HD5870 vs GTX480 two years later.

[MrK6]

Is the problem that people don't understand graphics card architectures? The 5870 was a graphics card architecture from 2008 with a tessellation engine and a few other tweaks slapped on. You'll notice in games with heavy tessellation or an NVIDIA-only optimization (Civ5, etc.), the 5870 is behind. Otherwise, the GTX 480 is only 10-15% ahead, as it was at launch. Drivers have done little to differentiate the two cards, only the architectures have.

 

[Grooveriding]

In BF3, GTX560ti 1GB is faster than HD6950 2GB and GTX570 1250MB is faster than HD6970 2GB. NO, BF3 doesnt need more than 1GB buffer at 1080p Ultra, otherwise GTX560 would not be faster than HD6950 2GB.

About Civ 5

Everyone talks about the Multithreading drivers. I havent seen anyone quoting Ryan's latest evaluation about Civ 5 performance in HD7970 review. Let me quote it here,

http://www.anandtech.com/show/5261/amd-radeon-hd-7970-review




About Batman: AC

Clearly the latest Unreal Engine 3 with DX-11 needs massive amounts of compute power, something Evergreen(HD5870) lacks.
Take for instance the HD6970 vs HD7970. GCN was designed for GPGPU, that gives it a huge boost in games with DX-11. In Batman: AC HD7970 is 64% faster than HD6970.

Fermi architecture was designed for GPGPU and DX-11, it has much more compute power than HD5870, that is the reason of the huge performance difference in Batman: AC.

Do you start to see a pattern here ?? in both of this games (Civ 5 and Batman: AC), HD7970 is 60%+ faster than HD6970.


About Shogun 2

By what you are saying, i shouldn't have put Shogun 2 in the mix. HD5870 has the same performance with GTX580, really ??
Why you didnt say anything about that game ?? Well it seams that the developers made a nice job coding the game for VLIW architectures and you know what ?? that's fine by me, because there are a lot of people with AMD cards playing that game.

Just because one game perform way faster than the other card, we dont discard it from benchmarks. It really shows us the differences in Graphics Card Architectures. If you dont like to see the differences or you dont play that game/games dont count them. But when we evaluate the performance of a card we need to take in to consideration all current DX-11 games.

Most of the DX-11 games will need more compute power in the near future. GNC is the right direction but im afraid that Tahiti is not enough to compete with GK100. Because of that, GK104, a middle end NV card could perform very close with 79xx series in those DX-11 games.

2012 is a DX-11 year, and you wont see the same small performance differences we had in 2009-2010 with DX-9 games in the mix. If GK100 will double its compute power it will have significant more performance in those DX-11 games than Tahiti. AMD will have to design a bigger chip some time in the near future in order to compete in DX-11 games and in GPGPU.

Are you lumping compute in with tessellation ? Because in the games you've used with the largest differences; tessellation looks to be the deciding factor in the differences. You expect this to play out the same way with Tahiti vs Kepler ? I suppose that's possible if nvidia can reduce the performance impact tessellation inflicts on their cards more so than it does currently, but that sounds like a stab in the dark. You might want to re-read the results of multi-threaded rendering for nvidia. Performance doubled, that's a significant gain.

You may not want the one benchmark that helps to prop up your whole opinion more than the rest to be discarded, but when there is a result so supect and defying logic, it needs to be looked at closer. Have you played B:AC ? It looks like a console port and the graphics are not groundbreaking, contrast it to BF3, Metro 2033 or Crysis 2, games that show a reasonable performance disparity and visually are exponentially more impressive, not the ridiculous 100%. The 100% jump seen there begs suspicion in a game that is one of the least impressive graphically in that entire list. It's also a suspect result that most skews your entire summation.

Per Shogun 2, that is the one game you are using where it is not running on its maximum settings. It's also another game that cannot be run on a 1GB card at 1200p maxed out, another bench better left out because you took it from another source and it's not being run on its highest settings.

Again, I don't disagree with the 480 being a better card than the 5870, again - we knew this 2 years ago. But, the way you are trying to draw a comparison between the 480 and 5870 and use that to predict how kepler will compare to tahiti is not valid. Firstly, because someone could choose a list of DX11 games that gives a completely different summation than your arbritarily decided on list has. Secondly, because you are assuming performance characterists with no information available at all about how an unreleased card will perform.

