[H]Rise of the Tomb Raider Video Card Performance Review

[raghu78]

In single GPU R9 390 does very well in price/perf. For multi GPU R9 390X CF provides awesome price/perf as it gets close to GTX 980 Ti SLI at 4k and 1440p (apples to apples comparison). good job Crystal Dynamics on getting CF/SLI working soon. :thumbsup:

 

[Bacon1]

SMAA has close to none impact on performance.

Ah well... thats probably because its not actually working.

https://www.reddit.com/r/pcgaming/comments/434w9l/just_a_heads_up_dont_run_smaa_on_rise_of_the_tomb/

Just tested myself as well, AA all over even with SMAA turned on, FXAA / SSAA 2x removes them.

SSAA had huge perf hit (obviously) and FXAA had no perf hit.

So looks like SMAA is just not working, as even the in game settings say "Anything over FXAA will incur a large performance hit".

 

[tential]

Could be. This isn't the first time I've seen glitches reported with Fiji Crossfire.

Although we do see in the patch notes improved SLI but nothing for Crossfire. AMD might require some game patching too for better results.



Notice I said what I would do. I'm not recommending it for others (not that I wouldn't). I just wouldn't want ~500W dumped into my case. 350W from Nano would be about my comfortable limit.

This doesn't discount any of your points. Hawaii is great value and of course if someone already owned one then the cost is dramatically less.



See above.

I do think Crossfire Nano is the way to go if you need/want the extra performance and have the cash though.

AMD really should talk about their great CF scaling. I think with Polaris, a LOT of people will consider crossfire builds where they wouldn't have before if the power consumption is as low as people think it is.

 

[moonbogg]

What I'm noticing is that in this game Fiji is matching GM200. I said from the start that Fiji was close enough that before it was over it would be the faster card. I still believe that.

That's because Nvidia drives hard in the beginning to grab a bunch of sales, then they slowly start to care less and less until the next cards come along. Then they make them look good for a few weeks and then start to care less and less as people buy them less and less until they eventually let them rot and get wrecked by AMD cards costing 1/4 the price.

 

[OVerLoRDI]

The performance on crossfire 290x is pretty excellent, but I've been getting some weird flickering that is only present in RoTR w/Xfire. There are some notes in the latest 16.1.1 drivers having flickering in certain areas under Windows 7. Hopefully next driver will resolve them.

 

[Silverforce11]

Let's talk a little bit about PureHair. In the last Tomb Raider game Crystal Dynamics leveraged AMD's TressFX, which did look great. It moved realistically, and made hair finally look "real" in a game.

In Rise of the Tomb Raider Crystal Dynamics went down its own path this time and created PureHair.

[H] being very revisionist. PureHair = TressFX3.0+, in fact made by AMD collaboration with the studio as noted by GDC presentations.

 

[3DVagabond]

[H] being very revisionist. PureHair = TressFX3.0+, in fact made by AMD collaboration with the studio as noted by GDC presentations.

That's what happens when you do something like OpenGPU. The positive of it is you don't have Hairworks destroying playability.

 

[Silverforce11]

The performance on crossfire 290x is pretty excellent, but I've been getting some weird flickering that is only present in RoTR w/Xfire.

They rushed out the 16.1.1 to get CF support into this game, it's got issues with Fury X not scaling well & stuttering. R290X/390/X CF is reportedly excellent on the Steam forums. Very strange. haha

You would expect them to rush out support for the top end configurations and get that to be good then focus on the rest of the stack, but they seem to go at Tonga/Hawaii optimizations first. Makes no sense.

 

[tg2708]

They rushed out the 16.1.1 to get CF support into this game, it's got issues with Fury X not scaling well & stuttering. R290X/390/X CF is reportedly excellent on the Steam forums. Very strange. haha

You would expect them to rush out support for the top end configurations and get that to be good then focus on the rest of the stack, but they seem to go at Tonga/Hawaii optimizations first. Makes no sense.

Why not? If a performance issue that is present can be fixed more quickly and or easily for a certain card why not do it first? To me its better to fix 50 bugs quickly than trying to do the same with 100 because another issue can always crop up thus the never ending cycle of optimizations and bug fixes.

 

[Silverforce11]

Because as a flagship product with a premium, if I were owners of such I would expect priority.

Btw, Fury X prior to 16.1.1 also stuttered in single GPU, as found by computebase.de and pcgameshardware.de, its a problem unique to this GPU for this game. The hotfix drivers fix the issue for single GPU but not for CF.

 

[raghu78]

Because as a flagship product with a premium, if I were owners of such I would expect priority.

Btw, Fury X prior to 16.1.1 also stuttered in single GPU, as found by computebase.de and pcgameshardware.de, its a problem unique to this GPU for this game. The hotfix drivers fix the issue for single GPU but not for CF.

I think 4GB VRAM is a issue in Rise of the Tomb Raider. The game can use up more VRAM and R9 390X and GTX 980 Ti SLI have the best performance at 4k and Fury X CF and 980 SLI really struggle with stuttering and inconsistent frames.

 

[Silverforce11]

I think 4GB VRAM is a issue in Rise of the Tomb Raider. The game can use up more VRAM and R9 390X and GTX 980 Ti SLI have the best performance at 4k and Fury X CF and 980 SLI really struggle with stuttering and inconsistent frames.

Nope, plenty of people play on R290/X with very high textures don't get stutter. Even review sites don't get stutter, except on Fury X.

