• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Quantum Break another example of poor optimisation from console?

C

CSI_PC

2[H]4U
Joined
Apr 3, 2016
Messages
2,193
Not only poor optimisation but also rushed for release on PC.

Watching and listening to the Digital Foundry report regarding PC version of Quantum Break I feel brings home this trend of what is happening with regards to priority development and optimisation on console that is then followed up on PC and both rendering engines and other optimisations end up being reduced.

My favourite example of how consoles are influencing PCs (and some times IMO directly GPUs) was how Witcher III graphics render engine that initially looked great at its 1st demo presentation had to be redone to be efficient with consoles.
That engine was used on both consoles and PC, and lead to many being upset about graphics being degraded/parity even if they were not aware that they had to change the engine for valid reason.

Back to Quantum Break.
Now we are seeing some face palm moments with the PC project development-release of Quantum Break on PC.
It is using the same identical rendering engine, that not only has its optimisation based around console but critically still is at 720p and upscaled (temporal reconstruction) affecting texture-objects-aliasing-etc.

Watching the video is painful, and kind of reminds me of the days when developers could not use Tessellation correctly causing massive impact for AMD hardware.
A good example in the video is where they show a lot of the issues come from the lighting/particles, which if this was also optimised for NVIDIA could be done in ways to seriously improve performance in DX12.
On the plus side it is good on AMDs in comparison and that is something to cheer them up :)

Enjoy the video, or wince :)


Got to give AMD credit in wrapping up the console market as a soft benefit (beyond hard cash) is now coming to fruition IMO.
This does not mean NVIDIA is in trouble, it does mean a lot more work needs to be done by developers and possibly with NVIDIA, although some of this may also be less of an impact on Pascal but we will have to wait and see.
Cheers
 
Last edited:
I bet it's more of a driver issue than a hardware issue. Quite sure that Nvidia can optimize their drivers for DX12 titles on the PC much better than this. Across the board Nvidia has been lacking at DX12. Title after title this has been the theme. So it's nothing new at this point, but still sad as their customers deserve better. It's Remedy Entertainment who created great games like Max Payne on the PC. Look at their track record; they aren't into a Red vs Green rivalry. Their games work well across all platforms. And since the game runs great on AMD obviously it has a working path.

Nvidia needs to optimize their drivers and maybe ask Remedy to change something about the lighting to give their GPUs a break. The lighting affects AMD cards also in certain situations as my game got choppy for a few seconds in one particular area just before the end of Act 1.
 
Just a question for those with the game, on AMD if this game uses tesselation, does limiting it to 16x affect the performance and any visual changes?
 
cageymaru said:
I bet it's more of a driver issue than a hardware issue. Quite sure that Nvidia can optimize their drivers for DX12 titles on the PC much better than this. Across the board Nvidia has been lacking at DX12. Title after title this has been the theme. So it's nothing new at this point, but still sad as their customers deserve better. It's Remedy Entertainment who created great games like Max Payne on the PC. Look at their track record; they aren't into a Red vs Green rivalry. Their games work well across all platforms. And since the game runs great on AMD obviously it has a working path.

Nvidia needs to optimize their drivers and maybe ask Remedy to change something about the lighting to give their GPUs a break. The lighting affects AMD cards also in certain situations as my game got choppy for a few seconds in one particular area just before the end of Act 1.
Click to expand...
I doubt the driver is going to do anything regarding the way the rendering engine/lighting/etc is implemented and still same as console implementation.
Drivers may add stability that is erratic with NVIDIA but IMO the game needs a lot of developer intervention to improve its performance, and especially so on NVIDIA cards.
Why this goes beyond just an NVIDIA driver, while the 390/390x perform and behave better than NVIDA they are still below where they should be for the visual fidelity of this game..

