Mirror's Edge Catalyst 'Hyper' Settings Made Possible by 8GB VRAM

[Sweepr]

Making full use of their blistering performance, and 8GB framebuffer, DICE is able to dramatically increase environmental detail and shadow quality, and further improve the quality of reflections, visual effects, environmental maps, image based lighting, motion blur, and Resolution Scale’s downsampling. Classed as “Hyper” settings, these enhancements enable users to explore a richer, more detailed Glass City on the very best PCs and graphics cards, delivering the definitive Mirror’s Edge Catalyst experience.

For the technically-minded, here’s a quick overview of the many features enhanced by Hyper:

Lighting Quality: Shadow Map Resolution, Spotlight Shadow Map Resolution, and Shadow Draw Distance greatly increased
Mesh Quality: Object Draw Distance, Character Draw Distance, Object LOD Quality, and Object Shadow Draw Distance greatly increased
Reflection Quality: Fidelity further improved
Texture Quality: Environment Mapping Quality improved and Image-Based Lighting Resolution increased
Motion Blur: Sample count increased
Render Scaling: Downsampling upgraded to Lanczos Separable from Bicubic Sharper

www.geforce.com/whats-new/articles/mirrors-edge-catalyst-geforce-gtx-1080-1070-hyper-settings


GameGPU tested at 'Very High Quality', but they don't mention which settings from the game it corresponds to (Ultra?).

According to GamersNexus you need at least 6GB VRAM to run at Hyper Settings, even at 1080P. 7455MB VRAM consumed with a Geforce GTX 1080!

“High” runs about 14.36% higher in AVG FPS than “Ultra” at 1080p during the research stages. “Ultra” then runs about 38.4% faster than “Hyper.” We have been unable to get “Hyper” reliably executing on some cards – like the R9 Fury X & GTX 970 – and that instability seems to coincide with devices running 4GB VRAM. Devices running 6-8GB VRAM did not have this issue.

The Fury X and GTX 970 refused to run with any stability – we think that's a VRAM limitation – and so the chart only shows the 390X, 1080, 980 Ti, and 1070. All of these cards have 8GB of VRAM, except the 980 Ti and its 6GB.

The GTX 1080 runs 29.5% slower with “Hyper” graphics than with “Ultra.” The GTX 1070 trails as expected, based upon our GTX 1070 review. The R9 390X runs at 54FPS on Hyper, versus 79.7FPS on “Ultra.” For most devices, 1440p/Ultra is fairly close in performance with 1080/Hyper – but the 980 Ti shows the choke-point: Memory. Being the only card with 6GB VRAM on this particular chart, the 980 Ti scales poorer than the 8GB counterparts. It still does well, don't get us wrong; but the 980 Ti does serve as a point of clarity for when Mirror's Edge is most hungry for VRAM.

We haven't spent too much time fully validating system RAM and CPU consumption, but had a chance to look at VRAM utilization during our test passes. In just the few minutes tested on each configuration, 1080p / Hyper seems to be nearly maxing-out the GTX 1080 (7455MB consumed). 4K / High had us sitting at 6615MB, with 1080p / Ultra resting more easily at 5020MB. We'll need to run endurance tests to see if these numbers increase with longer play sessions. For now, that's what we were getting from 5-minute, quick-and-dirty measurements.

www.gamersnexus.net/game-bench/2471-mirrors-edge-catalyst-graphics-card-benchmark-gtx-1080-1070-390x

Hyper textures require a minimum of 6 GB graphics card memory

...It requires at the "Hyper-textures" in Full HD at least a graphics card with six gigabytes and from 2,560 × 1,440 then must it be eight gigabytes.

...The Radeon R9 390 easy to lose 25 percent higher and the Radeon R9 Fury X 27 percent. The latter has the problem that the four gigabyte memory is no longer sufficient. The FPS-loss thereby is indeed low, but the game not running smoothly.

www.computerbase.de/2016-06/mirrors-edge-catalyst-benchmark

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.

http://forums.anandtech.com/showpost.php?p=38284089&postcount=34

We've re-tested Mirror's Edge Catalyst in order to get to the bottom of how the hyper settings and VRAM limitations affect performance on GTX 970 and R9 390 (8GB). The bottom line: ultra works best for both cards, but only the R9 390 (8GB) can only fully enable hyper settings.

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


PCLab results:

http://pclab.pl/art70140.html


Will future DICE games (Battlefield 1) also benefit/need 8GB+ VRAM VGAs to run at max settings?

 

[Bacon1]

You should probably tag this with 390, 390x or Hawaii since you are bothering to tag it with GP104.

Also throw in Polaris / 480 since it will have 8gb cards.

 

[Sweepr]

You should probably tag this with 390, 390x or Hawaii since you are bothering to tag it with GP104.

Also throw in Polaris / 480 since it will have 8gb cards.

I thought about it but Radeon 390/390X's performance is sub-par (<30 FPS dips) in 1080P Hyper according to Gamer's Nexus. But your point still stands, I think the extra bucks for the 8GB version of Radeon RX 480 will be well worth it in 2016 and beyond.

 

[Yakk]

If a single 390x can achieve 54 fps, should be interesting for the 8GB RX480.

Looks like the 970 is already being left in the dust by nvidia.

 

[Bacon1]

I thought about it but Radeon 390/390X's performance is sub-par in 1080P Hyper according to Gamer's Nexus. But your point still stands, I think the extra bucks for the 8GB version of Radeon RX 480 will be well worth it in 2016 and beyond.

It's getting 54 fps average, not bad for a chip originally from 2013 with stock clocks vs the OC'd Nvidia cards.

 

[96Firebird]

Looks like the 970 is already being left in the dust by nvidia.

