Metro Exodus Benchmark Performance, RTX & DLSS

[W1zzard]

Metro Exodus launches this week. We thoroughly tested the game's technical aspect, on 3 resolutions, with 16 graphics cards, including the whole RTX lineup and Radeon VII. We also have a ton of screenshots and side-by-side image comparison using NVIDIA RTX and DLSS.

Show full review

 

[jabbadap]

Out of no where, and bam Metro goodness... Did I miss something or where is the DLSS numbers on performance page?

 

[ArbitraryAffection]

Ouch somewhat poor Radeon performance.

But by the time i buy this maybe the drivers and/or patches will improve things. Either way I have just bought a Freesync 1080p monitor so even if my 570 can only do 40~ fps it should still be playable and smooth,..

 

[raptori]

Thanks for the Performance review, Interesting that some pictures in RTX off looks better and nice to see that developers didn't make RTX-off looks bad on purpose ... like the first picture I'd say that RTX -off looks better and it seems that ray-tracing doesn't give enough global illumination.

Does DX11 give better performance ?

 

[dirtyferret]

awesome, thank you

 

[Al Chafai]

the game runs better on Nvidia Hardware,How the 3 Gb 1060 coming that close to The RX 590.
do u think AMD will release a Game ready driver to boost performance?
also thnx for the amazing analysis.

 

[nickbaldwin86]

makes me want a 2k series card before buying the game... oh well should play just the same without the never ending eye candy on my current rig setup

 

[Al Chafai]

Thanks for the Performance review, Interesting that some pictures in RTX off looks better and nice to see that developers didn't make RTX-off looks bad on purpose ... like the first picture I'd say that RTX -off looks better and it seems that ray-tracing doesn't give enough global illumination.

i agree,but i think that the RTX On looks better in the outdoor screen shots,and looks bad indoor.
idk but i think that's what i noticed.
also is it me or the DLSS screens shots are so blurry compared to the normal ones?

 

[W1zzard]

Did I miss something or where is the DLSS numbers on performance page?

Didn't have time for full benching of all cards with DLSS, the DLSS screenshot comparison page has performance numbers with RTX 2080 Ti in the titles of the images

also is it me or the DLSS screens shots are so blurry compared to the normal ones?

Yeah seems more blurry than what we've seen in Final Fantasy I think

do u think AMD will release a Game ready driver to boost performance?

I'm sure they will release a driver, just not sure how much improvement

 

[IceShroom]

Ouch somewhat poor Radeon performance.

But by the time i buy this maybe the drivers and/or patches will improve things. Either way I have just bought a Freesync 1080p monitor so even if my 570 can only do 40~ fps it should still be playable and smooth,..

What you expect from Nvidia sponsored title?? Poor performence on Readons.

 

[londiste]

Interesting that some pictures in RTX off looks better and nice to see that developers didn't make RTX-off looks bad on purpose ... like the first picture I'd say that RTX -off looks better and it seems that ray-tracing doesn't give enough global illumination.

This is much more likely a question of building the level and lighting. Assuming that RTX still does pointlights just with rasterization methods, this highlights the differences the lighting methods. RTX is likely to be more "correct" which the original level/lighting design does not account for.

 

[jabbadap]

Didn't have time for full benching of all cards with DLSS, the DLSS screenshot comparison page has performance numbers with RTX 2080 Ti in the titles of the images


Yeah seems more blurry than what we've seen in Final Fantasy I think


I'm sure they will release a driver, just not sure how much improvement

What do you compare DLSS, no-AA or some other AA mode? FF had horrible TAA, which made DLSS look better.

 

[kastriot]

Did you tried low settings with 2080Ti+RTX @4K and does it go over 60fps?

 

[champsilva]

Thanks for the Performance review, Interesting that some pictures in RTX off looks better and nice to see that developers didn't make RTX-off looks bad on purpose ... like the first picture I'd say that RTX -off looks better and it seems that ray-tracing doesn't give enough global illumination.

Does DX11 give better performance ?

No, exactly the same (i've checked 2 reviews)

 

[jabbadap]

No, exactly the same (i've checked 2 reviews)

Which is rather good thing. Is there cpu scaling anywhere?

 

[illli]

Interesting, my thoughts on the rtx/non-rtx images:

most non-rtx images look better than the rtx ones. while some rtx might be more 'accurate' some are way too dark to see anything.
some rtx images look nearly indistinguishable from the non-rtx ones.
a few rtx images look better than the non rtx ones.

maybe in a few years it'll look better once game studios have more experience working with it.

 

[Markosz]

Well well well, just before reading through this I watch the DLSS Port Royal Benchmark video NVIDIA posted a week ago.

The performance gain is really nice and much welcome (But not as great as in the bencmark), but also there was no image quality improvement which they highlighted in that video.
In fact the whole image was blurrier, which brings up an interesting question. Since DLSS is machine learning based, the more learning it does, the results get better.
They either did tons of machine learning for the benchmark, hence it's results, or they purpusefully cause image quality loss for the 'no DLSS' image.

But either way, how are we supposed to know how long will they support a game with this? The results for each game might be very different and some might be even worse quality than DLSS off.

 

[champsilva]

Which is rather good thing. Is there cpu scaling anywhere?

Same.

 

[ArbitraryAffection]

What you expect from Nvidia sponsored title?? Poor performence on Readons.

yeah it seems fishy here honestly I suspect 4A may have been drinking some green poison when they implemented RTX....

 

[metalfiber]

For once, glad to see it's not just a console port.

 

[nickbaldwin86]

yeah it seems fishy here honestly I suspect 4A may have been drinking some green poison when they implemented RTX....

Yeah how DARE they use the most current technology to make an amazing game with AMAZING graphics.

NV is paving new roads with new graphics technology. while AMD has a new card out that is a refresh of old and has NO new tech on board. hold your breath and turn blue doing so because AMD will have RayTracing of some kind by 2020

 

[ArbitraryAffection]

Yeah how DARE they use the most current technology to make an amazing game with AMAZING graphics.

NV is paving new roads with new graphics technology. while AMD has a new card out that is a refresh of old and has NO new tech on board. hold your breath and turn blue doing so because AMD will have RayTracing of some kind by 2020

They can implement RTX and they can implement RTX AND exclusde GCN specific optimisations, you know, to cripple performance on Radeon cards? I wouldn't put it past NVIDIA to push something like that.

Also Guru3D have Radeons doing MUCH better, I'm so confused:

 

[W1zzard]

They either did tons of machine learning for the benchmark

3DMark is running on rails, so it's easy to optimize for, because only a few ten thousand of images are shown for the full benchmark, no user control

 

[SystemMechanic]

TPU. Please fix ur benchmarks. First Anthem and now this...how is guru3d getting 20fps lower at 4k than yours ???

 

[ArbitraryAffection]

TPU. Please fix ur benchmarks. First Anthem and now this...how is guru3d getting 20fps lower at 4k than yours ???

I guess different areas of the game tax different bits of the GPU differently or something. I'm just confused that TPU has the 3GB 1060 beating 580 when Guru3D has it a hair off the 1070.