Resident Evil 7 Benchmarks

Ian Evey · 26 Jan 2017 at 17:53

I don't understand why they don't leave it on for amd cards that gain fps and likewise for Nvidia and turn it off if any card gets lower fps.

Seems the only fair way imo, maybe the dev has the answers why a 390x beats a 1080 with it on?

Silent_Scone · 26 Jan 2017 at 17:54

maonayze said:
Shader Cache/Shadow Cache whatever...just semantics. Is that all you got left Silent?

No plenty more, it's just fun watching the vitriol. Let's look at somethings in the review, shall we...

Now we look at the Radeon RX 480 8GB - as you can see, extremely smooth. That is just very impressive for a 239 USD card !

Hardly what one would say when tying to put a vendor in bad light.

When take the plot results for the two modes on the Fury and transition them towards relative FPS, yes ... that was causing the massive drop-off in performance. As you can see at the 50% you'd have your average framerate. With Shadow Cache on that would be say 70 FPS. With Shadow cache off we are passing 80 FPS. Also look at the end of that orange line, a huge drop-off, that's low framerates as a results of the stutters.

Objectively showing how the FuryX (AMD's flagship card) benefits from having Shadow Cache disabled.

But most importantly;

** Update #1: Nvidia released a new driver which massively impacts performance. The results have been updated but we'd not be surprized to see something similar happening with AMD drivers. hence the results today remain subjective. The game is incredible hard to measure objectively.

** Update #2:

After lots of weird results with graphics card that have less then 4 GB of VRAM we found out that one problematic setting is the Shadow cache.

If you have an up-to 4 GB graphics card, please turn it off.
However if you have a 6GB or 8GB card, leave the shadow cache activIn closing,
ated.

There have been an unfortunate three revisions of the benchmarks with results all over the place. The last batch (revision 3 as you see them) are valid. The Shadow Cache settings disabled offers the best game-play for the majority of cards hence it is the preferred and recommended setting, for 6GB and upwards shadow cache enabled canoffer better results, use it at your own peril. In-between revision 1 and 3 of the review a lot has happened, new drivers 5 minutes after I posted the initial article, totally weird anomalies but also an error on my side that boosted perf on the earlier tests. I think this was the cause of my having interlaced modus activated (mistakenly) on some of the cards as I have been goofing around with that on day 1. Totally my bad, but I am not even 100% sure either that was issue as the game remains difficult to measure. We will retest the game in a few weeks when the drivers on both sides have matured and when the game has had a patch or two.

1) He openly explains where he may have gone wrong in testing (including using interlaced which would explain the jump in frame rate for certain lower end SKU)
2) He explains why shader cache, sorry, Shadow Cache, has a derogitory impact on some cards.
3) He acknowledges due to the setting the game is difficult to review objectively depending on what card you own (including a fued with AMD SKU)

Ultimately, people need to stop throwing slandarous claims if they don't have the nous, nor evidence to back them up.

As the kids put it these days...rekt.

MyBrainz · 26 Jan 2017 at 17:57

Ian Evey said:
I don't understand why they don't leave it on for amd cards that gain fps and likewise for Nvidia and turn it off if any card gets lower fps.

Seems the only fair way imo, maybe the dev has the answers why a 390x beats a 1080 with it on?

Lets turn HBAO+ off in Arma 3 then so the 1060 looks better than the RX 480?

Maxed Ultra should mean EVERYTHING on its highest settings and all other options switch to ON.

The End.

Raelgun · 26 Jan 2017 at 17:59

Silent_Scone said:
No plenty more, it's just fun watching the vitriol. Let's look at somethings in the review, shall we...

Hardly what one would say when tying to put a vendor in bad light.

Objectively showing how the FuryX (AMD's flagship card) benefits from having Shadow Cache disabled.

But most importantly;

1) He openly explains where he may have gone wrong in testing (including using interlaced which would explain the jump in frame rate for certain lower end SKU)
2) He explains why shader cache has a derogitory impact on some cards.
3) He acknowledges due to the setting the game is difficult to review objectively depending on what card you own (including a fued with AMD SKU)

Ultimately, people need to stop throwing slandarous claims if they don't have the nous, nor evidence to back them up.

As the kids put it these days...rekt.

