Mirror's Edge Catalyst Benchmarks Megathread

[Bacon1]

Is this game DX12? That's the biggest gap between 980 TI and 1080 I've seen yet. Seems the 1440P bench from computerbase was not done at Hyper but Ultra so VRAM issues should not have come into play here either.

Its DX11

 

[Bacon1]

Digital Foundry made a video showing the 970 beating the 390 @ Hyper, but its obvious to anyone who watched the video that the 970 wasn't running in Hyper mode.

Need to make sure reviewers do IQ checks!

https://youtu.be/6JhDRtA8Vg4?t=74

 

[Kenmitch]

Digital Foundry made a video showing the 970 beating the 390 @ Hyper, but its obvious to anyone who watched the video that the 970 wasn't running in Hyper mode.

Need to make sure reviewers do IQ checks!

https://youtu.be/6JhDRtA8Vg4?t=74

Yep. But too be fair it looks like crap on both rigs.

I'd imagine if it was a highly anticipated game that it would at least have it's own thread in the PC Gaming sub-forum.

Hyper mode looks to be over hyped viewing the screenshots.

 

[Bacon1]

Yep. But too be fair it looks like crap on both rigs.

I'd imagine if it was a highly anticipated game that it would at least have it's own thread in the PC Gaming sub-forum.

Hyper mode looks to be over hyped viewing the screenshots.

Completely agree, reviewers need to be doing IQ comparisons and actually testing to see if the performance hit is worth it or not. Hyper vs Ultra? Some shadows look different (hard vs soft) but is that worth a massive 25% performance hit??

Hell many games its hard to tell high vs very high for many settings but you can get massive amounts of performance.

This whole "max everything run benchmark record results" without any analysis is terrible and does a disservice to anyone reading / watching. It seems like they are just trying to upsell you by not bothering to see if the performance hit is actually worth it for better IQ.

 

[caswow]

Problem is a lot of people already have seen this bullcrap and blamed it yet again on amd and their drivers...

 

[crisium]

So I decided to buy the game. I'll try Hyper with the memory management off for kicks, but I'm planning to run just on Ultra. I'll let you know how the 4GB handles it. I'll fit some time in after work today.

 

[Silverforce11]

Is this game DX12? That's the biggest gap between 980 TI and 1080 I've seen yet. Seems the 1440P bench from computerbase was not done at Hyper but Ultra so VRAM issues should not have come into play here either.

Not surprised...

NV has shifted focus on "Game Ready" drivers to Pascal already.

25% lead goes to 40%.

It's what I said about Pascal a few months back. That launch reviews will show a ~25% lead, but the gap will grow very quick in new games.

 

[tential]

Not surprised...

NV has shifted focus on "Game Ready" drivers to Pascal already.

25% lead goes to 40%.

It's what I said about Pascal a few months back. That launch reviews will show a ~25% lead, but the gap will grow very quick in new games.

"how will they gimp maxwell when the architectures are similar!" they asked. Never under estimate Nvidia...

 

[crisium]

Hyper mode is a stutterfest if you uncheck "GPU Memory Restriction" on my Fury, even when I tested only at 1080p it is always below 30fps. Not realistically possible to play the real Hyper with 4GB.

While using the GMR it is possible to use Hyper, and it uses basically the full 4GB (more than Ultra which is usually around 3.5-3.9 at 1440P).

I would guess Hyper with GMR is a mix of Ultra and Hyper, dynamically scaling more Ultra when it is low on memory. Given that I'm already fairly close to 4GB on Ultra, it's probably much more Ultra than Hyper on a 4GB card.

In the Digital Foundry video it is likely that they played with Hyper GMR enabled, and the 970 was actually running nearly entirely Ultra due to the dynamic scaling. With a full 8GB, the 390 was likely running the full Hyper.

DF has quite the reputation for analyzing any tiny PS4 vs Xbone detail difference. It is quite embarrassing they did not realize this, and are now likely spreading misinformation.

 

[Bacon1]

Yep the DF test is completely wrong, yet they've left the video up for hours now even after admitting they might have made a mistake. They should have taken it down or at the very least put up a warning message on top of the video. People aren't going to read the description.

