Resident Evil 7 Demo Benchmarks

[Bacon1]

Fury X vs 1070 FE vs 1080 FE:

https://www.youtube.com/watch?v=quskymgJAYA

 

[Head1985]

Furyx aging very good.Its faster than GTX1070 again.And that is only 1080p..in 1440P the gap will be even greater.
I think in 2017 we will see similar situation with Furyx and GTX1070 like 390x vs gtx970
GTX1070 is just GTX980TI with far worse OC headroom so we can expect furx will beat GTX1070 sooner or later.

 

[bystander36]

Furyx aging very good.Its faster than GTX1070 again.And that is only 1080p..in 1440P the gap will be even greater.
I think in 2017 we will see similar situation with Furyx and GTX1070 like 390x vs gtx970
GTX1070 is just GTX980TI with far worse OC headroom so we can expect furx will beat GTX1070 sooner or later.

Are you referring to Bacon1's post, your post is showing a 970 vs 390. Bacon1's post shows a single, unreleased game.

 

[Carfax83]

There's a new patch out that supposedly increases performance considerably when HBAO+ is enabled, and also enables support for CPUs without SSE4.1.. Can't believe they support CPUs that old?!

 

[psolord]

I just finished uploading my benchmarks if anyone is interested.

The benchmark consists of almost the whole duration of the demo. SMAA used for all.

V.high (-) means high shadows for the 970 and 7950, because they couldn't handle very high shadows setting.

970 and 1070 benches were made with the new patch. Didn't notice any performance improvement on the 970 in comparison to the initial release, but the new patch may have improved the Radeon's performance. The delta seems to be higher than it usually is, so HBAO+ may have handicapped the 7950's performance, which was still quite playable. Will test again later, since it's 970's turn to run benchmarks on, this period.

Resident Evil 7 Biohazard 1920x1080 v.high GTX 1070 @2Ghz Core i5 2500k @4.8GHz - 166fps

Resident Evil 7 Biohazard 1920X1080 v.high(-) GTX 970 @1.5Ghz CORE i7-860 @4GHz - 121fps

Resident Evil 7 Biohazard 1920x1080 v.high(-) 7950 @1.1Ghz CORE i7-860 @4GHz - 70fps

As I said, GTX 970 and Radeon 7950 had to be run with shadows at high, because using very high shadows, made the game come to a crawl on both cards. I also made the following video to illustrate the issue.

Resident Evil 7 : Biohazard - Quick tip, don't use very high shadows


Thankfully, the GTX 1070 couldn't care less about shadows setting.





Here are the respective frametimes percentile graph from all runs.

 

[dogen1]

So, I guess the interlacing option is like killzone? Wonder why they went with that instead of a checkerboard rendering pattern. Wish there were more details about it somewhere.

 

[Thinker_145]

Fury cards should never be compared with reference cards. It's completely BS really.

 

[Raduque]

I have everything set to max and my lowest FPS was 64, highest was 98. I just wandered through the house looking around and watching the FPS counter in from Steam.

i7-4720HQ
GTX970m

edit: Shadows were on High. Very High gets me console-level performance - 32-40FPS.

 

[Head1985]

GTX980TI beat GTX1070 in this game and GTX980 is faster than GTX1060.GTX1080 is +- 30% faster than GTX1070(again..)

I really want know how in reviews is GTX1070 only 20% slower than GTX1080 when in last 6months almost in every new AAA game the gap is close to 30%.

 

[Thinker_145]

GTX980TI beat GTX1070 in this game and GTX980 is faster than GTX1060.GTX1080 is +- 30% faster than GTX1070(again..)

I really want know how in reviews is GTX1070 only 20% slower than GTX1080 when in last 6months almost in every new AAA game the gap is close to 30%.

Cuz you conveniently ignore the situations where the difference is less than 20%. You can go look at the individual games in the TPU charts yourself.

 

[Head1985]

TPU have half games very old.I am talking about recent 6 months.
Deus ex 27%
doom 26%
mafia3 27%
BF1 22%
Other games are old.Games like COD IW, dishonored2, GOW4 have close to 30% gap.

 

[tg2708]

GTX980TI beat GTX1070 in this game and GTX980 is faster than GTX1060.GTX1080 is +- 30% faster than GTX1070(again..)

I really want know how in reviews is GTX1070 only 20% slower than GTX1080 when in last 6months almost in every new AAA game the gap is close to 30%.

really good showing for the 980ti since it clocks are lower than what most can reach so a few extra frames is left to be had

 

[Head1985]

really good showing for the 980ti since it clocks are lower than what most can reach so a few extra frames is left to be had

Yeah but 1200Mhz for 980TI is very high because in reviews they run at 1050mhz.Thats why GTX1070 is little faster than them.Aftermarket 980TI is always faster than GTX1070.And when both OC to max the gap is even greater.

As you can see here GTX1070 is faster than GTX980TI=GTX980TI must run way bellow 1100Mhz to be slower than GTX1070.Thats why i dont like any reviews that uses only reference cards.

