• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

[benchmarks] The Division - Steam Release

Page 4 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
Maxwell's 18 month subscription has expired. Sorry guys. Pascal will be here soon though. Then you can play this new game properly. Unless you have a 980ti. You have some time left until expiration.
 
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...

Ask me about replacing my 970 Gaming G1 as featured in that review with a Vapor-X 290 better than the one in that review and playing at 3440x1440. Best 30% boost, would "sidegrade" again.

Incidentally is this one of the games that desprado said AMD cards would underperform in? Because if so, I'm looking forward to them underperforming their NV counterparts into the ground even more.

Jokes aside, that's utterly shameful.
 
will have to wait for few more sites who post actual gameplay performance. gamestar and PCGH did actual gameplay runs. few sites like guru3d use in built benchmark. I would like to see results from few more sites which do actual gameplay benchmarks. anyway both Nvidia and AMD cards seem to be doing well. 980 Ti OC is the king of the hill as we all have come to expect.
 
Huh, PCGH shows much different results @ 1080p than the review in the OP...

It does. It's a very serious outlier in the PCGH review because it's high while the 970 otherwise starts low and scales well with resolution increases, which is why I think he discarded it.

However, it looks like Guru3D's review has the 970 holding on till after 1440 and only falling off a cliff at 4K, which makes me think there's something really strange going on.
 
It seems GTX970 gets problems if the game uses more than 3,5gb ram. Depending on the scene they benched this can happen maybe in lower or higher resolution.

What i find also interesting is how strong the 380X is. 11% faster than the 280x despite lower clocks and just 10% slower than 390. Seems this game profits a lot from Tongas good Front end.
 
It does. It's a very serious outlier in the PCGH review because it's high while the 970 otherwise starts low and scales well with resolution increases, which is why I think he discarded it.

However, it looks like Guru3D's review has the 970 holding on till after 1440 and only falling off a cliff at 4K, which makes me think there's something really strange going on.


It makes sense that it would incur a large hit at 4k, don't you think? Note the 380 4GB swaps positions with it at 4k too, ironic.
 
It seems GTX970 gets problems if the game uses more than 3,5gb ram. Depending on the scene they benched this can happen maybe in lower or higher resolution.

I don't think VRAM is an issue, since the 780Ti 3GB is on par with the 390X 8GB @ 1440p looking at PCGH's review... I didn't look into the 4K results much, as the framerates aren't even close to playable.

Edit - Oh, it looks like you guys are talking about Guru3D's review? I haven't looked at that yet...
 
Last edited:
The 380X seems to do relatively well in high resolutions for some reason.

In the pcgameshardware division benchmarks at 2560x1440 it almost ties a 290x. At techpowerup you see it beat a 290 on 4K resolution.
 
I don't think VRAM is an issue, since the 780Ti 3GB is on par with the 390X 8GB @ 1440p looking at PCGH's review... I didn't look into the 4K results much, as the framerates aren't even close to playable.

Edit - Oh, it looks like you guys are talking about Guru3D's review? I haven't looked at that yet...

I talked about pcgh, but you're right. The 3 Gb gpus don't struggle so much. But pcgh mentions a important feature. They write the game uses a lot of texture streaming and depending on the amount of ram, it's changing the level of detail in the distance. As i understood cards with lower amount of ram only get good textures nearby, even with the same texture details enabled. (Which could make the game useless for benches with different amounts of ram if you have different details)
So my theorie: With the 780Ti the game is pushing for 3gb textures, not more usind the LOD. But as GTX970 has 4gb, it's going for 4Gb and this is killing the performance because you can only use 3,5gb usefully.
 
Last edited:
Techpowerup shows the 380X beating the 290 at 4K as well:

perfrel_3840_2160.png


However, it doesn't do this well in other reviews, and its performance at lower resolutions is far less.
 
That texture quality thing would explain things although I'm surprised the minimum framerates aren't lower once it falls off the cliff.
 
980 Ti OC is impressive at 1080p and 1440p easily beating Fury X by 15-18%. Fury X matching 980 Ti OC at 4k is also commendable. It points to some huge bottlenecks in Fiji which are either due to hardware design or software (driver CPU overhead) or a combination of both.
 
Techpowerup shows the 380X beating the 290 at 4K as well:

However, it doesn't do this well in other reviews, and its performance at lower resolutions is far less.

That's a really weird result. You might think it could be a VRAM issue, where the games tested at settings where 4GB isn't enough for the 290 but the 380X gets by because of color compression.
https://www.techpowerup.com/reviews/ASUS/R9_380X_Strix/1.html
It could be the new mix of games, but in the 380X review the ref 290 is 27% faster @ 4k then the overclocked Strix model. It wins every single game benchmark.

Unfortunately they don't have the 290 or 380X in the individual scores of the new Matrix review, but if you look at the 390X in the common games between the two reviews, the scores don't change. It doesn't seem like different settings between the reviews.
 
That's a really weird result. You might think it could be a VRAM issue, where the games tested at settings where 4GB isn't enough for the 290 but the 380X gets by because of color compression.
https://www.techpowerup.com/reviews/ASUS/R9_380X_Strix/1.html
It could be the new mix of games, but in the 380X review the ref 290 is 27% faster @ 4k then the overclocked Strix model. It wins every single game benchmark.

Unfortunately they don't have the 290 or 380X in the individual scores of the new Matrix review, but if you look at the 390X in the common games between the two reviews, the scores don't change. It doesn't seem like different settings between the reviews.

Is it a reference 290?
 
Color compression happens between the VRAM pool and output; the full texture still needs to be in the VRAM with either case. It helps with slower memory, not memory capacity.

They may be doing driver magic for GCN1.2 cards that is not done on previous iterations to allocate textures more efficiently. Or something else in the architecture is better at 4K...
 
I've been running the inbuilt benchmark and I get results like avg cpu: 64 and avg gpu: 83. But it won't hold a constant 60fps. I wish it had peak numbers too.

Edit: just tested using gpu-z there are moments it hits 99% gpu. But only one or two seconds.
 
Last edited:
Back
Top