[benchmarks] The Division - Steam Release

[Silverforce11]

Unless ~5 years ago (when we assume they laid out Pascal's design) they didn't really know about DX12/Vulcan/Mantle.

Possible, but we will still get many DX11 games, such as The Division.

I see some people said that NV is starting to gimp Maxwell, but I thought at the time, no way, cos Pascal is still far away, there's no reason for them to do that. So if its happening, Pascal could be here sooner than we think. *queues mysterious music*

 

[flynnsk]

NV Driver 364.47 , AMD Driver 16.2.1
http://www.gamestar.de/hardware/praxis/technik-check/3269013/the_division_pc_technik.html

Spoiler

http://i.imgur.com/inPUHmA.jpg

http://techfrag.com/2016/03/07/amd-...ision-radeon-gpus-perform-exceptionally-well/

Spoiler

Wrapping things up, AMD’s Radeon GPUs perform exceptionally well in The Division, with Radeon R9 390 outperforming the GeForce GTX 970 at all resolutions

 

[tential]

I am afraid to use 364.47 from the problems I've been reading about.

Do NOT use it. I already have texts from friends who had to rollback their PCs.

 

[escrow4]

Pffft, poor single player (and you can barely call it that) and I suspect Kepler has been gassed once again. I really should just sell this gaming box, these 30-50GB console ports are laughable.

 

[Erenhardt]

Not testing legendary 280X feels like a major insult.

 

[sxr7171]

Runs great. Just disable PCSS or HFTS or whatever and it will run great.

 

[Mahigan]

Do NOT use it. I already have texts from friends who had to rollback their PCs.

Pffft

It's a new NVIDIA Gameworks title called "Rollback". Super fun. The PhysX effects are life-like. It's almost as though you're really there, rolling back your driver, a true GeForce Experience. A+ NVIDIA. AMD sucks, stupid peasant cards

Threadcrapping and trolling are not allowed
Markfw900

 

[thesmokingman]

I think that driver is 6 months early. Oh snap?

 

[airfathaaaaa]

Do NOT use it. I already have texts from friends who had to rollback their PCs.

most of the problems was because people stacked the new driver with the old one without using ddu to delete them

 

[Pantalaimon]

most of the problems was because people stacked the new driver with the old one without using ddu to delete them

Isn't that how the GeForce Experience updates the drivers? You're saying NVDIA is not doing it right?

 

[jantjeuh]

I don't think Maxwell performing worse than GCN in some recent titles is due to Maxwell being intentionally gimped by nVidia (though the Kepler dropoff might be due to that). I think it's simply a case of old game engines being phased out, and newly written/updated engines being used in recent titles. These engines might perform better on GCN due to them being targeted for a subset of PS4/XB1/PC, where GCN simply is a major common factor.

 

[airfathaaaaa]

Isn't that how the GeForce Experience updates the drivers? You're saying NVDIA is not doing it right?

the nvidia reddit page is full of people saying that they had to delete the previous one and then install the new
some even say they had to go into safe mode and delete their monitor drivers too and then install one by one and doing a restart in the process

 

[3DVagabond]

most of the problems was because people stacked the new driver with the old one without using ddu to delete them

That's probably half of all driver problems. People are too lazy to do a proper driver update. You can often get away with it, but not always. Then things go to custard and they start whining on the internet.

 

[Silverforce11]

That's probably half of all driver problems. People are too lazy to do a proper driver update. You can often get away with it, but not always. Then things go to custard and they start whining on the internet.

I've been just updating new AMD drivers directly over the older versions, did same for Crimson.

I thought we have long moved beyond the days where you needed a 3rd party driver sweeper routine just to update drivers!

 

[airfathaaaaa]

I've been just updating new AMD drivers directly over the older versions, did same for Crimson.

I thought we have long moved beyond the days where you needed a 3rd party driver sweeper routine just to update drivers!

windows doesnt always delete properly (especially from the registry)the older files its always the best option to actually delete EVERYTHING before installing the new driver

 

[3DVagabond]

I've been just updating new AMD drivers directly over the older versions, did same for Crimson.

I thought we have long moved beyond the days where you needed a 3rd party driver sweeper routine just to update drivers!

Like I said, you can often get away with it. You take the percentage of times it will FUBAR though and multiply it by the internet and you are going to have a crapload of people screwing up their systems.

windows doesnt always delete properly (especially from the registry)the older files its always the best option to actually delete EVERYTHING before installing the new driver

^^^This^^^ Especially if you are switching brands.