If someone had used a comparison of the 3870 to the 8800GTX to come to a performance estimate on how the 4870 would compare to the GTX280 they would of been way out in the weeds. You've done more than compare GPU architectures here; you've compared VRAM buffers, driver differences, a suspect benchmark showing 100% gains and different testbeds and testiing methodologies here.

Appreciate your enthusiam and efforts but you've just reiterated the results of 2 year old benchmarks and added in a dose of speculation on an unreleased product with no proven foundation in the data you have here.

I get that there has been nothing new for a long while now to talk about with nvidia graphics cards. So we're going to back 2 years and reiterate on the 480 vs 5870 and try to make it relevant for a new architecture vs an unreleased and unknown architecture ? Really ? Hopefully this does not pan out to be true or we'll be hearing more wild speculations like these for the next seven months until September.

 

[LOL_Wut_Axel]

Is the problem that people don't understand graphics card architectures? The 5870 was a graphics card architecture from 2008 with a tessellation engine and a few other tweaks slapped on. You'll notice in games with heavy tessellation or an NVIDIA-only optimization (Civ5, etc.), the 5870 is behind. Otherwise, the GTX 480 is only 10-15% ahead, as it was at launch. Drivers have done little to differentiate the two cards, only the architectures have.

This. Fans and fanboys alike always like to nitpick benchmarks. The reality is that even taking DX11 into account the GTX 480 is only 15-20% faster on average. It also consumed around 50% more power and ran a lot hotter and louder, too, so that's why it was considered meh (and to this day I still consider it that way). Only card I'd really consider buying from NVIDIA now is the GTX 560 Ti. Hopefully Kepler will change that since there's no way I'm buying AMD's more affordable VLIW4 cards. Maybe if the 1.5GB 7950 ends up at $350.

 

[96Firebird]

...or an NVIDIA-only optimization (Civ5, etc.) that AMD has been too lazy to enable the same feature for quite some time now...

FTFY. :sneaky:

 

[Lepton87]

The good thing about tessellation in current games is that you can just turn it off and barely notice a difference in most cases. (cough Crysis 2 cough)

 

[Makaveli]

Cool, fanboy arguments 2 years after the fact!

I don't really consider this a fanboy argument. Nothing wrong with looking at the cards now after two years of drivers to see how they hold up.

No one is coming in here starting a flame war everyone knows where they cards stand when it comes to performance.

 

[blackened23]

FTFY. :sneaky:

To be fair, 1 game that uses it? And its Civ 5 at that. I'm pretty sure 3 year old hardware can run civ 5 just fine, unless you're really interested in running it at 200 fps.

Personally I find Civ 4 to be the superior game, civ 5 is a big disappointment. Nope, no craps given here about MT rendering.

 

[96Firebird]

To be fair, 1 game that uses it? And its Civ 5 at that. I'm pretty sure 3 year old hardware can run civ 5 just fine, unless you're really interested in running it at 200 fps.

Personally I find Civ 4 to be the superior game, civ 5 is a big disappointment. Nope, no craps given here about MT rendering.

MrK6 said it was an Nvidia-only optimization, when it reality AMD can just as easily use it too. That was the point I was trying to make.

And I played Civ 5 for a bit, but it was the first of the Civilization series that I've played. Hell, the last strategy game I played before that was Ages of Empire III. I didn't play Civ 5 long however, after a few passes through it got boring to me.

 

[AtenRa]

Are you lumping compute in with tessellation ? Because in the games you've used with the largest differences; tessellation looks to be the deciding factor in the differences. You expect this to play out the same way with Tahiti vs Kepler ? I suppose that's possible if nvidia can reduce the performance impact tessellation inflicts on their cards more so than it does currently, but that sounds like a stab in the dark. You might want to re-read the results of multi-threaded rendering for nvidia. Performance doubled, that's a significant gain.

You may not want the one benchmark that helps to prop up your whole opinion more than the rest to be discarded, but when there is a result so supect and defying logic, it needs to be looked at closer. Have you played B:AC ? It looks like a console port and the graphics are not groundbreaking, contrast it to BF3, Metro 2033 or Crysis 2, games that show a reasonable performance disparity and visually are exponentially more impressive, not the ridiculous 100%. The 100% jump seen there begs suspicion in a game that is one of the least impressive graphically in that entire list. It's also a suspect result that most skews your entire summation.

Per Shogun 2, that is the one game you are using where it is not running on its maximum settings. It's also another game that cannot be run on a 1GB card at 1200p maxed out, another bench better left out because you took it from another source and it's not being run on its highest settings.