You have to wonder why they didn't test Ultra settings and instead went all the way down to medium in the later part of the comparison... probably because the 970 framerate was the exact same? :\

 

[littleg]

Hyper mode is a stutterfest if you uncheck "GPU Memory Restriction" on my Fury, even when I tested only at 1080p it is always below 30fps. Not realistically possible to play the real Hyper with 4GB.

While using the GMR it is possible to use Hyper, and it uses basically the full 4GB (more than Ultra which is usually around 3.5-3.9 at 1440P).

I would guess Hyper with GMR is a mix of Ultra and Hyper, dynamically scaling more Ultra when it is low on memory. Given that I'm already fairly close to 4GB on Ultra, it's probably much more Ultra than Hyper on a 4GB card.

In the Digital Foundry video it is likely that they played with Hyper GMR enabled, and the 970 was actually running nearly entirely Ultra due to the dynamic scaling. With a full 8GB, the 390 was likely running the full Hyper.

DF has quite the reputation for analyzing any tiny PS4 vs Xbone detail difference. It is quite embarrassing they did not realize this, and are now likely spreading misinformation.

Is there an optimised driver for this yet from AMD? I'm wondering if that might help with the HBM cards as iirc they need to optimise the memory usage specifically on a per game basis to get the best performance at that level.

 

[crisium]

Yes, I'm using 16.6.1 which has optimizations.

 

[Bacon1]

Is there an optimised driver for this yet from AMD? I'm wondering if that might help with the HBM cards as iirc they need to optimise the memory usage specifically on a per game basis to get the best performance at that level.

There might be another driver for Fury series, but honestly the Hyper settings aren't worth the performance hit regardless. Massive (20%+) hit for "hard" shadows which you probably won't even notice as you fly through the game running as fast as you can.

 

[boozzer]

this thread actually showed the other thread to be complete fud.

 

[RussianSensation]

this thread actually showed the other thread to be complete fud.

User reviews:

PC - 3.4/10
http://www.metacritic.com/game/pc/mirrors-edge-catalyst

PS4 - 4.1/10
http://www.metacritic.com/game/playstation-4/mirrors-edge-catalyst

XB1 - 3.4/10
http://www.metacritic.com/game/xbox-one/mirrors-edge-catalyst

Also, the number of reviews for such a high profile game is dismal. I am also not aware of any PC game worth buying that benefits from 8GB of VRAM without requiring a Titan X/1070/980Ti SLI to run it at 1440/4K anyway. Chances are those cutting edge PC gamers on Fury X CF would upgrade to 1070/1080 SLI anyway even without Mirror's Edge Catalyst, simply to lower power usage while preserving the resale value.

 

[Bacon1]

Digital Foundry redid their test, added it to the OP but here it is as well:

https://www.youtube.com/watch?v=Bm5ZCJah-FY

 

[RussianSensation]

Digital Foundry redid their test, added it to the OP but here it is as well:

https://www.youtube.com/watch?v=Bm5ZCJah-FY

:thumbsup: But I thought GTX970 was outperforming a Fury X at 1440p on Ultra at GamersNexus.

 

[Bacon1]

:thumbsup: But I thought GTX970 was outperforming a Fury X at 1440p on Ultra at GamersNexus.

Yep its clear that GamersNexus was incorrect in all of their testing as well.

 

[caswow]

this thread quickly died down after the retest

 

[Bacon1]

http://community.pcgamingwiki.com/p...-reports/pc-report-mirrors-edge-catalyst-r195

 

[zlejedi]

http://www.purepc.pl/karty_graficzn..._pc_test_wydajnosci_kart_graficznych?page=0,7

benchmarks done with memory restriction turned off

Suprising how AMD barely matches non overclocked maxwell equivalents in Frostbite game.

 

[lyssword]

unrelated to video bench, but it does scale well to 8 threads vs 4 lol seems latest games utilize cpu better

 

[crisium]

http://www.purepc.pl/karty_graficzn..._pc_test_wydajnosci_kart_graficznych?page=0,7

benchmarks done with memory restriction turned off

Suprising how AMD barely matches non overclocked maxwell equivalents in Frostbite game.

That's your interpretation?

 

[zlejedi]

That's your interpretation?

That's what we call reality - when reference clocked 980 matches Fury then overclocked one will be faster due to bigger overclocking headroom on maxwell cards.

 

[linkgoron]

[delete] mistake.