DOOM (2016) performance thread

[Evildeffy]

Thanks OP for the interesting findings playing Doom 4. I didn't realize this was out yet and I'll have to grab a copy. I still remember the day I bought Doom "shareware" for 3$ on a floppy disk at the book store in my local strip mall. Oh how times have changed.

I'm curious if you are running at 1080p just as a baseline for benchmarking performance or if this is actually the resolution you intend to play the game at? I am truly not trying to be rude or inflammatory but this just feels like a very low resolution to be using given your hardware. Anecdotally in this forum I've noticed a lot of discussion around FPS and everyone always talks about running at 1080p. I guess I'm just not familiar enough with this community to know if this is intentionally downgraded to make comparisons easier or if people actually still commonly use this resolution.

Thanks again for the interesting write up. Can't wait to try this game on my new rig!

Dear Lord... floppies! Hell I actually have a floppy/card reader combo built in my PC (through the floppy mobo connector) just because I was always capable of reviving a PC from death with them even if the ports were broken or not supported in the BIOS for booting.

Also.. DOOM on floppy was ossom... but those "shareware" screens ... HATED those with a passion to say the least.

The simple answer is as thus:
Playing @ 1080p is my actual game-play resolution, if I had 1440p I would use that as a benchmark but I can tell you right now that 1440p is not happening with an R9 390X without dropping quality .. at least not on the OpenGL API, Vulkan may be a different beast but we'll see.

The rig I have currently, as you can see from the specs, is 7 years old and by today's standards very much dated. (CPU upgrade in 2011 and GPU upgrade half a year ago for 12 hours of work basically)
Personally I feel it can still cope with most I throw at it so I have never really felt the need to truly upgrade it and now that I'm unfortunately without a job (trying to find a new job though) I don't have the money to upgrade either.

F.ex. the 2 screens I'm running with are the iiyama ProLite B2712HDS and as you can see those are 6 years old as well.
1440p was almost a grand a screen then and I would've loved to have em but... $$$ is a lot.

And since I worked as a Store Manager whom had good ties with iiyama at the time I got these 2 for a good price.
They stuck with me ever since and I can tell you that whilst it's large for 27" with 1080p it is not nearly as bad as people claim.
The viewing angles are deffo better than average and no discolourations from any normal viewing angles that wouldn't make it playable anyway.

And no.. you cannot see the pixels... considering the screens I've seen ranging from A to Z I can say these are deffo the better TN panels.
I've seen a lot of 22 - 24" 1080p "quality" screens from a great deal of brands and they mostly couldn't hold a candle against these unless you involved IPS.

[Denpepe]

Well it runs at 1440p somewhere between 30-50 fps with some mix of settings, put a lot on high in the advanced options, going to wait with ultra till the new GPU's are out and I got one. Did not have the CPU info on, only put it on medium, will try that some other time.

I guess thx to G-sync everything was pretty smooth mostly, but this is pushing my GTX780 a bit much on 1440p, game is fun though. 2 things thuis far annoy me, had to pull out my x-box controller or the game would not let me start it, fallout 4 does that too, and using the scroll wheel to go trough weapons makes my browser window on the other screen move, kinda distracting.

[Evildeffy]

Well it runs at 1440p somewhere between 30-50 fps with some mix of settings, put a lot on high in the advanced options, going to wait with ultra till the new GPU's are out and I got one. Did not have the CPU info on, only put it on medium, will try that some other time.

Please post your exact specs please, would be nice to have correlation points.

I guess thx to G-sync everything was pretty smooth mostly, but this is pushing my GTX780 a bit much on 1440p, game is fun though. 2 things thuis far annoy me, had to pull out my x-box controller or the game would not let me start it, fallout 4 does that too, and using the scroll wheel to go trough weapons makes my browser window on the other screen move, kinda distracting.

That's an MS driver issue with games that support controllers.
Pull out the connector that binds your PC with the controller (wired or wireless) and it should function properly.

