[HardOCP] Tomb Raider Performance Review w/ new drivers

[Stuka87]

Great job by Nvidia getting the driver fix out quickly! Performance is much improved. It is also really good to see that it would seem AMD did nothing to cripple a competitor's GPU.

Yeah, AMD could have easily pulled a "Crysis 2" on nVidia with this game. My guess is nVidia had already been working on drivers before the game was released, which is how they got them out so early.

 

[BallaTheFeared]

Yeah, AMD could have easily pulled a "Crysis 2" on nVidia with this game. My guess is nVidia had already been working on drivers before the game was released, which is how they got them out so early.

How? Nvidia has higher raw tess performance :hmm:

DirectCompute? They tried that, TressFX.


Your guess is wrong, Nvidia was already working on drivers prior to release, for another build of the game. What Nvidia was working on and what was released changed.

 

[Stuka87]

Your guess is wrong, Nvidia was already working on drivers prior to release, for another build of the game. What Nvidia was working on and what was released changed.

Are you saying TressFX was completely changed between beta and release? Because there is no way the developer would put some huge change in after already going to beta. Its also very unlikely that nVidia turned a driver release around in 10 days, 4 of which were weekends.

 

[BallaTheFeared]

Did I say TressFX?

I didn't think I did, if I did I'm sorry, I didn't mean to say they changed TressFX only that they changed the build of the game enough that it messed up Nvidia's drivers - According to the horses mouth.

 

[parvadomus]

How? Nvidia has higher raw tess performance :hmm:

AFAIK NV uses shaders to do tessellation, meanwhile AMD uses separate hardware. I dont know what would happen if shaders are also taxed with something else, and not a pure tess benchmark.

 

[BallaTheFeared]

Since when? GF100/110 had a tess engine for each SM cluster, 5xxx had 1 tess engine and 6xxx had 4, never heard of either of them using shaders for tess.

 

[notty22]

Yeah, AMD could have easily pulled a "Crysis 2" on nVidia with this game. My guess is nVidia had already been working on drivers before the game was released, which is how they got them out so early.

So much wrong with your guesses.

Crysis2 dx11 was a optional upgrade PATCH. Including larger textures. Which also was not very playable for 1gb cards. Besides weaker cards on tessellation. A in-game changeable option.
edit: my point also, is that the dx11 patch was not a game launch day option/issue.
TR2

Game needed a engine patch, which is also publicly known. It's not helpful spreading conspiracy theories.

Lara Croft Joins GeForce with 45% Performance Gain


The launch of Tomb Raider last week delivered an exciting reboot to a legendary franchise.
GeForce gamers have come to expect an awesome Day 1 experience. But in the case of Tomb Raider, that didn’t come to pass. We hadn’t received the game in time to assure them of this – which was disappointing for them and for us.
But our software engineers, like our GPUs, are the best in the industry. And they’ve been working closely with Crystal Dynamics since the game’s release to help improve the experience for GeForce gamers.
So, in conjunction with a game patch that Crystal Dynamics released earlier today, we recommend that GeForce users update to the just-posted GeForce 314.21 drivers. They’ll provide major gains in reliability and performance in the game – delivering an average of 45% and up to 60% performance improvements (as measured with GeForce GTX 680 at 1920×1080 and 2560×1600 resolutions

 

[ad3pt7]

I'd like to get this game pretty soon. Looks like I should get solid performance running SLI 670's at 1920x1080. Hopefully lol, I even get pretty crazy drops in Far Cry 3 and Crysis 3 sometimes.

 

[Tsavo]

TressFX looks pretty neat on the hair. Is that all it does? (I've gathered that is about all?) Edit. Now I see the 40+ page thread on this that I've been ignoring. XD

It reminds me of physx, just a tack on thing. Physx is gimmicky and just adding a few curtains waving in the breeze and scraps of rock that bounce and then lay above the ground before fading away isn't doing much to improve IQ.

Screenshots don't do TressFX justice, and it's far from tacked on. Hair moves fairly well, but most importantly, catches light in a similar way to real hair.

PhsyX is great in Batman, BL2 and that FTP online mech game.

 

[BallaTheFeared]

Screenshots don't do TressFX justice, and it's far from tacked on. Hair moves fairly well, but most importantly, catches light in a similar way to real hair.

PhsyX is great in Batman, BL2 and that FTP online mech game.

Mine jumps around and makes little to no sense for the most part, that's just the tail the bangs do nothing they just sit there.

Clips weapons, and is generally giving off goofy looking light reactions, like I was blonde when sliding down a tunnel at one point as an example.

 

[Bateluer]

Anyone else really think that HardOCP article was just really poorly done? Same cards were all tested at different resolutions, there wasn't a clear chart with all cards tested at the different resolutions and settings.

 

[Dankk]

Mine jumps around and makes little to no sense for the most part, that's just the tail the bangs do nothing they just sit there.

Clips weapons, and is generally giving off goofy looking light reactions, like I was blonde when sliding down a tunnel at one point as an example.

Odd. You're the only person I've heard mention problems like that. Are you using the latest drivers and everything?

 

[chimaxi83]

Anyone else really think that HardOCP article was just really poorly done? Same cards were all tested at different resolutions, there wasn't a clear chart with all cards tested at the different resolutions and settings.

That's the way they test. First they do "highest playable settings", where they find the max settings that provide a subjective playable experience to them. Typically just resolutions and AA levels change for that test. They then do apples to apples, with all settings matched like in the typical review out there.

 

[Tsavo]

Anyone else really think that HardOCP article was just really poorly done? Same cards were all tested at different resolutions, there wasn't a clear chart with all cards tested at the different resolutions and settings.

Bidness as usual.

About as pointless as TR's scatter plots.