'The Shadow Cache settings disabled offers the best game-play for the majority of cards hence it is the preferred and recommended setting'

If he does this, then his article is no longer a performance benchmark, the article was never designed to be what setting you should play this game at, the article was designed to show what GPU is the best at the game. If you fiddle with non vendor specific settings, then he needs to redo the whole article with the intention in bold in the title to the specifics he wants.

If the intent of the article is to produce the best performance results and, then the article is nothing but a tweaking guide.

Benchmarks is about bringing hardware to their knees even if the average FPS on a Titan XP is 20 FPS, its a VALID score, just means the cards are not good enough for the titles highest setting.

Raelgun · 26 Jan 2017 at 18:01

MyBrains said:
Lets turn HBAO+ off in Arma 3 then so the 1060 looks better than the RX 480?

Maxed Ultra should mean EVERYTHING on its highest settings and all other options switch to ON.

The End.

You are right, however HBAO+ is meant to be a Nvidia specific technology, SSAO is non vendor specific.

Otherwise I do agree with you, Max is Max, there is no Max +1.

Lokken86 · 26 Jan 2017 at 18:02

Ian Evey said:
I don't understand why they don't leave it on for amd cards that gain fps and likewise for Nvidia and turn it off if any card gets lower fps.

Seems the only fair way imo, maybe the dev has the answers why a 390x beats a 1080 with it on?

The 390x has a great architecture still and with 8gb it sees a large benefit here. The game generally favours AMD anyway so not really so surprising. Some people find it hard to embrace that older cards designed for the future on release can still challenge newer cards years down the line. They see a second Updated "review" that shows certain cards back "where they should be" and toss out any legitimate reason for it. After order is restored the thread is typically labelled a joke by anyone opposed to any challenge on the results. It really does get daft after a while.

Silent_Scone · 26 Jan 2017 at 18:03

Raelgun said:
'The Shadow Cache settings disabled offers the best game-play for the majority of cards hence it is the preferred and recommended setting'

If he does this, then his article is no longer a performance benchmark, the article was never designed to be what setting you should play this game at, the article was designed to show what GPU is the best at the game. If you fiddle with non vendor specific settings, then he needs to redo the whole article with the intention in bold in the title to the specifics he wants.

If the intent of the article is to produce the best performance results and, then the article is nothing but a tweaking guide.

Benchmarks is about bringing hardware to their knees even if the average FPS on a Titan XP is 20 FPS, its a VALID score, just means the cards are not good enough for the titles highest setting.

You need to read properly before coming in half cocked. The conclusion explains the performance boost on earlier tests. Ergo you can't compare the first results objectively with the new ones.

Seems this is too much for people to comprehend.

AlamoX · 26 Jan 2017 at 18:04

Gregster said:
Guru was spot on with turning the Cache off in my opinion. It is fair to all cards when it is.

what would have been fair is enabling shadow cache on cards that benefit from it, and turn it off from cards struggling with it.
the feature have 0 impact on visuals contrary to hairwork/hbao+ etc, and not vendor specific.
so i dont see the point of doing all ON or all OFF, especialy when you knock off 40fps from certains cards with 0 visual impact.
so it certainly is not fair

Ian Evey · 26 Jan 2017 at 18:05

MyBrains said:
Lets turn HBAO+ off in Arma 3 then so the 1060 looks better than the RX 480?

Maxed Ultra should mean EVERYTHING on its highest settings and all other options switch to ON.

The End.

No as shadow cache is not a graphical enhancement to make the game look better it's just caching of shadows.

Leave it on if your fps are better or turn it off if it gains you more fps.

MyBrainz · 26 Jan 2017 at 18:06

Raelgun said:
You are right, however HBAO+ is meant to be a Nvidia specific technology, SSAO is non vendor specific.

Otherwise I do agree with you, Max is Max, there is no Max +1.

Well AMD has HDAO in Arma 3 but also can use HBAO too.. but yea max is max is max.. i dont know why there is an argument over this, just seems like some people are getting rather salty over the initial benchmarks.. yes benchmarks that were posted in the first place.

My 480 is a 4GB anyway.. also i wont be buying this game either lol

Lokken86 · 26 Jan 2017 at 18:06

MyBrains said:
Lets turn HBAO+ off in Arma 3 then so the 1060 looks better than the RX 480?

Maxed Ultra should mean EVERYTHING on its highest settings and all other options switch to ON.

The End.

That's true about ARMA but shader cache for this game is a setting for performance not for graphics quality.

maonayze · 26 Jan 2017 at 18:07

Silent_Scone said:
As the kids put it these days...rekt.

