industry.
Nope, 1070 is 30% faster than the 480 in their Vulkan bench.
Looking at TPU and PCGamer's 1070 averages, it should be about 50% faster.

Vulkan allows the 480 to close performance gap with Nvidia by approx 20%.

edit: This is according to Guru3D's bench. I still see people on Reddit claiming big improvements for their Nvidia GPUs in Vulkan.
 
Last edited:

jwcalla

2[H]4U
Joined
Jan 19, 2011
Messages
3,628
Nope, 1070 is 30% faster than the 480 in their Vulkan bench.
Looking at TPU and PCGamer's 1070 averages, it should be about 50% faster.

Vulkan allows the 480 to close performance gap with Nvidia by approx 20%.

Should a 1070 really be 50% faster considering it likely has the same die size / transistors as the RX 480 (when considering that 25% of the 1070 chip is disabled)? Or does it have a 50% advantage because AMD's non-DX12 / Vulkan drivers are awful?
 

pendragon1

Extremely [H]
Joined
Oct 7, 2000
Messages
39,803
isn't doom ogl or vulkan, no dx11/12?! was hoping to retry the demo to see the vulkan side but alas its for full version only, I think...
 

tybert7

2[H]4U
Joined
Aug 23, 2007
Messages
2,745
More like because nvidia has the best OpenGL drivers in the industry.


Yes. And now, that "advantage" will boost them less and less. More and more shackles on amd performance are being removed by the day. More nvidia engineers to throw at optimizing the cesspool known as opengl 4.5, great. And now less relevant than ever for the devs looking for performance who learn to optimize it themselves for vulkan instead. Granted, the indie dev creating squirrel simulator may not bother, and I'm sure tears will be shed for not being able to see a game like that run at 120fps. But for devs worth their salt who bother to optomize for the higher end titles, shackles off guys.
 

tybert7

2[H]4U
Joined
Aug 23, 2007
Messages
2,745
Should a 1070 really be 50% faster considering it likely has the same die size / transistors as the RX 480 (when considering that 25% of the 1070 chip is disabled)? Or does it have a 50% advantage because AMD's non-DX12 / Vulkan drivers are awful?


People keep saying this, but how well does it track?


How much space on a gpu die is devoted to resources that do not scale with chip size? Is there literally zero difference performance wise from an nvidia chip designed for a 300mm die size on the same architecture that was cut down by say, 25% for an "effective" die size of 225mm compared to a full nvidia chip designed from the ground up to fill out a 225mm die size?

There is zero difference between a cut down gpu vs a full chip gpu? Or are there some extra advantages to being cut down to a similar effective die size? I don't know, asking, because this keeps being parroted as if one ought to expect perfectly linear scaling between cut down chips and full chips of equal size and presumed equivalent hardware resources.
 

jwcalla

2[H]4U
Joined
Jan 19, 2011
Messages
3,628
Yes. And now, that "advantage" will boost them less and less. More and more shackles on amd performance are being removed by the day. More nvidia engineers to throw at optimizing the cesspool known as opengl 4.5, great.

Unless a rendering scenario is CPU-limited due to threading limitations, I'm not sure why you think Vulkan would perform better than OpenGL.

Of course I've been asking this question for three years and have received nothing but hand-waving, so I'm not expecting an answer now either.
 

heatlesssun

Extremely [H]
Joined
Nov 5, 2005
Messages
44,154
Hmmm, the Vulkan isn't compatible with Flawless Widescreen that's necessary for the correct POV on a widescreen setup. I love my new rig but the combination of SLI and Surround is becoming iffy. But I love my 3 monitors. Argg.
 

jwcalla

2[H]4U
Joined
Jan 19, 2011
Messages
3,628
People keep saying this, but how well does it track?


How much space on a gpu die is devoted to resources that do not scale with chip size? Is there literally zero difference performance wise from an nvidia chip designed for a 300mm die size on the same architecture that was cut down by say, 25% for an "effective" die size of 225mm compared to a full nvidia chip designed from the ground up to fill out a 225mm die size?

There is zero difference between a cut down gpu vs a full chip gpu? Or are there some extra advantages to being cut down to a similar effective die size? I don't know, asking, because this keeps being parroted as if one ought to expect perfectly linear scaling between cut down chips and full chips of equal size and presumed equivalent hardware resources.

I don't know. I would expect a 1070 to perform significantly better, but 50% strikes me as a bit much considering the underlying physical hardware. Maybe we could draw better comparisons when 1060 benchmarks come out, and we can see what 1060 vs 1070 looks like in both performance and die size.
 

Panmaster

Limp Gawd
Joined
May 31, 2013
Messages
228
People keep saying this, but how well does it track?


How much space on a gpu die is devoted to resources that do not scale with chip size? Is there literally zero difference performance wise from an nvidia chip designed for a 300mm die size on the same architecture that was cut down by say, 25% for an "effective" die size of 225mm compared to a full nvidia chip designed from the ground up to fill out a 225mm die size?

There is zero difference between a cut down gpu vs a full chip gpu? Or are there some extra advantages to being cut down to a similar effective die size? I don't know, asking, because this keeps being parroted as if one ought to expect perfectly linear scaling between cut down chips and full chips of equal size and presumed equivalent hardware resources.
Performance is roughly proportional to the die-size raised to the 1/2 power.
 