Haven't had that other thing happen though... nice bug

[Denpepe]

i7 6700k at 4.4 GHz, MSI z170A Gaming 5 Mobo, 16 GB 3000 mhz Corsair DDR4, Asus GTX780 DCUII, Asus PG279Q, Soundblaster x-fi fatality, win 10 on Intel 535 240 Gb SSD - Doom on OCZ Trion 150 480 GB SSD

[Evildeffy]

i7 6700k at 4.4 GHz, MSI z170A Gaming 5 Mobo, 16 GB 3000 mhz Corsair DDR4, Asus GTX780 DCUII, Asus PG279Q, Soundblaster x-fi fatality, win 10 on Intel 535 240 Gb SSD - Doom on OCZ Trion 150 480 GB SSD

Not bad.. that GTX 780 is being pushed to it's limits though, I did expect less strength from them.
Then again it could be due to nVidia's better OpenGL performance vs. AMD's worse OpenGL performance that my expectations are lowered.

This makes the Vulkan API all the more interesting to try out when it's implemented!

[Artorius]

Seems like there's a performance bug with shadows on AMD right now, lowering the shadows are giving people double the performance lol.

- - - Updated - - -

That's pretty much what I stated in a nutshell but elaborated further except that Windows can see which are real and which are fake, just you do not.
DOOM in OpenGL 4.3 API will only utilize the physical cores and not the HyperThreads.

There are no physical cores and fake cores, they're all logical cores. They act as 2 when you can make them act as two, and they act as one big core if you can't.

It's simple to understand.

[Denpepe]

Not sure you made a typo or something with your running on OpenGL4.3, but mine is running on OpenGL 4.5, did get thrown out of the game once for no apparent reason, can say I got some of the same results as you that FPS is better in open spaces then more confined ones.

[Evildeffy]

Not sure you made a typo or something with your running on OpenGL4.3, but mine is running on OpenGL 4.5, did get thrown out of the game once for no apparent reason, can say I got some of the same results as you that FPS is better in open spaces then more confined ones.

Tomorrow I'll drop in a screenshot where it shows OpenGL 4.3 when I wake up and start DOOM up.

Perhaps it's a difference between AMD and nVidia ... But I doubt it.. Perhaps I'm a blind bat.

[Shinzai]

I made a quick 10 minute video of gameplay on my system:



Using an overclocked 980 1500mhz core, including boost and i7 3770K @ 4.5Ghz.

[Artorius]

Tomorrow I'll drop in a screenshot where it shows OpenGL 4.3 when I wake up and start DOOM up.

Perhaps it's a difference between AMD and nVidia ... But I doubt it.. Perhaps I'm a blind bat.

AMD runs on OpenGL 4.3 and Nvidia runs on OpenGL 4.5, yes it's different.

I'm not exactly sure why...

[Evildeffy]

Seems like there's a performance bug with shadows on AMD right now, lowering the shadows are giving people double the performance lol.

- - - Updated - - -


There are no physical cores and fake cores, they're all logical cores. They act as 2 when you can make them act as two, and they act as one big core if you can't.

It's simple to understand.

I'll try lowering the shadows and have a look.

Regarding HyperThreading .. Yes it technically is as it's a mask that identifies as such.. There are only 6 cores on my chip and no additional hardware to make it larger.

It is nothing but an added function to increase the utilization of the CPU core to it's max resources.
Your definition would literally mean that the core would br able to be split up and hold 50/50 each which it cannot.

Yes they are all identified as logical cores but Windows can tell the difference between the HyperThread and the actual core thread. If that weren't the case each core would have significantly less horsepower than it's i5 variant, the fact that you can even turn off hyperthreading shows that it's a mask.

Do programmes that multithread care? Nope for them it's the more the merrier.

You are however making a mistake when you state "act as one big core if you can't" as that would signify a dynamic cpu core control and that is (currently) impossible for any OS, they work with fixed threads and processID to Logical Core assignments.. Otherwise Intel would've succeeded long ago with their plans for "inverse HyperThreading" where 4 cpu cores would be seen as a single core if the application wasnt multithread capable.

Imagine what that would do for WoW.

The HyperThread is only there to allow for a single core to which it's paired with to allow itself to be fully utilised.

If they were "just" logical cores then Windows, who governs every single instruction sent to the cpu, would be unable to tell the difference and randomly assign tasks to the HyperThreads and it'd ve executed far more slowly.
In fact this used to even be the case until MS and Intel changed this behaviour to stop this from happening somewhere during Windows XP/Vista era.

Hell if you want to test this right now then ID your HyperThreads and set WoW's executable affinity settings to those cores only and see what happens.

I can give you a hint .. You won't be happy.

[Artorius]

Yes they are all identified as logical cores but Windows can tell the difference between the HyperThread and the actual core thread. If that weren't the case each core would have significantly less horsepower than it's i5 variant, the fact that you can even turn off hyperthreading shows that it's a mask.

