Resident Evil 7 Benchmarks

Kaapstad · 28 Jan 2017 at 16:36

Silent_Scone said:
Who knows. Maybe because the Fury-X is on an AIO, and isn't dumping hot air into the system.

It is possible but the difference does seem too high for that.

CAT-THE-FIFTH said:
Either the GTX1070 run was after the Fury card or it could be the Nvidia drivers pushing a greater CPU load as they thread better IIRC.

Or it could be because the cards are using different settings.

This needs to be explained.

My CPU does not show anywhere near that amount of different however many cards I am using.

Silent_Scone · 28 Jan 2017 at 16:42

The most likely explanation given the total load is very similar, is the overlay is reading the package temp. Package temp will read the average core temperature over a polling period.

Like Cat says, the system might have just had longer to warm up. Especially if he's using an AIO cooler, or air cooling.

CAT-THE-FIFTH · 28 Jan 2017 at 16:43

Kaapstad said:
It is possible but the difference does seem too high for that.

Or it could be because the cards are using different settings.

This needs to be explained.

My CPU does not show anywhere near that amount of different however many cards I am using.

AMD drivers tend to push less threads and Nvidia drivers do,so we would need to look at CPU usage graphs. If the Nvidia card is pushing more threads,it will be causing the CPU to heat up more.

But again maybe the GTX1070 was run after the Fury X so its simply the effect of the system warming up.

shankly1985 · 28 Jan 2017 at 17:04

So I guess from now on then we test games with features turned off that benefit one brand the most over the other ??
The only setting I can understand turning off it PhysX or a feature that is tied into the hardware from Nvidia Gameworks. HBAO+ runs a lot better on AMD hardware these days that is an Nvidia in house feature that am sure runs better on Nvidia than AMD so why not switch this off then for example? Leave SSAO.
The Shadow cache feature isn't AMD or Nvidia feature so it should be left alone!

Stupid benchmarking tbh Guru3D lost a lot of respect here

** Update #2: after lots of weird results with graphics card that have less then 4 GB of VRAM we found out that the problematic settings is Shadow cache. If you have an up-to 4 GB graphics card, please turn it off and you'll be playing properly. The charts are now updated to reflect that.

What a load off RUBBISH! 4GB card here and I see massive gains.

Silent_Scone · 28 Jan 2017 at 17:18

shankly1985 said:
** Update #2: after lots of weird results with graphics card that have less then 4 GB of VRAM we found out that the problematic settings is Shadow cache. If you have an up-to 4 GB graphics card, please turn it off and you'll be playing properly. The charts are now updated to reflect that.

What a load off RUBBISH! 4GB card here and I see massive gains.

At what resolution? Maybe you can show us some frametime data.

shankly1985 · 28 Jan 2017 at 17:24

Silent_Scone said:
At what resolution? Maybe you can show us some frametime data.

1440p

LoadsaMoney · 28 Jan 2017 at 17:49

humbug said:
Maybe something to do with HBM? Shankleys 290 is also 4GB.

Might be as im not on the Resi optimized drivers, still on the 16.3 hotfixes.

EDIT: Although can't be, as Hilbert showed the Fury stuttering like a mofo on the latest drivers.

Lokken86 · 28 Jan 2017 at 18:21

Should be a more optimised driver for AMD soon for this game.

fs123 · 28 Jan 2017 at 18:26

Lokken86 said:
Should be a more optimised driver for AMD soon for this game.

The game appears to be running the same or slightly faster than a 1070 on my 390 in the demo when I compare it to some youtube videos. Surely something wierd going on with the Nvidia drivers or is it simply more optimized for AMD?

LoadsaMoney · 28 Jan 2017 at 18:47

Yeah we know, the Nvidia owners are bloody livid!

TheRealDeal · 28 Jan 2017 at 19:18

fs123 said:
The game appears to be running the same or slightly faster than a 1070 on my 390 in the demo when I compare it to some youtube videos. Surely something wierd going on with the Nvidia drivers or is it simply more optimized for AMD?

Nope it's pretty much because AMD cards with 8 gb vram can use Shadow Cache, this super charges them making them punch way above there weight class.

fs123 · 28 Jan 2017 at 19:41

TheRealDeal said:
Nope it's pretty much because AMD cards with 8 gb vram can use Shadow Cache, this super charges them making them punch way above there weight class.

The shadow cache makes some difference mainly in the kitchen with the light shafts but all the other areas run at around 130fps even when it's turned off.

In the kitchen when looking from the doorway at the light shafts, disabling SC drops fps from 105fps to around 85fps. The game runs in the 130's the majority of the time so Guru3D seems to have taken near the minimum fps for his final 390X result.

Silent_Scone · 28 Jan 2017 at 20:03

fs123 said:
The shadow cache makes some difference mainly in the kitchen with the light shafts but all the other areas run at around 130fps even when it's turned off.

In the kitchen when looking from the doorway at the light shafts, disabling SC drops fps from 105fps to around 85fps. The game runs in the 130's the majority of the time so Guru3D seems to have taken near the minimum fps for his final 390X result.

Overclock3D ran the "high" settings also, btw.

*Rattling of pitch forks being grabbed*

...Not again.

LoadsaMoney · 28 Jan 2017 at 20:08

Just uncapped my frames, and tested with the Cache off/on, i do get higher frames with it on (120+ mainly), but when running along, i get quick pauses, every so often, and when looking/spinning around, i get a slight stuttering at first, but then once ive done a complete look/spin, the stuttering stops, goes buttery.

With the cache off, i get lower frames (112+ mainly), but no quick pauses when running, or stutters when looking/spinning around, its just buttery all the time.

Silent_Scone · 28 Jan 2017 at 20:12

LoadsaMoney said:
Just uncapped my frames, and tested with the Cache off/on, i do get higher frames with it on (120+ mainly), but when running along, i get quick pauses, every so often, and when looking/spinning around, i get a slight stuttering at first, but then once ive done a complete look/spin, the stuttering stops, goes buttery.

With the cache off, i get lower frames (112+ mainly), but no quick pauses when running, or stutters when looking/spinning around, its just buttery all the time.

Yep, paging to disk will do that, ruins the experience. Nobody wants to play like that...

siriq111 · 28 Jan 2017 at 23:42

akarypid said:
I thought this game is DX11?

If i am not mistaken, i did refer to new games as well.

Silent_Scone · 29 Jan 2017 at 13:57

Shadow Cache is extremely memory intensive. I'm seeing as much as 10GB requested by the GPU at 1080p on my TX. It's a luxury that's not required, and I think reviewers are wise to turn it off.

shankly1985 · 29 Jan 2017 at 14:22

Silent_Scone said:
Shadow Cache is extremely memory intensive. I'm seeing as much as 10GB requested by the GPU at 1080p on my TX. It's a luxury that's not required, and I think reviewers are wise to turn it off.

It will cache how much its allowed.. Because you have 16GB its allowed to cache that much doesn't mean on my system it will use 10GB. Its away of boosting performance and it should be left enabled!

queamin · 29 Jan 2017 at 14:50

They should do both charts on and off, or just on and off for cards with 6/8GB memory or more.

Silent_Scone · 29 Jan 2017 at 14:57

shankly1985 said:
It will cache how much its allowed.. Because you have 16GB its allowed to cache that much doesn't mean on my system it will use 10GB. Its away of boosting performance and it should be left enabled!

That's not how these things work, don't pretend you understand how stack overflow works based on what you've read from other users. The hitching experienced is enough to tell you there is an issue with cards with less than 4GB, which is where this argument started. Also Titan XP has 12GB.