CSI_PC

2[H]4U
Joined
Apr 3, 2016
Messages
2,193
Important heads-up.
To enable Async Compute you must either use TSSAA or No AA.
Anything else disables Async Compute.
youtube.com/watch?v=cjgFNkNW8zI … DOOM GL vs Vulkan on an awesome AMD 480. Heads up benchmarkers, use TSSAA or no AA (else Async Compute is disabled)

Anyone using Nvidia, will be interesting to compare using TSSAA and say one of the other AA options just to see what happens if anything (should not affect behaviour but who knows)
Those with AMD, make sure you use only those two AA options.

More generally, seems those with 980ti are seeing performance boosts with Vulkan, no idea how well the boost is for other Nvidia cards - this is separate to the Async Compute related AA as it is not yet implemented for Pascal.
Cheers
 

TaintedSquirrel

[H]F Junkie
Joined
Aug 5, 2013
Messages
11,140
Performance changes seem to be related to CPU, for Nvidia anyway. DSOGaming gained anywhere from 20-60fps switching to Vulkan.

DOOM benefits greatly from Vulkan - Vulkan versus OpenGL Performance Comparison - DSOGaming

Vulkan: http://www.dsogaming.com/wp-content/uploads/2016/07/D-Vulkan.jpg
OpenGL: http://www.dsogaming.com/wp-content/uploads/2016/07/D-OpenGL.jpg

I think we need to see comparable cards benchmarked to judge how CPU bottlenecks might be impacting fps ceiling (IE: not RX 480 vs 1070). When you're looking at excess of 150+ FPS it's not really surprising the CPU starts doing heavy lifting.
 

sleepybp

Limp Gawd
Joined
Feb 5, 2014
Messages
389
Good point, squirrel

The reviews should show useful scenarios that matter to gamers. Like a card that games @1080P going from 45 fps to 55 or 60 fps, or 1440P card going from 65 fps to 80 fps.

Useless to see if a gtx 1080 goes from 160 fps @1080p to 190 fps @ 1080p, who cares bc gtx 1080 is not a 1080p card and like you said that is just a cpu bottleneck that doesnt matter
 

Presbytier

[H]ard|Gawd
Joined
Jun 21, 2016
Messages
1,058
My GTX 1080 hit 200fps without Vulcan but dropped frequently to 120 ish with Vulcan it consistently stays higher with some frame drops. Considering the open gl implementation on AMD cards for Doom was already broken we can not assume the performance gain is indicative of a nvidia weakness, but instead should just be happy AMD is improving their pipeline.
 

GameLifter

Gawd
Joined
Sep 4, 2014
Messages
699
After more testing I still didn't see a difference on my 1080. Then it CTD'd on me again. I went back to playing it on the Xbox One, lol!
 

Ozymandias

[H]ard|Gawd
Joined
Jun 21, 2004
Messages
1,610
I ain't buying it then considering I have a 3x 980Ti setup for 4k.
Your loss. It's a fun game. I just-about-maxed it out (no Nightmare textures or AA) on a Fury at 4K so one 980Ti should be fine.
 

TaintedSquirrel

[H]F Junkie
Joined
Aug 5, 2013
Messages
11,140
WCCF has a video up comparing async on/off for the 480.
Although it involves using different forms of AA so there will be some difference.

I would also like to point out they have TWO videos on RX 480's Doom Vulkan performance and none on Nvidia GPUs. Just food for thought.

 

JustReason

razor1 is my Lover
Joined
Oct 31, 2015
Messages
2,483
My GTX 1080 hit 200fps without Vulcan but dropped frequently to 120 ish with Vulcan it consistently stays higher with some frame drops. Considering the open gl implementation on AMD cards for Doom was already broken we can not assume the performance gain is indicative of a nvidia weakness, but instead should just be happy AMD is improving their pipeline.
Sorry but try as you might you can't deny what Nvidia lacks. Now whether it impacts them greatly or not is up for discussion.
 

TheBlueChanell

Supreme [H]ardness
Joined
Apr 15, 2005
Messages
4,660
Most important thing I say was Steam Controller support! Sweet!

Played through the entire game at launch with the Steam Controller, I'd say it's about 80-90% as accurate as a KB/M using the Gyro. Good to see official support though.
 

Presbytier

[H]ard|Gawd
Joined
Jun 21, 2016
Messages
1,058
Sorry but try as you might you can't deny what Nvidia lacks. Now whether it impacts them greatly or not is up for discussion.
They say they support async compute their has been no conclusive evidence to the contrary. AMD sees such large gains because in this game the version of opengl they used on AMD cards was sub-par, who's fault that is I do not know, but we literally do not have enough evidence anywhere to suggest nvidia suffers from low level api's.
 

JustReason

razor1 is my Lover
Joined
Oct 31, 2015
Messages
2,483
They say they support async compute their has been no conclusive evidence to the contrary. AMD sees such large gains because in this game the version of opengl they used on AMD cards was sub-par, who's fault that is I do not know, but we literally do not have enough evidence anywhere to suggest nvidia suffers from low level api's.
Saying one supports a feature does not mean they support it at a hardware level. They mean, in the case of Nvidias current lineup, that they emulate the feature via software. There is plenty of proof all over the internet that proves Nvidia lacks true hardware support of asynchronous compute+graphics.
 
  • Like
Reactions: N4CR
like this
Top