No no no.

There are no actual core thread. Both are logical cores and a single real core has 2 threads. That is the case and the core can work in both ways, if you try to run something that doesn't support multithreading you can throw it at any of the 8 logical cores and it'll utilize 100% of the physical core that it is running at.

If for some odd reason you try to put something else on the other logical thread of the same physical core then the performance of the precious thing will suffer since now the core will have to do work for 2 tasks that are running at its two logical cores.

What Windows is able to do is split tasks in a way that it doesn't waste unused cores. If you have something running at core 0 and core 1, 2 and 3 aren't doing crap, if Windows couldn't recognize which logical cores belong to each core then you'd have a chance of Windows throwing a process at a logical core of the same core therefore impacting performance of both applications while the other 3 cores are still doing nothing. But since it knows which logical cores belong to their respective physical cores, Windows can split things between different cores and only put different tasks to be done at the same physical core (but different logical cores) if doing so would result in a performance boost.

Go read: https://en.wikipedia.org/wiki/Hyper-threading

"It is possible to optimize operating system behavior on multi-processor hyper-threading capable systems. For example, consider an SMP system with two physical processors that are both hyper-threaded (for a total of four logical processors). If the operating system's thread scheduler is unaware of hyper-threading, it will treat all four logical processors the same. If only two threads are eligible to run, it might choose to schedule those threads on the two logical processors that happen to belong to the same physical processor; that processor would become extremely busy while the other would idle, leading to poorer performance than is possible by scheduling the threads onto different physical processors. This problem can be avoided by improving the scheduler to treat logical processors differently from physical processors; in a sense, this is a limited form of the scheduler changes that are required for NUMA systems."

It's not about real cores and fake cores, you have 4 physical cores and 8 logical cores.

[Evildeffy]

I made a quick 10 minute video of gameplay on my system:



Using an overclocked 980 1500mhz core, including boost and i7 3770K @ 4.5Ghz.

Post exact specs + game settings and res plx.

[Shinzai]

Post exact specs + game settings and res plx.

http://www.3dmark.com/fs/8382244

Computer:
GPU: NVIDIA GeForce GTX 980 @ 1.5Ghz Core, 7.2Ghz on Vram
CPU: Intel Core i7-3770K Processor @ 4.5Ghz
Motherboard: ASRock Z68 Extreme4 Gen3
Ram: G.Skill RipjawsX 16GB (2x8GB) DDR3 PC3-12800 1600MHz
Hard Drive: Seagate SSHD 2GB ST2000DX001-1CM164

Drivers: 365.19

Game Settings: 1080p Ultra + Custom -

All settings left at Ultra defaults, except Anisotropic Filtering, which I upped to 16.

[Artorius]

Didn't you have a Qnix 2560x1440 monitor @Shinzai?
I think you mentioned it when we were playing BnS ages ago

[Jakexe]

Games runs very very smoothly for me.

I'll update when I actually run tests later today but from playing alone it's looking great. Which is great because it scared the fuck out of me when the 'bethesda' logo would barely play.

[Shinzai]

Didn't you have a Qnix 2560x1440 monitor @Shinzai?
I think you mentioned it when we were playing BnS ages ago

Yup, just used 1080p for the video as a benchmark reference.

[Deleted]

Running at 100+ fps on everything at max settings, 3440x1440p @ 100hz.

I7 3770K @ 4.2GHZ, GTX980 with 10% overclock

[Artorius]

This is so tryhard that I don't even know what to say...

[Evildeffy]

<Lots of HyperThreading info>

You are saying exactly what I'm saying and then saying something else.
It's a bit weird, your very quote is agreeing exactly what I'm saying...

But to rephrase and put your mind to rest:
The game is ignoring the logical threads that offer the least performance gain as the main thread of said physical core is more powerful due to the CPU's scalar architecture.

Happy now? :P

- - - Updated - - -

This is so tryhard that I don't even know what to say...

Ehrm I seriously doubt AMD's hardware can outperform nVidia's hardware in OpenGL mode, Vulkan probably but not OpenGL.

- - - Updated - - -

Update:

It is 4.3 and lowering shadows brings the FPS up to between 90 - 100 (and above at times) but the weird lag points etc. I had previously are gone.
I will post the exact settings I'm running with in 2 screenshots links shortly.

Don't think I'm going to match Shinzai's performance though .. but there could be disparity in settings.