[HardOCP] Nvidia cards much faster than AMD cards in "Rage" game.

[biostud]

I really don't understand the obsession with the performance in a single game. nVidia and AMD cards are somewhat equally fast, some games are faster on AND hardware and some on nvidia. And in Rage both sides can play the game fine.

 

[RobertR1]

I've noticed that AMD cards tend to have more severe and frequent dips in games than nvidia cards (atleast at the 580 vs 6970 level which is all I pay attention to). Any technical explanation for that?

What I mean by this is that nvidia tends to have less peaks and valleys and relatively consistent performance where as amd cards tend to be spike up and down. Both end up around the same avg just go by different means of getting there.

 

[blackened23]

I've noticed that AMD cards tend to have more severe and frequent dips in games than nvidia cards (atleast at the 580 vs 6970 level which is all I pay attention to). Any technical explanation for that?

What I mean by this is that nvidia tends to have less peaks and valleys and relatively consistent performance where as amd cards tend to be spike up and down. Both end up around the same avg just go by different means of getting there.

What? Rage runs well even on dated hardware so i'm not sure what you're talking about (unless you're jumping on the hate of all things AMD bandwagon) If your'e gaming at 2560x1440, MAYBE if you enable driver post processed effects (which is stupid, especially in rage.). Lets make one thing clear, rage is pretty grahpically tepid even on dated hardware so I have no idea what you're talking about. A 6970 destroys rage just like a 6950, 560ti, 570, or so on. I game at 1080p and rage is pretty much a joke compared to crysis 2 or metro 2033.

 

[blackened23]

I've noticed that AMD cards tend to have more severe and frequent dips in games than nvidia cards (atleast at the 580 vs 6970 level which is all I pay attention to). Any technical explanation for that?

What I mean by this is that nvidia tends to have less peaks and valleys and relatively consistent performance where as amd cards tend to be spike up and down. Both end up around the same avg just go by different means of getting there.

What? Rage runs well even on dated hardware so i'm not sure what you're talking about. If your'e gaming at 2560x1440, MAYBE if you enable driver post processed effects (which is stupid, especially in rage.). Lets make one thing clear, rage is pretty grahpically tepid even on dated hardware so I have no idea what you're talking about. A 6970 destroys rage just like a 6950, 560ti, 570, or so on. I game at 1080p and rage is pretty much a joke compared to crysis 2 or metro 2033.

My 580 box doesn't even approach anything close to 100% gpu usage, I guess rage is a pretty good CPU benchmarking program though?

 

[blackened23]

I've noticed that AMD cards tend to have more severe and frequent dips in games than nvidia cards (atleast at the 580 vs 6970 level which is all I pay attention to). Any technical explanation for that?

What I mean by this is that nvidia tends to have less peaks and valleys and relatively consistent performance where as amd cards tend to be spike up and down. Both end up around the same avg just go by different means of getting there.

Rage runs well even on dated hardware -- If your'e gaming at 2560x1440, MAYBE some cards would dip with PP IQ settings enabled (which is stupid, especially in rage.). Rage is pretty grahpically tepid even on dated hardware -- A 6970 destroys rage just like a 6950, 560ti, 570, or so on. I game at 1080p and rage is pretty much a joke compared to crysis 2 or metro 2033.

At least you can bump up the detail and turn on tessellation and physx in rage though right? Oh wait, nevermind ()

My 580 box doesn't even approach anything close to 100% gpu usage, I guess rage is a pretty good CPU benchmarking program and worthless for everything else.

 

[RobertR1]

It's like you didn't read my post and yet replied to it 3 times anyway.

 

[formulav8]

OK blackened we got the point, you don't need to post it a fourth time.

 

[(sic)Klown12]

I've noticed that AMD cards tend to have more severe and frequent dips in games than nvidia cards (atleast at the 580 vs 6970 level which is all I pay attention to). Any technical explanation for that?

What I mean by this is that nvidia tends to have less peaks and valleys and relatively consistent performance where as amd cards tend to be spike up and down. Both end up around the same avg just go by different means of getting there.

I don't have any evidence backing this up, but I feel it might have something to do with Nvidia's superior primitive rate compared to AMD. Fermi can do 4 per cycle while Cayman is limited to a maximum of 2 with an actual rate of around 1.8 in games.

 

[Silverforce11]

I've noticed that AMD cards tend to have more severe and frequent dips in games than nvidia cards (atleast at the 580 vs 6970 level which is all I pay attention to). Any technical explanation for that?

Got any bench to back that up?

And an easy explanation for you: One card is a lot cheaper than the other, they don't even belong in the same performance price class.

 

[PingviN]

That Cuda Gpu texturing must really help..

It looks like AMD's minimums are really bad.

"Apples to Apples. No GPU transcoding"

 

[Ackmed]

id has always been in nVidias pocket, this is not news. Goes back to Doom 3 and beyond.

Perhaps id should focus on producing a more bug free game, without graphical complications instead of making the mess they did at launch. Oh wait, the PC was a port of a console game, I forgot.

And for the record, I am actually enjoying the game with the preview 3 AMD drivers. No popup, or tearing. Smooth gameplay with my 2500k and 6850 too, 1920x1200. The first week was brutal though.

 

[96Firebird]

So because AMD drivers fixed the problem, the blame is put on iD being in Nvidia's pocket...

Interesting.

 

[happy medium]

Got any bench to back that up?

And an easy explanation for you: One card is a lot cheaper than the other, they don't even belong in the same performance price class.