 

[DiogoDX]

Anyone else really think that HardOCP article was just really poorly done? Same cards were all tested at different resolutions, there wasn't a clear chart with all cards tested at the different resolutions and settings.

In HardOCP tests you always have to look at apples to apples because that is the test with the same settings.

In all the forums I read have reports that the benchmark is heavier than the gameplay and surprisingly HardOCP play the whole game and choose a part of the gameplay that 7970GHZ presents worse performance than the benchmark.:whiste:

To test Tomb Raider we first played through the entire game. We looked for scenes, levels, or areas which produced lower framerates than others. Our run-through takes place at the location the "Chasm Ziggurat."

 

[BallaTheFeared]

In all the forums I read have reports that the benchmark is heavier than the gameplay and surprisingly HardOCP play the whole game and choose a part of the gameplay that 7970GHZ presents worse performance than the benchmark.:whiste:

And then you go and post graphs from pre game update & pre nvidia driver update :whiste:

 

[tviceman]

In all the forums I read have reports that the benchmark is heavier than the gameplay and surprisingly HardOCP play the whole game and choose a part of the gameplay that 7970GHZ presents worse performance than the benchmark.:whiste:

Wut? If they are benching one of the most demanding areas of the game, it's going to perform worse on Nvidia cards too. I don't think there will be areas of the game that will benchmark noticeably better on one particular vendor for an extended period of time, and then other areas where it will behave the opposite and perform noticeably better on the OTHER vendor for an equally long period of time. Just doesn't work like that. If a game favors one particular type of GPU, it will do it consistently throughout the entire game. And this is an AMD gaming evolved game, soooooo chances are it's got a bit more optimizations specific towards GCN than Kepler.

And, as Balla said, your graphs are all before Nvidia's driver fixes. So your entire post is both wrong and invalid.

 

[DiogoDX]

And then you go and post graphs from pre game update & pre nvidia driver update :whiste:

How Nvidia drivers will affect 7970GHZ performance?

 

[BallaTheFeared]

Odd. You're the only person I've heard mention problems like that. Are you using the latest drivers and everything?

White

 

[3DVagabond]

for AMD its hugely embarassing. Nvidia cards are running better on GE titles. they better have some performance improvements in their upcoming drivers.

If AMD wanted GE games to run better on AMD hardware it would be easy to do. It's obviously not their aim.

 

[Jaydip]

If AMD wanted GE games to run better on AMD hardware it would be easy to do. It's obviously not their aim.

You can't be serious.It will be a huge embarrassment for them if NV trumped them in GE titles and vice versa. Dirt-showdon,Sniper Elite V2 comes to mind.

 

[nextJin]

If these numbers stick (after they both finalize their driver tweaks) I'll actually be pissed.

 

[antihelten]

HardOCP has some benchmark comparisons with the old patch at the end of the article. This allows us to evaluate their chosen benchmark spot (chasm ziggurat), with that of the default benchmark (using the numbers provided by DiogoDX previously).

At 1920x1080 with 2xSSAA, HardOCP gets 35.5 FPS with the 7970GHz, compared to 40 and 40 for the sites running the default benchmark, or a drop of about 11%. This isn't particularly surprising as HardOCP specifically looked for the most demanding part using a 7970GHz.

At the same settings, HardOCP also gets 35.5 fps with the GTX 680 using the old patch, compared to 31 and 33 for the sites running the default benchmark, or an increase in performance of about 11%. This means that in contrast to the 7970GHz, this spot is actually easier for the GTX 680 with the old patch, than the default benchmark.

In conclusion, using the spot HardOCP uses, instead of the default benchmark, provides the GTX 680 (with old patch) a total boost, relative to the 7970GHz, of 25%, which coincidently is exactly the same as the difference HardOCP shows between the 7970GHz and the GTX 680 with the new patch (35.5 FPS vs. 44.4 FPS).

So clearly there is a rather significant difference between the default benchmark and HardOCP's spot, that completely changes the rankings, but the question then is which one is more representative of actual gameplay.

 

[BrentJ]

HardOCP has some benchmark comparisons with the old patch at the end of the article. This allows us to evaluate their chosen benchmark spot (chasm ziggurat), with that of the default benchmark (using the numbers provided by DiogoDX previously).

At 1920x1080 with 2xSSAA, HardOCP gets 35.5 FPS with the 7970GHz, compared to 40 and 40 for the sites running the default benchmark, or a drop of about 11%. This isn't particularly surprising as HardOCP specifically looked for the most demanding part using a 7970GHz.

At the same settings, HardOCP also gets 35.5 fps with the GTX 680 using the old patch, compared to 31 and 33 for the sites running the default benchmark, or an increase in performance of about 11%. This means that in contrast to the 7970GHz, this spot is actually easier for the GTX 680 with the old patch, than the default benchmark.

In conclusion, using the spot HardOCP uses, instead of the default benchmark, provides the GTX 680 (with old patch) a total boost, relative to the 7970GHz, of 25%, which coincidently is exactly the same as the difference HardOCP shows between the 7970GHz and the GTX 680 with the new patch (35.5 FPS vs. 44.4 FPS).

So clearly there is a rather significant difference between the default benchmark and HardOCP's spot, that completely changes the rankings, but the question then is which one is more representative of actual gameplay.

I would say actual gameplay is representative of actual gameplay.

 

[BrentJ]

Anyone else really think that HardOCP article was just really poorly done? Same cards were all tested at different resolutions, there wasn't a clear chart with all cards tested at the different resolutions and settings.

We start with highest playable settings, which takes the cards, and we find out what in-game quality settings/resolution are playable on each, since that is what gamers do with video cards when they play games, first figure out what settings are playable. Then, we do apples to apples tests at same settings, for those that find value in that. Both methods are presented in every evaluation.