Death Stranding Benchmark Test & Performance Analysis

[pbupm]

DLSS2.0 makes this game better on nvidia anyway.

RTX 2060 Super can play this game 60+fps on 4K DLSS quality mode



I would not recommend upgrading just for DLSS2.0 because not many games support DLSS2

I would wait for RTX 3080. Turing is not worthy upgrade over 1080ti

i wont upgrade to turing but Ampere, i know that ampere is around the corner.

 

[wtfbbqlol]

I would not recommend upgrading just for DLSS2.0 because not many games support DLSS2

I would wait for RTX 3080. Turing is not worthy upgrade over 1080ti

Not sure what the resale value is for 1080Ti these days, but I definitely thought my "sidegrade" was worth it in 2019. I sold my 1080Ti for $500 and bought the 2070SUPER for $550. So I basically get the same non-RT performance (within 10%), while getting a good experience in RT-enabled games like Control, Wolfenstein Youngblood, and Quake 2 RTX (I just love it).

But today, yeah I would wait for the 30-series RTX cards.

 

[Ashtr1x]

Great write up. As for the game itself, the art style looks cool and innovative, the HUD and how options / menu design (DOOM Eternal downgraded a lot vs DOOM2016 for eg and looks meh vs DS). Decima Engine as mentioned looks beautiful and for environments it works great but for the characters it falls short. RE Engine does it much better, the movement is really clunky from the videos of gameplay.

The most horrible part is, this game is like a movie. Literally 30minutes of cutscene right when player starts the game and so many cutscenes all the time, if we watch gameplay, for more than 15 hours, one does the same thing, delivering packages like a post apocalyptic Fed Ex man. Too much hype and way too much of expectations., this game blew a lot of its budget to that hollywood caste, marketing unfortunate, wish they made this into a great action title or such.

 

[harm9963]

There are huge differencies in frame rates in your review vs one on guru3d?

Saw that to !

definitely different test scene, did they use the latest drivers?

latest driver 451.67 and AMD as well.

 

[kardeon]

No way you can see a substantial part of the game in the first 2 hours. And you’re not missing much. I would have refunded too

Of course 1 hour is not enough but if a game cannot captive your attention the first hour, it is already a bad start.
I made the mistake with RDR2 and now i have this boring game (with old gaming mechanics) in my librairy. I played only 6 hours and lost 60 euros.

 

[wolf]

DLSS is really starting to prove it's worth, 1.0 was promising from a performance perspective but lacked in the image quality, now 2.0 give better visuals and better performance (on quality mode), and substantially boosts performance on performance mode with similar or slightly degraded visuals.

If this really is so 'easy' for Dev's to include, a card capable of DLSS will be a must have for my next GFX card purchase.

Nvidia gambled on RTRT and DLSS 2 years ago and I think the payoff is clear to see.

Although the results are far more 'meh', it's nice to see AMD put in some time to make a workable and easily toggled in game settings solution to boost FPS. It's something anyone could do themselves if the game ships with a resolution scale slider but yeah, nice to see something nice and easy thrown in.

 

[watzupken]

At the get go, I feel there is too much hype about this game. And it seems true that its only air, but little substance. Graphically, I think it is good, but gameplay, its not for me.

DLSS 2.0 works very well here, I feel its primarily because of the extensive collaboration between the game maker and Nvidia. For the real test of DLSS 2.0 results, ideally we should be just randomly throwing a game and see if the machine learning/ AI cores can significantly improve IQ on the fly. That is how I envision machine learning to do, and not through some heavy optimization in the back end.

Not sure what the resale value is for 1080Ti these days, but I definitely thought my "sidegrade" was worth it in 2019. I sold my 1080Ti for $500 and bought the 2070SUPER for $550. So I basically get the same non-RT performance (within 10%), while getting a good experience in RT-enabled games like Control, Wolfenstein Youngblood, and Quake 2 RTX (I just love it).

But today, yeah I would wait for the 30-series RTX cards.

The unfortunate fact is, you need to be using the latest Nvidia cards in order to get the best performance. In other words, by the time the RTX 3xxx gets released, do expect some features to either not be supported, or perform subpar on the RTX 2xxx.

I don't believe a GTX 1080 Ti is any slower than say a RTX 2070 Super. In this game, I see it has fallen behind fairly significantly, which I attribute to Nvidia not bothering to optimize drivers for Pascal or older cards. The good thing is at least the drivers for these older cards are still stable.

 

[kinjx11]

here's my benchmarks of the game , 4k 8k DLSS TAA FACS all of them

8k

4k

i think my setup is doing a lot better than TP's

 

[TKnockers]

Saw that to !


latest driver 451.67 and AMD as well.

They tested radeons with 20.7.1... and then retested only 5700, 5700xt with 20.7.2. They saw no difference with rx580 20.7.2 I think so they didn't retest other radeons with that latest driver.

 

[the54thvoid]

They used the latest ones for 5700xt... but even with 20.7.1 all radeons have higher fps..?

That's selective bias you're using there.

1080p

TPU - 5700XT = 132.9
Guru - 5700XT = 153


TPU - 2080ti = 157.6
Guru - 2080ti = 187


Nividia have higher fps as well at Anand, sorry Guru. So, across both the reds & greens, Anand Guru has higher fps. You said it as though only the radeons were lower on TPU. Not the case. As W1zz says, different scene.

 

[TKnockers]

That's selective bias you're using there.

1080p

TPU - 5700XT = 132.9
Anand - 5700XT = 153


TPU - 2080ti = 157.6
Anand - 2080ti = 187


Nividia have higher fps as well at Anand. So, across both the reds & greens, Anand has higher fps. You said it as though only the radeons were lower on TPU. Not the case. As W1zz says, different scene.