You can see here: http://www.computerbase.de/2016-01/rise-of-the-tomb-raider-benchmarks/4/

On 16.1.1 AMD fixed the stutter for Fury X (single GPU), but they didn't fix it for CF.

From During Crimson 16.1 in some scenes on the Fiji graphics cards and has come to time to falter, the Crimson 16.1.1 runs noticeably better on the two Fury- and nano-graphics card and behaves there as the other AMD graphics cards.

You notice this in [H]'s own benchmarks.

1440p, Fury X has great performance in single GPU. Add a 2nd and boom, stutter and low min fps.

If it was a vram bottleneck, it would neuter single GPUs with 4GB vram as well.

SoM uses more than 6GB on Titan X. It runs fine on 4GB GPUs. Likewise for GTA V, COD AW, FC4 etc. These games have dynamic asset loading, if VRAM allows it, they can go very high.

Btw, if I had to guess, I would say Fiji requires specific optimization for it's HBM config in the drivers, rather than just standard GCN optimizations.

 

[3DVagabond]

I think 4GB VRAM is a issue in Rise of the Tomb Raider. The game can use up more VRAM and R9 390X and GTX 980 Ti SLI have the best performance at 4k and Fury X CF and 980 SLI really struggle with stuttering and inconsistent frames.

Fury runs 4K maxed out with higher mins than Titan X.

While @ 1440 the Fury Crossfire struggles with mins but the 4GB 980 doesn't.

I think it's simply Crossfire with this game. Probably needs driver work or game patch. The patch latest patch says it addressed SLI.

 

[crisium]

Good to see more parity in this title than from earlier and/or shorter benchmarks. I have to get critical though, because I'm still puzzled by some of HardOCP's conclusions here on recommendations. I see irrationality.

"For single-GPU our order of recommendation for this game is AMD Radeon R9 380X 4GB, AMD Radeon R9 390, AMD Radeon R9 390X, GeForce GTX 980, GeForce GTX 980 Ti. "

Wait, try again? The recommend the 980 over the 390X?

"Technically the AMD Radeon R9 390X is faster than the GeForce GTX 980, by about 7%, but that isn't enough to change settings."

The 980 costs more and is slower, and nails the recommendation. Say again?

"We have also found that AMD video cards below the R9 390X are doing quite well in Rise of the Tomb Raider compared to the NVIDIA counterparts. The only video card we were not impressed with is the Radeon R9 Fury. For the price, and the fact it is based on the latest GCN technology, it doesn't perform up to our expectations. The Radeon R9 Fury X is better, but both of these video cards are limiting at 4K with their 4GB of VRAM."

You recommend the 980 over the cheaper, and more VRAM heavy 390X? It costs more and is slower. But you are disappointed in the Fury, which costs more but at least is faster than the 980? And no comment on the 980 VRAM, just Fiji.

I know the Fury is a hard sell over the X and 980 Ti, but one wonders how in what universe the 980 comes away with a recommendation here.

 

[3DVagabond]

Good to see more parity in this title than from earlier and/or shorter benchmarks. I have to get critical though, because I'm still puzzled by some of HardOCP's conclusions here on recommendations. I see irrationality.

"For single-GPU our order of recommendation for this game is AMD Radeon R9 380X 4GB, AMD Radeon R9 390, AMD Radeon R9 390X, GeForce GTX 980, GeForce GTX 980 Ti. "

Wait, try again? The recommend the 980 over the 390X?

"Technically the AMD Radeon R9 390X is faster than the GeForce GTX 980, by about 7%, but that isn't enough to change settings."

The 980 costs more and is slower, and nails the recommendation. Say again?

"We have also found that AMD video cards below the R9 390X are doing quite well in Rise of the Tomb Raider compared to the NVIDIA counterparts. The only video card we were not impressed with is the Radeon R9 Fury. For the price, and the fact it is based on the latest GCN technology, it doesn't perform up to our expectations. The Radeon R9 Fury X is better, but both of these video cards are limiting at 4K with their 4GB of VRAM."

You recommend the 980 over the cheaper, and more VRAM heavy 390X? It costs more and is slower. But you are disappointed in the Fury, which costs more but at least is faster than the 980? And no comment on the 980 VRAM, just Fiji.

I know the Fury is a hard sell over the X and 980 Ti, but one wonders how in what universe the 980 comes away with a recommendation here.

[H] has been really PO'd at AMD for a while now. They were one of the sites not sent a Nano when AMD was calling sites out for their review practices. It's sad but it seems like their subjective analysis isn't lining up with their benchmarks.

 

[Bacon1]

"For single-GPU our order of recommendation for this game is AMD Radeon R9 380X 4GB, AMD Radeon R9 390, AMD Radeon R9 390X, GeForce GTX 980, GeForce GTX 980 Ti. "

Wait, try again? The recommend the 980 over the 390X?

I'm not sure, but maybe he was stating it the opposite way, as in "First -> Last"? Based on price/perf? If not, then yeah it makes zero sense.

And yes, they beat the "only 4gb" for Fury on every single review, even when they can't provide any proof of it being the cause for issues as SLI v CFX usually pulls in favor of Fury.

Also Nano is a great price/perf card as long as you allow it the air needed to not throttle itself.

Perf wise its about even with the Fury air, but it is the same price as the 980.

 

[tential]

[H] has been really PO'd at AMD for a while now. They were one of the sites not sent a Nano when AMD was calling sites out for their review practices. It's sad but it seems like their subjective analysis isn't lining up with their benchmarks.

They contradict themselves a lot in their analysis compared to actual results