The fact that the PC version is still using the temporal reconstruction core to the graphics engine is pretty damning, and just a big neon sign IMO on how PC is secondary in the development-optimisation of this title.
Cheers
 
cageymaru said:
I bet it's more of a driver issue than a hardware issue.
Click to expand...
Might be more of a Microsoft issue than a driver issue. That strange double buffering thing they have going on might be throwing off the rendering.

CSI_PC said:
The fact that the PC version is still using the temporal reconstruction core to the graphics engine is pretty damning, and just a big neon sign IMO on how PC is secondary in the development-optimisation of this title.
Click to expand...
I'm a bit mixed on this. On one hand they're re-using a performance boosting technique that probably saves development time, on the other they may not have released the game for PC.

This might be worth looking at in a month if/when Microsoft fixes their vsync issue.
 
CSI_PC said:
I doubt the driver is going to do anything regarding the way the rendering engine/lighting/etc is implemented and still same as console implementation.
Drivers may add stability that is erratic with NVIDIA but IMO the game needs a lot of developer intervention to improve its performance, and especially so on NVIDIA cards.
Why this goes beyond just an NVIDIA driver, while the 390/390x perform and behave better than NVIDA they are still below where they should be for the visual fidelity of this game..

The fact that the PC version is still using the temporal reconstruction core to the graphics engine is pretty damning, and just a big neon sign IMO on how PC is secondary in the development-optimisation of this title.
Cheers
Click to expand...
If I search your post history will I see you complaining and calling out the developer about Tomb Raider performance? Or is that game's DX12 implementation good enough since Nvidia wins the benchmarks?

I mean if performance like this is bad (even though it has far better minimums)

QB_2560.jpg


Then why aren't you complaining about performance like this, with far worse minimum framerate?

tr_2560_12.jpg


And please don't cop out saying you can run the game in DX11 for better performance.
 
Last edited:
If you followed my post you would also know I showed the benchmarks by Brent from this site :)
Just to re-iterate again Brent tested:
Febuary :
Performance - Single GPU - 1440p - Rise of the Tomb Raider Video Card Performance Review
Late March with latest game build. This now takes the place of our apples-to-apples performance:
Benchmark - NVIDIA - Rise of the Tomb Raider DX11 vs. DX12 Review
Benchmark - AMD - Rise of the Tomb Raider DX11 vs. DX12 Review
NVIDIA quite a lot ahead now and it is clear a performance swap has occurred.

TBH I really really am not a fan of the site you linked the benchmark for, also the reason I used here is because Tomb Raider was tested twice and with different game builds.
And yes this IS about DX11 because that is the only one we have a realistic trend or even analysis over the years, and the only true api we can use to look at past and present on PC, and then also with console still DX11 as it shows optimisation IF done by developers for all platforms (DX12 which is relevant when forward looking but even then what we have so far does not help as they are not using all the feature sets - Intel has worked with 2 developers on CR/ROV so lets see what happens there and also DX12 comes down to requiring a strong understanding of developing low level on both AMD and NVIDIA for optimal performance).
But please if you want to argue about this, why not take it to the Tomb Raider thread where I last posted my thoughts and asked Brent some questions :)

So regarding Tomb Raider, it shows that AMD has an easier time with the console game being taken to PC and seems it aligns better with the optimisation done on the console version.
It requires work from both the developer and NVIDIA to get the optimisation one would expect, and obviously this is extra work so AMD these days has an edge I feel due to dominating console market and goes beyond the DX12 performance seen so far.

Case in point the latest build of Tomb Raider includes features within 11.3 associated with CR-VXAO and NVIDIA related.

Cheers

EDIT:
Tidied performance chart links as a couple were wrong.
 
Last edited:
Well pcgameshardware is seeing same thing as Digital Foundry.
Usually the German site is also one of the ones I feel seem pretty good with their analysis.
Will need to translate: Quantum Break PC jetzt mit echten Benchmarks: Technik-Test [Update]
The GPU comparison is about 3/4 of the way down, with the earlier sections showing some issues within the game.