Not sure about anand i was talking about guru3d.. not biased, had many gpus in the last 25 years of gaming...navidias and radeons. Had great gpus from both teams in their own time. From simple observation that frame rates differ and what drivers have been used came a whole debate about someone being biased because both teams have lower frame rates on a website I didn't even mention at all. Not everything is meant to be "hostile"

 

[the54thvoid]

Not sure about anand i was talking about guru3d.. not biased, had many gpus in the last 25 years of gaming...navidias and radeons. Had great gpus from both teams in their own time. From simple observation that frame rates differ and what drivers have been used came a whole debate about someone being biased because both teams have lower frame rates on a website I didn't even mention at all. Not everything is meant to be "hostile"

Meant Guru - used pics from post #29 to see fps.

I picked up on the specific words you used, wanted to emphasise both teams have worse scores at Guru. You did focus on Radeon's having higher fps with Guru. And, as you've been around a long time, you'll know that's often the first shot in a snipe about review bias. But I'm glad that's not the case. FWIW, Guru's fps have always been way higher than W1zz's reviews. I know a lot of sites use a specific benchmark. W1zz does use gameplay (similar section of game) to test 'real-world' performance.

 

[W1zzard]

W1zz does use gameplay (similar section of game) to test 'real-world' performance.

I also try to find "demanding" scenes to test, not worst case, but definitely won't use something that's "light", but easy to bench.
I do play the game for several hours (which often means I'm not the first one posting), but that gives me a better feel for performance across several maps, and also gameplay, and fun .. well wasn't fun this time t.t

 

[bubbleawsome]

Wonder if nvidia will think it's worth it to allow DLSS at native. Running 1440p DLSS on 1440p to get rid of aliasing but without the blur. Or is this something you can already do?

 

[TKnockers]

Meant Guru - used pics from post #29 to see fps.

I picked up on the specific words you used, wanted to emphasise both teams have worse scores at Guru. You did focus on Radeon's having higher fps with Guru. And, as you've been around a long time, you'll know that's often the first shot in a snipe about review bias. But I'm glad that's not the case. FWIW, Guru's fps have always been way higher than W1zz's reviews. I know a lot of sites use a specific benchmark. W1zz does use gameplay (similar section of game) to test 'real-world' performance.

I never understood the whole fan base thing .. whichever piece of hardware performs better for the money that's what I buy. If I had some of their stocks then I'd root for certain team but as long as I work my ass off to be able to buy these toys I'll go for the most rewarding solution. Btw I love both sites, tpu and guru and have been reading both sites' reviews for a long time. Keep up the good work guys! Regards from Zagreb

 

[Dredi]

wait for rtx3000

I hope they keep working on it.If this is what 1st gen turing can do,delivering more detailed picture than native with huge peerformance gains,then imagine what happens with better cards and better support.

I wish presets were free to choose tho.Nothing seems like an obstacle and current segmentation is artificial.Only at 4K I can choose dlss to upscale from 1080-1440p.at QHD I'm stuck with 960p.I wanna be able to upscale from 1080p image at 1440p.

View attachment 162402

You should be able to set up DSR in nvidia drivers. That way you can effectively set the DLSS internal resolution to whatever you want. For example DSR 4x with DLSS performance will have the internal resolution at native resolution. In your case it would render at 2560x1440, scale it up with DLSS performance to 5k and then downscale with DSR back to 2560x1440. I don’t have a RTX card to test this with, but please do tell if it does not work!

 

[bobsled]

In all honestly, the conclusion is odd. The review title of "Death Stranding Benchmark Test & Performance Analysis" indicates you'd be benchmarking performance, not providing a subjective piece of what you think of the game, spoilers included.

 

[W1zzard]

not providing a subjective piece of what you think of the game

Yeah a bunch of articles ago I started with some basic thoughts on gameplay, people kept asking for more.. it actually is somewhat refreshing to write a different kind of review from time to time that's not hardware

spoilers

Where?

 

[dayne878]

Yeah, the game looks nice, but I have no interest in the premise of the game at all, and your helpful article just confirmed that.

 

[the54thvoid]

Yeah a bunch of articles ago I started with some basic thoughts on gameplay, people kept asking for more.. it actually is somewhat refreshing to write a different kind of review from time to time that's not hardware


Where?

Maybe when you revealed it was 'Backpack Walking Simulator 2020'

One off my list.

 

[W1zzard]

Maybe when you revealed it was 'Backpack Walking Simulator 2020'

That's not a spoiler

 

[the54thvoid]

That's not a spoiler

Oh yeah, I wasn't being serious. Though I genuinely didn't know that mechanic was a thing.

 

[Bee9]

Maybe when you revealed it was 'Backpack Walking Simulator 2020'

The author said it's a walking simulator game.

I've just installed it and played for 90 mins. The environment looks really nice. Gameplay feels like I'm watching 3D movie, so I refunded the game. I'd rather watch youtube videos from other people playing it though. The story caught my attention.

 

[nguyen]

Wonder if nvidia will think it's worth it to allow DLSS at native. Running 1440p DLSS on 1440p to get rid of aliasing but without the blur. Or is this something you can already do?

Yes so can use Dynamic Super Resolution (DSR) at 2.25x to make DLSS use native resolution to upscale it then downsampling back to native res.
If you have a 1080p screen it's better to do this because DLSS is not as good when rendering 1080p compare to 1440p and 4k.

 

[cucker tarlson]

If you have a 1080p screen it's better to do this because DLSS is not as good when rendering 1080p compare to 1440p and 4k.

watch from the start up until 11:00.