 

[Majcric]

From my understanding, the 980ti's boost around 1200 right out the gate. (i know the reference one I have does) 1400 and beyond is what I would personally consider a very high overclock.

It should probably be noted that since the patch those charts are moot and performance has increased significantly.

 

[Thinker_145]

At the time of the 1070 release there was a max vs max OC comparison with the 980Ti and it was a tie.

 

[Thinker_145]

TPU have half games very old.I am talking about recent 6 months.
Deus ex 27%
doom 26%
mafia3 27%
BF1 22%
Other games are old.Games like COD IW, dishonored2, GOW4 have close to 30% gap.

They only use 2015 and 2016 games which are absolutely relavent.

 

[AtenRa]

Ok im currently exploring under-voltage with my XFX RX 480 GTR Black

XFX RX 480 GTR Black
1105MHz with 0.937v
Memory 2000MHz

System specs
Core i7 3770K @ 4.4GHz
4x 4GB Kingston DDR-3 2133MHz 1.65v
Win 10 pro 64bit
Crimson 16.12.2

Bellow is the GPUz and Fraps from the benchmark (same sequence as gameGPU),

 

[Bacon1]

Ok im currently exploring under-voltage with my XFX RX 480 GTR Black

What's the power usage difference between that and stock voltage? Any difference in performance?

 

[Keysplayr]

Ok im currently exploring under-voltage with my XFX RX 480 GTR Black

XFX RX 480 GTR Black
1105MHz with 0.937v
Memory 2000MHz

System specs
Core i7 3770K @ 4.4GHz
4x 4GB Kingston DDR-3 2133MHz 1.65v
Win 10 pro 64bit
Crimson 16.12.2

Bellow is the GPUz and Fraps from the benchmark (same sequence as gameGPU),

What for? You post this in the wrong thread?

 

[Bacon1]

What for? You post this in the wrong thread?

No, he is showing you his game performance in the same segment that gamegpu did.

 

[AtenRa]

What's the power usage difference between that and stock voltage? Any difference in performance?

Here is the same system, same game sequence with XFX RX480 GTR Black at default 1338MHz clocks and voltages.

Total system peak power usage was 312W and GPU Chip at 163W.
Average fps was 123 and minimum at 94

If we compare with the under-volted under-clocked version we have

Average fps = 33,44% faster for the default clocks
But the GPU Chip used 93,24% more power (close to double)
and
Total System power was 222W (for the under-volted) vs 312W for the 1338MHz or 40,5% more

 

[Bacon1]

Here is the same system, same game sequence with XFX RX480 GTR Black at default 1338MHz clocks and voltages.

Interesting thanks

So 1105 vs 1338 and 0.9375 vs 1.20 right?

21% higher clocks / ~17% reduction in clocks but 33% perf difference is crazy.

I think your clocks were actually lower though in the original test, looking at the GPUZ graph it has a reduction ~25% into the test. So that 1105 would have only been the peak? Might want to retest that one and see what the dip goes down to, wonder what clocks its running at there. Guessing closer to 870-900?

This kinda thing is what Radeon chill is supposed to be all about though, as you can obviously downclock a ton and still get great performance from this game.

Whats the lowest voltage you can go while maintaining 1300+ clocks?

 

[AtenRa]

Interesting thanks

So 1105 vs 1338 and 0.9375 vs 1.20 right?

21% higher clocks / ~17% reduction in clocks but 33% perf difference is crazy.

I think your clocks were actually lower though in the original test, looking at the GPUZ graph it has a reduction ~25% into the test. So that 1105 would have only been the peak? Might want to retest that one and see what the dip goes down to, wonder what clocks its running at there. Guessing closer to 870-900?

This kinda thing is what Radeon chill is supposed to be all about though, as you can obviously downclock a ton and still get great performance from this game.

Yea i believe i had power at -50% when i run the undervoltage settings, that is why it didnt keep the clocks at 1105MHz all the time

Whats the lowest voltage you can go while maintaining 1300+ clocks?

At 1300MHz it seems to be stable at 1.1V, but i have to do more testing to verify this.

What is strange is that at 1300MHz and 1.1V the performance is equal with the default at 1338MHz 1.2V.
Nice power reduction without performance loss, at least in this game.

1300MHz 1.1V Power at 0

 

[Bacon1]

What is strange is that at 1300MHz and 1.1V the performance is equal with the default at 1338MHz 1.2V.
Nice power reduction without performance loss, at least in this game.

Cool thanks for testing again , yeah that is a solid drop from undervolting. I've noticed that AMD tends to go with high voltage and most people can undervolt w/o losing performance. I do on my Fury as well. This game looks like it will have solid performance at launch with how well the demo runs.

 

[Headfoot]

What for? You post this in the wrong thread?

Yeah, his benchmark run of Resident Evil 7 Demo definitely doesn't belong in this thread about Resident Evil 7 Demo benchmarks.