Final Fantasy XV: Windows Edition Benchmark Thread!

[Justinus]

Heeeeeeeeeeeeeeeeeeeeeeeeere we goooooooooo!

Run the benchmark on 1080p high - 4k high if you're extra dangerous.

AMD GPU USERS: Rerun again after the first run, shaders compile the first time and performance increases on subsequent runs.

There is a way to exploit the benchmark and generate artificially high scores. Please do not post these, leave the thread to legitimate scores only.

http://benchmark.finalfantasyxv.com/na/

Looks like the results page intends to put the category at the top of the page, but does not do that.

5960x at 4.7GHz
32GB DDR4 3200 CL 14
1080ti at 2113 core/12500 mem

http://benchmark.finalfantasyxv.com...73d6b30a247&Resolution=1920x1080&Quality=High

http://benchmark.finalfantasyxv.com...73dbc475f85&Resolution=3840x2160&Quality=High

 

[Muhammed]

Having scores instead of fps sucks

 

[alcoholbob]

Ditto, no context for anything.

 

[IllogicalGlory]

1080p high:

GTX 1060 6GB: 4605 (138)
RX 480: 3328 (100)

 

[tamz_msc]

There are a couple of Gameworks settings that are turned on by default on the high preset, while the medium preset turns them all off.

 

[Justinus]

1080p high:

GTX 1060 6GB: 4605 (138)
RX 480: 3328 (100)

You have to re-run any AMD GPU a second time, the first run compiles shaders. It should be more in line with the 1060 after that.

There are a couple of Gameworks settings that are turned on by default on the high preset, while the medium preset turns them all off.

The aforementioned shader compiling is AMDs way of dealing with those Gameworks features, performance is much better after the first run.

 

[Aikouka]

If you're curious, Gamers Nexus posted a video on the benchmark that look into a few aspects of the benchmark in regard to CPU and GPU.

 

[dogen1]

The aforementioned shader compiling is AMDs way of dealing with those Gameworks features, performance is much better after the first run.

Seems unusual, but if you're seeing higher scores on subsequent runs I'll have to take your word for it.

 

[Muhammed]

Scores converted into fps from ComputerBase

https://www.computerbase.de/2018-02...diagramm-final-fantasy-xv-benchmark-1920-1080

 

[Hitman928]

GN results at medium are pretty different than CB, but from the limited amount of online reported results I've seen, I think the CB results are more accurate. Unless "standard" settings was lost in translation and doesn't mean "medium".

 

[Hi-Fi Man]

GN results at medium are pretty different than CB, but from the limited amount of online reported results I've seen, I think the CB results are more accurate. Unless "standard" settings was lost in translation and doesn't mean "medium".

The bench is only in English and Japanese so no translation was done, not sure why they wrote the graph in English and then used the word standard instead of medium.

3221 (UHD, HDR, 12bit, high) on a Core i7 5960X @ 4.5GHz with 2.8GT/s quad channel RAM and a GeForce GTX 1080 @ 2.1GHz/10.8GT/s.

Interesting note, the battle scene is dynamic and changes slightly each run.

 

[Crono]

1950X at 4GHz, 1070 Ti at 2025MHz core/ 4151MHz memory.

1080p, High:

4K, High:

 

[ZGR]

If the battle scene is different each time, should we run multiple tests and take the mean score?

 

[Justinus]

If the battle scene is different each time, should we run multiple tests and take the mean score?

My personal experience has seen very consistent scores between runs, like within 0.5% back to back. I know GN says theres quite a variance, but I am only concerned about AMD GPUs having a raw first run due to the shader thing.

 

[Crono]

If the battle scene is different each time, should we run multiple tests and take the mean score?

Benchmark is long enough that there doesn't seem to be that much variance in score running multiple passes (except for aforementioned AMD/Gameworks reason).

 

[SPBHM]

http://benchmark.finalfantasyxv.com/result/

seems very well optimized for Nvidia, even Kepler and Fermi are performing pretty well compared to AMD cards.

 

[96Firebird]

i5-8600k @ stock
GTX 1080 Ti FTW3 @ stock
16GB DDR4-2800

4K, High

 

[Muhammed]

https://www.gamersnexus.net/game-bench/3223-ffxv-gpu-benchmark-technical-graphics-analysis

 

[Alpha One Seven]

There's a XV in that old series?? When will they let it die and create something new?

 

[SteveGrabowski]

Running this benchmark at 1080p high with my GTX 970 + Xeon E3-1231v3 + 16GB DDR3-2400 was a mess. It seemed like it was just above 30 fps a lot of the time and there were some really nasty stutters which would make the game unplayable even with a 30 fps lock. On 1080p standard I was mostly around 60-70 fps but with lots of dips to around 50 fps. For what it's worth my 970 gets about 2% higher score than the 970 score they have in their graph. Pretty disappointing to think I have to start running games at medium, though I guess it makes sense with a 3.5 year old card. Still, I'm not especially impressed by FF XV's graphics at standard.

 

[tamz_msc]

https://www.gamersnexus.net/game-bench/3223-ffxv-gpu-benchmark-technical-graphics-analysis

So Gameworks makes it unplayable for those with high refresh-rate monitors.

 

[Muhammed]

So Gameworks makes it unplayable for those with high refresh-rate monitors.

I guess it does, but IMO the increased shadowing, particles and tessellation greatly enhances the image quality.

 

[tamz_msc]

I guess it does, but IMO the increased shadowing, particles and tessellation greatly enhances the image quality.

Eh, the game looks great on the Standard quality preset. The only things that you immediately notice on the High preset are increased shadow LOD and more grass.

 

[SteveGrabowski]

Eh, the game looks great on the Standard quality preset. The only things that you immediately notice on the High preset are increased shadow LOD and more grass.

The pop in is terrible on standard quality IMO. Not surprising considering Square Enix's history with PC ports though. Like the older FF remasters that run at 30 fps or NieR Automata that runs at subnative resolution without using the FAR mod.

 

[tamz_msc]

The pop in is terrible on standard quality IMO. Not surprising considering Square Enix's history with PC ports though. Like the older FF remasters that run at 30 fps or NieR Automata that runs at subnative resolution without using the FAR mod.

Well one can hope that the game allows for adjusting LOD without being tied to GameWorks settings in the final release.