Do you also believe AMD is leaving Fury X in the dust as well? Hell, that is less than a year old and cost twice as much as the 970!

Capability of all graphics cards is limited in some fashion...

 

[Bacon1]

Interesting to see the 970 outpacing both 390x and Fury X considering its usually sitting next to the 390, not 390x.

 

[Mercennarius]

I can see all 4GB cards really struggle with new games on high settings over the next year.

 

[tential]

I wonder what the benefit is side by side for this "amazing" boost.

 

[tential]

Also, lets look at the real story here. A low clocked 980Ti is almost on par with an OC'd GTX 1080? Why is that even possible? Put a normal OC on the GTX 980Ti to even 1400mhz and it's above an OC'd GTX 1080. Talk about a failure.

 

[Yakk]

Do you also believe AMD is leaving Fury X in the dust as well? Hell, that is less than a year old and cost twice as much as the 970!

Capability of all graphics cards is limited in some fashion...

This is an nvidia sponsored game and option it looks like. It will be interesting to see if AMD updates their dynamic VRam allocations in the drivers to allow this option to run on 4gb cards.

 

[crisium]

Definitely a game to dial back the settings, especially 1440P users.

Surprising to see such demand after Battlefront easily hits 60fps on many cards and under 4GB too.

Neutral vs Gameworks, I guess. I don't have an issue with "future cards needed" for the absolutely highest settings necessarily, I'm just surprised to see it from DICE. I'm sure the High Texture setting combined with Ultra everything else still looks quite good and should get 60fps on my rig. I'll check out comparisons of settings later.

 

[Bacon1]

You also left off that gamegpu tests were much different than the ones you posted:

http://gamegpu.com/action-/-fps-/-tps/mirror-s-edge-catalyst-beta-test-gpu

Huge performance gain on Nvidia cards in the gamersnexus test, 20-30% each.

 

[Thala]

But your point still stands, I think the extra bucks for the 8GB version of Radeon RX 480 will be well worth it in 2016 and beyond.

I don't think so. We are talking about settings for which even an OCed GTX 1080 is unable sustain 60fps in 1080p!

 

[linkgoron]

Interesting to see the 970 outpacing both 390x and Fury X considering its usually sitting next to the 390, not 390x.

The 1070 is basically beating the 1080 in that benchmark.

Also, note that the nvidia cards are all overclocked while the AMD cards are all basically stock (+50 mhz for the 390, less than 5%).

that 970 is 15% over stock.

 

[Sweepr]

Also, lets look at the real story here. A low clocked 980Ti is almost on par with an OC'd GTX 1080? Why is that even possible? Put a normal OC on the GTX 980Ti to even 1400mhz and it's above an OC'd GTX 1080. Talk about a failure.

No it wouldn't, even if you add +20% to Geforce GTX 980 Ti @ 1228 MHz at 1080P Hyper / 4K High:

And I'm pretty sure that's not the best example of Geforce GTX 1080 overclocking out there.

 

[96Firebird]

TIL 20-30% is "almost on par"...

 

[Face2Face]

Interesting to see the 970 outpacing both 390x and Fury X considering its usually sitting next to the 390, not 390x.

I'm sure that will change with some future driver updates. Isn't this a NVIDIA sponsored games as well?

Graphics look great, but gameplay looks meh...

 

[Sweepr]

You also left off that gamegpu tests were much different than the ones you posted:

http://gamegpu.com/action-/-fps-/-tps/mirror-s-edge-catalyst-beta-test-gpu

Huge performance gain on Nvidia cards in the gamersnexus test, 20-30% each.

Did you even read the OP?
4GB VRAM is not a problem at up to 1440P Ultra, even for Geforce GTX 970. You need more for Hyper settings (1080P and up) and 4K.

 

[96Firebird]

You also left off that gamegpu tests were much different than the ones you posted:

Much newer drivers from GamerNexus...

 

[Bacon1]

Did you even read the OP?
4GB VRAM is not a problem at up to 1440P Ultra, even for Geforce GTX 970. It is required for Hyper settings and probably 4K.

Did you even read my post?

GameGPU showed the 970 getting 39 fps avg, 32 min @ 1440p

GamersNexus showed 970 getting 64 avg, 45 min @ 1440p

That is an increase of over 60%

980 TI went from 47/38 to 92/59

That is an increase of almost 100%

Do you honestly see nothing wrong with these numbers? When a 970 goes from being tied with 290x (where it normally is, next to a 390) to being faster than a Fury X @ 1440p where it is normally having issues?

Also still waiting for you to either remove gp104 from tags or add all the other >6gb cards.. or else your bias is showing pretty strong

 

[Bacon1]

Much newer drivers from GamerNexus...

60% gain for 970, almost 100% gain for 980 TI

970 doing better than Fury X @ 1440p

You don't see any issues with those?

 

[Sweepr]

Did you even read my post?

Yes, and I also bothered to check drivers first:

Nvidia GeForce/ION Driver Release 364.72

AMD Radeon Crimson Edition 16.4.1

The latest AMD drivers (16.6.1 with Mirror's Edge Catalyst support) were used for testing. NVidia's 368.39 drivers were used.

Not only this but many Geforce VGAs are (factory) overclocked in GamersNexus. There goes your conspiracy theory.

 

[96Firebird]

Realize GameGPU's Nvidia drivers are from March 28th...

I fully expect you to believe there is something going on behind the scenes, as you continually do.

 

[lilltesaito]

60% gain for 970, almost 100% gain for 980 TI

970 doing better than Fury X @ 1440p

You don't see any issues with those?

I am sure it will be AMD fault, driver team just kicking back drinking beers. Once the game is out, they will get to work and make drivers for the game and the FPS will go up. But that will not happen until after the reviews are out for the game.