Death Stranding: PC graphics benchmark perf analysis review

I have no intention of getting this boring looking game, but this game has a really awesome DX12 implementation. Guerilla Games should be proud!

 

was waiting for this game for a long time, then I saw "win 10", dx12, sigh...

 

It's really interesting how well the 390X is still doing, leaving behind the Polaris cards. But then again, I always thought the Polaris specs were quite limited, like having only 32 ROPs. On the other hand I suppose Fury is hurt by the lack of memory, keeping it behind the 390X.

5500XT really jumped with the driver update, it seems. 5700XT also again made Radeon VII look worse.

 

No TAA = Jaggies everywhere. If it's properly implemented there is no comparison. DLSS 2.0 is a better alternative, but until Nvidia finds a way to control-panel it in every game with TAA, I would take TAA + a sharpening filter over any shimmering.

 

390X is still a beast at lower mid range level. If it had better tech like higher dp hdmi version plus hevc, h265, 4k it would be an undying gpu.

 

Polaris is also bandwidth limited so 390x brute force it in this game. Still, that happens rarely 580 is faster in most games.

 

So expecting around 70-80FPS at 3440x1440 on a 5700XT.

 

Thanks for the review.
It is a good looking game and it does seems to be very well optimized.

How is the CPU core/threads scaling? It seems that this game does not need much CPU power.
---

I think there is a small error in the Table of Contents (dropdown list): It says "Image Quality and System Requirements", but I think it should be "Image Quality and Benchmark System".

 

Hi @Hilbert Hagedoorn . First, thank you for the review, great job!

While reading the article I had just a doubt related with your testing methodology, and perhaps you could clarify a bit more this point:

Please, could you share and detail all the specific energy-saving functions you disabled for such mobo and processor? I have same processor and chipset platform and I'm interesting on benchmarking subjects, so if could clarify this point I'd highly appreciate it.

Does it mean you disable those (not specified yet) energy-saving features and set a fixed working clock speed for all cores too? Like setting a fixed all cores 4.7Ghz freq or just disable the energy-saving features and leave the rest at stock? Should I assume you didn't apply any level of CPU OC in order to ensure consistent benchmarks results as well?

Thanks in advance! Kind regards

 

Big thanks and respect to Hilbert for adding GTX970 to the test per my request!! Really appreciated.

Now... 64fps on maxed 2020 game in 1080p (Thats 14% faster than GTX 1650 super) - thats what Im talking about and thats why I still have this card, it runs 99% of the games at very high settings at 60fps. I have 60hz monitor, so what more do I REALLY need? I use MSI afterburner and Im using PC since it was created, so I can fine tune the games really easily to run great and look great. Also, gtx970 is a really good overclocker, 10%+ performance is not a problem and thats almost gtx980 teritorry.

Only reasonable alternative upgrade card for me is RX 5700 XT, I keep an eye on it, its the 1080ti basically, with more modern design (RDR2 is the example) and you can actually buy it new in the shops, unlike the 1080ti.

But I will indeed wait for the new incoming cards, than I might upgrade.

Anyway. Thanks GTX 970 OCed for life

 

I've still got the 290, which is basically just a slightly slower version of the 390X, and it surprises me how well it has aged. I intend to keep it until I switch to 4K.

 

For a true 4K experience and/or high refresh rates for 1440p, we need a gpu released fully in 2020 or 2021. Newer monitors will be capable of 4k 120hz and 1440p 240hz. There is no power available at the moment. By the end of the year it will be logical to drop gpus released 5 or more years ago.

 

Sounds like things have been changing at AMD RTG over the past year quite... a ... bit.... thanks Hilbert

Had one in a build until 3 months ago... this was for - was it almost 6/7 years already ??

 

Dude, the difference between the 4GB and 8GB cards with that GPU are not huge, like in so many other games.

Only an issue if you got slow arse PC to begin with, never had an issue fully loading the 970 I had.

In most games DLSS looks actually worse than TAA with sharpening used at 75-100% native resolution.

 

Except for, you know, games that use over 4GB of VRAM. Where GN found it could make almost a 30% difference.

Isn't the 970 part of the slow PC? lol...

In DLSS 1.0, sure - in 2.0 nope.

 

Ok dude. I saw the VRAM consumption and thought the 4GB would have an affect.

 

Games using more than 4gb vram even at 1080p looks like standard now. You can imagine those 1% lows for the 970.

 

Imagine 970 gpu and Ryzen cpu. *Imagine those 1% lows
Perfect combo for VERY 1% lows

 

I think ryzen is doing fine. It does not need a chiller to cool it like most intel cpus. I can live with few fps less at 1080p.