Worth noting though Razor, Digital Foundry highlighted the performance tanked where there is a lot of volumetric lighting and particles, so performance could be NVIDIA down some or down an insane amount.
Cheers
 
primetime said:
yep different cards, different results lol people just get a little excited amd performs better in some price groups....i hate nvidia in general but i seriously doubt there in any trouble that cant be fixed with driver updates - people that say otherwise are being silly
Click to expand...
The Bad optimization isnt going to be solved so easily with a better driver. look DX12 games are more dependent on Developers than AMD/nvidia using DX11

[techtimes] Vast Majority Of DirectX 12 Games Will Use AMD, Says AMD Exec - Page 5
Direct-X12-Panel-Slides-10.jpg
 
Last edited:
primetime said:
the bad thing is we get what we deserve...all the money is made on consoles game sales these days. I know not us exactly but if we owned the company we would prioritize were the profit is made just as they do now.
Click to expand...
And part of it is business risk analysis, they probably feel PC is riskier than consoles regarding return on investment developing-selling a game.
Chers
 
Lorien said:
If I search your post history will I see you complaining and calling out the developer about Tomb Raider performance? Or is that game's DX12 implementation good enough since Nvidia wins the benchmarks?

I mean if performance like this is bad (even though it has far better minimums)

QB_2560.jpg


Then why aren't you complaining about performance like this, with far worse minimum framerate?

tr_2560_12.jpg


And please don't cop out saying you can run the game in DX11 for better performance.
Click to expand...
Um they must be using 4x SSAA in that Tomb Raider test as those numbers are insanely low.

EDIT: I looked and they are using just SMAA which makes no sense as I get over 75 fps with every single setting manually maxed out at 1440 except AA.
 
Last edited:
misterbobby said:
Um they must be using 4x SSAA in that Tomb Raider test as those numbers are insanely low.

EDIT: I looked and they are using just SMAA which makes no sense as I get over 75 fps with every single setting manually maxed out at 1440 except AA.
Click to expand...

Yeah there is something profoundly wrong with those Tomb Raider results, with the exception of the texture resolution being set to only high (memory limitations) my 980 goes right past their 980 Ti results.
 
Kor said:
Yeah there is something profoundly wrong with those Tomb Raider results, with the exception of the texture resolution being set to only high (memory limitations) my 980 goes right past their 980 Ti results.
Click to expand...
After looking at it again they are testing in what I think is the most demanding part of the game. I believe I was in the low 50s there with my oced 980 ti where they tested which lines up with them getting 41 fps on stock 980 ti in that spot.
 
Crazy. My gtx 970 just goes 30 fps on this at my native res. It should be a driver thing.
 
May 28, 2016 test with the GTX 980ti vs GTX 1080. The new preemption and ASYNC DX12 driver improvements are freaking awesome! Never seen such a performance boost from generation to generation @1080p. misterbobby found the video.





Asus GTX 1080 Founders edition
Asus GTX 980 TI Reference

CPU - Skylake 6700K

Used shadowplay to record.
 
Last edited:
cageymaru said:
May 28, 2016 test with the GTX 980ti vs GTX 1080. The new preemption and ASYNC DX12 driver improvements are freaking awesome! Never seen such a performance boost from generation to generation @1080p.





Asus GTX 1080 Founders edition
Asus GTX 980 TI Reference

CPU - Skylake 6700K

Used shadowplay to record.
Click to expand...
Sarcasm? If not then you are way off for this DX12 game as the 1080 has a smaller lead over 980 Ti than it typically does in DX11 games.
 
I can't believe how poor the frames are, especially considering just how powerful both of those cards are. At 1080p I would have thought, it would be closer to 100 frames and not 40. Something is terribly wrong with that game.
 