 

[PhonakV30]

The first major Directx 12 release ran like crap on gcn. I think directx 12 will be worse for AMD than directx 11 was.

wow Major DX12? Really?

Seriously 2 core?

This is Major Dx12 :

 

[csbin]

http://www.pcgameshardware.de/The-Division-Spiel-37399/Specials/PC-Benchmarks-1188384/


PCGH Benchmarks


Geforce 364.47

Radeon Software 16.2.1



1080P

2560X1440

3440X1440

3840X2160

 

[beginner99]

It's all good, my R290X (~$250 USD) stomping on the 970 and matching highly OC 980 in these new games in 2016! \o/

Really, later this year, it's likely to perform great in Mirror's Edge and Deus Ex too. R290X got some legs...

Yeah, it was obvious 290(x) will be best performance/$ for a long time. New ones here never got very cheap but I got a used one for $250 as well it was a relatively new used one with warranty, bought after bitcoin mining on GPU made any sense). Now I'm actually CPU bound but that can also change with dx12.

wow, 380x beating 970. hmm, I guess maxwell tanking performance is no longer my own prediction.

Planned obsolesce. I predicted it in another thread that Maxwell will start to look bad after Pascal release but it looks like NV already changed GimpWorks hurt Maxwell to mentally condition Maxwell users to upgrade to Pascal. From a business perspective it's ingenious. I don't get however how consumers fall for this over and over again in the last 5 years.

 

[Silverforce11]

http://www.pcgameshardware.de/The-Division-Spiel-37399/Specials/PC-Benchmarks-1188384/

They did a manual 20s fraps run. There's no way it's believable that the 970 goes from that much down to such low levels with a move from 1080p to 1440p. Compare the other cards.

They must have stared at the floor for 20 seconds of their run at 1080p. -_-

However, the 390, 980 and Fury results are inline with the other site, when comparing the models tested.

 

[linkgoron]

http://www.pcgameshardware.de/The-Division-Spiel-37399/Specials/PC-Benchmarks-1188384/


PCGH Benchmarks


Geforce 364.47

Radeon Software 16.2.1



1080P

Those benchmarks are really impressive and reliable...

A fury X getting 2 FPS, the 980ti getting 2 FPS, as do the 390x, sapphire 290x etc. I can't believe they posted that.

 

[Silverforce11]

Those benchmarks are really impressive and reliable...

A fury X getting 2, the 980ti getting 2, as do the 390x, sapphire 290 etc. I can't believe they posted that.

It's live since they haven't added the other results yet.

They for some reason, force HQ AF in drivers when they test, I wonder if there's a difference since sites don't normally talk about it and leave it default. I noticed AMD's own testing, they always go 0x AF on both drivers... and they get better results than review sites. There must be something going on there, because I've seen pcgameshardware.de consistently get AMD performance about 10% or so worse than other sites like PCPER, TPU, Guru3d, Computerbase etc.

I wonder for those who have both AMD/NV GPU, if they can test to see if there's a performance impact of leaving it default vs forcing HQ AF in drivers? @gus?

Again, their 970 results for 1080p vs 1440p is just unbelievable and if they claim that's reality, it means their timed 20s manual play-run is borked for inconsistency.

Oh, they also didn't mention HBAO+ and PCSS+, just Ultra, so those are probably on and everything tanks in performance compared to the other review site which have those GW off.

 

[linkgoron]

nvm

 

[Dygaza]

Those benchmarks are really impressive and reliable...

A fury X getting 2 FPS, the 980ti getting 2 FPS, as do the 390x, sapphire 290x etc. I can't believe they posted that.

Those 2pfs numbers are placeholders. 980 Ti result have now appeared, so it seems they are updating it realtime when they are benching.

 

[linkgoron]

Those 2pfs numbers are placeholders. 980 Ti result have now appeared, so it seems they are updating it realtime when they are benching.

Yes, I missed the part where they wrote (via google translate):
"The following benchmark is still some placeholder values. All graphics cards that are specified with 2.0 / 1 FPS, are on our test agenda, the procedure has not yet been completed. Their values ​​follow the course of the day."

I must admit that I find it confusing when someone uploads an article before they actually have the final results, but being first is important I guess.