318$AR shipped for a 6970 and 407$ AR shipped for a gtx580 on Newegg. Thats 89$ more. Not bad for the fastest single gpu card.
I think Russian said the gtx580 is now about 20% faster than a 6970 @ 1080p.

But you are right the gtx570 is cheaper and on avg, just as fast @ 1080p ,thats a better comparison with a 6970

 

[tviceman]

I really don't care about anything hardocp does in reviews anymore. It's almost like they are extreme trolling one way or another. The majority of the time, their benchmarks do not jive with every other hardware website, and with what people say about hardocp's chief editor acting like Hitler on their forums, it's become extremely easy for me to avoid that website.

At least semiaccurate.com has entertainment value in how bad Charlie likes to exaggerate and hate on Nvidia.

 

[WMD]

This is not a gpu benchmark but rather cpu vs gpu transcoding benchmark. Running at high AA and resolution places more stress on the virtual texturing cause performance dips on AMD cards. They should have benchmarked with and without gpu transcode. Btw rage runs with no stutter or texture pop on my system. It uses only 35% of the gpu while locked at 60fps. So its actually pretty worthless as a gpu benchmark.

 

[(sic)Klown12]

Two of those graphs are without GPU transcoding(look at post 60), and there is large difference with the minimum framerate. It could just be drivers still, or Fermi's architecture is just a lot better with this game engine in this regard.

 

[3DVagabond]

I really don't care about anything hardocp does in reviews anymore. It's almost like they are extreme trolling one way or another. The majority of the time, their benchmarks do not jive with every other hardware website, and with what people say about hardocp's chief editor acting like Hitler on their forums, it's become extremely easy for me to avoid that website.

[H] reviews differently than other sites. that's why their results are different and why their reviews have value. Not because they are better or worse than others, but different. If all of the sites had the same results, as many of them do, what's the point of multiple sites?

As far as the Admins at [H], better to judge for yourself. I've only seen trolls get the kind of treatment you are referring to. Where as many sites allow it because the bickering generates lots and lots (and lots) of page hits.

At least semiaccurate.com has entertainment value in how bad Charlie likes to exaggerate and hate on Nvidia.

The best way to keep track of someone/something is by listening to their distracters. You have to sift through the BS, but you can get some valuable info. Bulldozer and Fermi are prime examples of who was giving the best info. It wasn't the respective company's fanbois. There's also some very informed posters @ S/A.

 

[Pantalaimon]

I'm talking specifically the numbers for the PC. With all the hype building up prior to release, 60K is pretty small. With all the bad press that has followed, I doubt sales numbers will pick up any steam outside of some huge price cut.

60K units sold on PC platform is pretty bad considering it's iD. I wonder if the low number has something to do with AMD's screw up with the incorrect drivers. If it is, that would make another reason for iD to hate AMD even more

ETA: Compared to Deus Ex: HR, according to VGA Chartz says Deus Ex has sold over 700K on XBOX360, over 500K on PS3, and about 180K on PC so far.

 

[3DVagabond]

60K units sold on PC platform is pretty bad considering it's iD. I wonder if the low number has something to do with AMD's screw up with the incorrect drivers. If it is, that would make another reason for iD to hate AMD even more

ETA: Compared to Deus Ex: HR, according to VGA Chartz says Deus Ex has sold over 700K on XBOX360, over 500K on PS3, and about 180K on PC so far.

AMD's fault, eh? I hear they're causing all of the earthquakes lately as well.

 

[Keysplayr]

AMD's fault, eh? I hear they're causing all of the earthquakes lately as well.

No. They're not.

 

[Seero]

60K units sold on PC platform is pretty bad considering it's iD. I wonder if the low number has something to do with AMD's screw up with the incorrect drivers. If it is, that would make another reason for iD to hate AMD even more

There is nothing surprising as it is an OpenGL game. Some may not remember but OGL games were always buggy with poor driver support.

It was expected to have some degree of problems but not this magnitute. Nvidia's driver is actually problematic as its OGL driver also lacks the appropriate optimization and features too, but AMD never put in the right version of the OGL pll in all its drivers, meaning it was never tested properly.

Nvidia have just released its WHQL driver which contains the appropriate OGL extensions, and if everything works as intended, than rage will be able to operate as intended on PC with Nvidia video card. AMD's 11.10 preview 3 still has problems and there is a OGL package that can be install on top of the preview 3 driver. Again, if everything works as intended on the final driver release, then rage will able to operate as intended on PC with AMD video card.

Rage's (PC) version should have waited out, but then it will take batman AC and BF3 heads on. Good decision? I don't know.

 

[Dribble]

CUDA GPU transcoding is GPU assisted transcoding of the heavily compressed megatextures used in Rage basically.

Anandtech had a bit of a write up about it:



Source..

ID also tried to use OpenCL to do the same thing on the Radeons, but the performance apparently wasn't good enough to justify including it in the final release.

This is interesting. Shows that AMD still have a long way to go with opencl - it's one thing to do a proof of concept where you can run opencl on your gpu, quite another to get it to work efficiently in anger.

 

[Silverforce11]

So random. Patched RAGE. "A few days after the release of the patch was released, extending the game graphics settings"

And no CPU bottleneck whatsoever, which is why this test is meaningful the way GameGPU always mix and match platforms. No CPU factor in fps performance.

 

[amenx]

^ Ah... vsync. The great equalizer of all hardware.

 

[Silverforce11]

^ Ah... vsync. The great equalizer of all hardware.

RAGE has 60 fps internal cap ^^