Maddness said:
I can't believe how poor the frames are, especially considering just how powerful both of those cards are. At 1080p I would have thought, it would be closer to 100 frames and not 40. Something is terribly wrong with that game.
Click to expand...
Doom is a better looking game and literally gets over THREE times the framerate.
 
cageymaru said:
May 28, 2016 test with the GTX 980ti vs GTX 1080. The new preemption and ASYNC DX12 driver improvements are freaking awesome! Never seen such a performance boost from generation to generation @1080p. misterbobby found the video.





Asus GTX 1080 Founders edition
Asus GTX 980 TI Reference

CPU - Skylake 6700K

Used shadowplay to record.
Click to expand...


I'd like to see that test done at 1440p, not 1080p.

Also not sure the drivers being used, NVIDIA did just release new drivers for everything, and possibly has a newer set coming this week.
 
I forgot I had this thread lol
I would had posted HardwareCanucks PresentMon analysis of Quantum Break with 1080FE.
Trend shows it is working better for Pascal than Maxwell , although AoTS shows even bigger jump between 980 and 1080 - again they used PresentMon and not the internal benchmark.

Just to emphasise, we really should use 980 rather than 980ti IF looking at this from a technical performance perspective rather than a buyers perspective.
The reason is that the overall architecture of 1080 is closer to the 980 in terms of GPC (clusters) and ROPs, the 1080 has more SM streams but that is only because Pascal increased that to 5 per GPC cluster rather than 4 in Maxwell)
The 980ti though still has more SM streams, GPC clusters, ROPs, TUs,etc.

So here is the Quantum Break comparison showing trend improvement for 1080.
GTX-1080-REVIEWS-76.jpg



So average fps is just about 95% faster than the 980, where trend usually put games between 65% to 75% faster.

GTX-1080-REVIEWS-51.jpg


The NVIDIA GeForce GTX 1080 Review - Page 19

AoTS is about 101% faster on 1080.....
The NVIDIA GeForce GTX 1080 Review - Page 18

I am using 1440p as it is the bridge between the two extremes of too small and too large influencing HW spec.

Importantly. just to add PresentMon was developed by Intel and closer to FRAPs (but not an identical replacement).

Cheers
 
Last edited:
Brent_Justice said:
I'd like to see that test done at 1440p, not 1080p.

Also not sure the drivers being used, NVIDIA did just release new drivers for everything, and possibly has a newer set coming this week.
Click to expand...

And on top of that I would prefer a publication as they do better analysis over a random youtube poster (in this instance using a 980ti without including a 980)..
Pretty sure most publications would also use a 980 for comparison and potentially additional analysis tools as well such as PresentMon.

As I mention with my above post, technically it would make more sense IMO to compare to a 980 rather than a 980ti due to the GPC/SM structure and ROPs disparity between them, which also still happens between a 1080 and 980ti.

So I think most of that youtube video channel comparison are meaningless unless looking from a buyers perspective rather than a technical comparison, IMO anyway.
Cheers
 
Brent_Justice said:
I'd like to see that test done at 1440p, not 1080p.

Also not sure the drivers being used, NVIDIA did just release new drivers for everything, and possibly has a newer set coming this week.
Click to expand...

Yes, I agree and hope that they fix whatever issues the game has. Remedy usually has great games like the Max Payne series on PC that run well out of the box. Game runs well (eye test) on my system. Really in the back of my mind I'm hoping that Remedy does a performance patch that knocks it out of the park. Since the game is very playable on my system I'm thinking that Nvidia is probably going to fix a great deal of the performance deficit via a driver.

You'll should do a short article after Remedy does their upcoming PC patch that they were talking about awhile ago, and Nvidia / AMD release drivers for it. The textures are unremarkable to say the least, but the lighting fits the areas very well. I can't say that it is the most advanced lighting I've seen, but it well done in the maps. The big problem is that turning on all of the lighting effects kills performance. I guess this would be akin to typical GameWorks lighting effects that murder frame rates also.
 
You must log in or register to reply here.
Back
Top