Yep totally agree. With both graphs shown (Original and Current) it would have been Nvidia mindshare that would have been rekt. (Okay, dented

)

Most people just skip to FPS and the graphs, which in this case would have shone a different light overall to the majority of users IF the original graph had been left in to go with the explanation. It certainly would have swayed a few looking into getting either a 1060 or a 480/390X

It alters the perception for sure in regard to current mindshare.

As for Guru3D, lets just say they know which side their bread is well and truly buttered on.

Raelgun · 26 Jan 2017 at 18:09

Silent_Scone said:
You need to read properly before coming in half cocked. The conclusion explains the performance boost on earlier tests. Ergo you can't compare the first results objectively with the new ones.

Seems this is too much for people to comprehend.

He literally says with it off, offers the best setting for most GPUs, telling people in a PERFORMANCE BENCHMARK to change the setting to get the most FPS by turning stuff off no longer makes it a PERFORMANCE BENCHMARK.

Tweaking settings to produce the highest FPS by turning stuff off becomes a tweaking guide, Guru3D has failed quite an extent on how to conduct a performance benchmark. Otherwise hes produced a good tweaking guide here.

MyBrainz · 26 Jan 2017 at 18:10

Ian Evey said:
No as shadow cache is not a graphical enhancement to make the game look better it's just caching of shadows.

Leave it on if your fps are better or turn it off if it gains you more fps.

Like i said, max is max.

He should have just done all cards with the cache ON and then all again with the cache OFF, then graphed them accordingly, with a page explaining about Shadow Cache, details of its use and how it affects VRAM and performance.

Pretty bloody simple really

humbug · 26 Jan 2017 at 18:10

Some are still trying to muddy what is very simple.

The performance on the 390X is at least = to a GTX 1080 with the Cache on.
With the Cache off the 390X lost 48 FPS while the GTX 1080 only 5.

Guru3D removed the review with Cache on, gone, out of sight completely and replaced it with a review with Cache off.

No user on those cards would turn cache off as it gimps performance, IMO in conclusion Guru3D removed the cache on review because nVidia told them to.
No other reason would Guru3D tell users to use a settings that is worse on both sides, especially deny 2/390X users a massive performance uplift.

Nexus18 · 26 Jan 2017 at 18:12

Some people just keep on....

Priceless!

Lokken86 · 26 Jan 2017 at 18:13

Raelgun said:
He literally says with it off, offers the best setting for most GPUs, telling people in a PERFORMANCE BENCHMARK to change the setting to get the most FPS by turning stuff off no longer makes it a PERFORMANCE BENCHMARK.

Tweaking settings to produce the highest FPS by turning stuff off becomes a tweaking guide, Guru3D has failed quite an extent on how to conduct a performance benchmark. Otherwise hes produced a good tweaking guide here.

Yes and test Doom only in OpenGL since that is the most taxing. You can't fault that kind of logic.

Silent_Scone · 26 Jan 2017 at 18:17

maonayze said:
Yep totally agree. With both graphs shown (Original and Current) it would have been Nvidia mindshare that would have been rekt. (Okay, dented )

Most people just skip to FPS and the graphs, which in this case would have shone a different light overall to the majority of users IF the original graph had been left in to go with the explanation. It certainly would have swayed a few looking into getting either a 1060 or a 480/390X

It alters the perception for sure in regard to current mindshare.

As for Guru3D, lets just say they know which side their bread is well and truly buttered on.

You're still not getting it...The graph posted in the OP was taken off the site because it is wrong. All is explained in the conclusion.

Some people just don't have the capacity to read what is in front of them, which is why arguments happen.

humbug · 26 Jan 2017 at 18:17

Actually this reminds me, on the GTX 1060 lauch Guru3D refused to review Doom in Vulkan mode vs the RX 480, their excuse was they didn't have a key, this despite doing it in the Demo version just days before.

Raelgun · 26 Jan 2017 at 18:17

Lokken86 said:
Yes and test Doom only in OpenGL since that is the most taxing. You can't fault that kind of logic.

If you consider Open GL a setting up from Vulcan then go ahead to think that.

I consider Vulcan a setting on, using Open GL in that game is literally taking a setting off.

However in all fairness on this point, a lot of reviewers actually provide both results and rightly so.

Guru3D has omitted the shadow cache on benchmark here, if he wanted to be neutral or fair, he should provide both results side by side.