Again, I don't disagree with the 480 being a better card than the 5870, again - we knew this 2 years ago. But, the way you are trying to draw a comparison between the 480 and 5870 and use that to predict how kepler will compare to tahiti is not valid. Firstly, because someone could choose a list of DX11 games that gives a completely different summation than your arbritarily decided on list has. Secondly, because you are assuming performance characterists with no information available at all about how an unreleased card will perform.

If someone had used a comparison of the 3870 to the 8800GTX to come to a performance estimate on how the 4870 would compare to the GTX280 they would of been way out in the weeds. You've done more than compare GPU architectures here; you've compared VRAM buffers, driver differences, a suspect benchmark showing 100% gains and different testbeds and testiing methodologies here.

Appreciate your enthusiam and efforts but you've just reiterated the results of 2 year old benchmarks and added in a dose of speculation on an unreleased product with no proven foundation in the data you have here.

I get that there has been nothing new for a long while now to talk about with nvidia graphics cards. So we're going to back 2 years and reiterate on the 480 vs 5870 and try to make it relevant for a new architecture vs an unreleased and unknown architecture ? Really ? Hopefully this does not pan out to be true or we'll be hearing more wild speculations like these for the next seven months until September.

First of all, the Shogun 2 slide is from the same review at ThechPowerUp as the rest of the slides. I have calculated its score in to the average performance. Check again the first post, i couldn't post more than 8 pictures (forum settings), so i have just put the link of the slide.

Compute and DX-11 comes together. You need a lot of compute power to calculate all those vertices created with Tessellation and you need compute performance for the rest of the DX-11 features, DX-11 is more than Tessellation.

From the AMD site,
http://sites.amd.com/us/game/technology/Pages/directx-11.aspx

Compute Shaders
Compute Shaders are programs executed on the graphics processor. With DirectX® 11’s DirectCompute feature, application developers can harness the massive parallel processing power of the AMD Radeon HD™ GPU to provide stutter-free playback of Internet and Blu-ray videos, improve video quality or upscale DVDs. In games, Compute Shaders can dramatically improve visual detail:
• Optimized post-processing effects: apply advanced lighting techniques to enhance the mood of any scene.
• High-quality shadow filtering: see shadows fade at their edges just as you would in real life.
• Depth of field: enables more realistic focal points by blurring objects that aren’t in focus—imagine looking down the sight of a rifle.
• Ambient occlusion: generates ultra-realistic lighting and shadow combinations.

In Batman:AC the HD6950 is almost 50% faster than HD5870. Caymans architecture is more Compute oriented than Evergreen and its Tessellation units have been enhanced over the Evergreen units. The result is HD6970 is 64% faster than HD5870. And again GCN is even more Compute oriented with more Tessellation enhanced units over Cayman and we get another 64% more performance from HD6970 to HD7970.

HD6970 64% faster vs HD5870
HD7970 64% faster vs HD6970

I believe it is getting clearer and clearer that HD5870 lack of Compute and Tessellation performance really shows its age in this game. There is no suspicion about the game performance of GTX480 vs HD5870 in this game, it is clearly an architecture design win for the GTX480 and for the GCN over Cayman.

Same thing with Crysis 2 DX-11, HD7970 is 63% faster than HD6970. But when GTX580 was 30% faster than HD6970, people believed that NVIDIA coded the DX-11 patch in order to hurt AMDs cards. GCN scaling vs Cayman in this game is amazing.

I’m not trying to predict Kepler vs Tahiti using GTX480 vs HD5870. I have only said that, we should not expect the Kepler vs Tahiti performance difference in those games to be like HD5870 vs GTX480 of 2010 with DX-9 games. This time AMD have lost its advantage (VLIW architecture) and they are going to play in Compute/Tessellation courtyard against a Compute/Tessellation heavyweight GK100 Kepler.

 

[blackened23]

MrK6 said it was an Nvidia-only optimization, when it reality AMD can just as easily use it too. That was the point I was trying to make.

And I played Civ 5 for a bit, but it was the first of the Civilization series that I've played. Hell, the last strategy game I played before that was Ages of Empire III. I didn't play Civ 5 long however, after a few passes through it got boring to me.

Gotcha. Yeah, Civ 5 wasn't NV only optimization, it just plays better with MT rendering. AFAIK its the only game that has it, I could be wrong though.

 

[blackened23]

First of all, the Shogun 2 slide is from the same review at ThechPowerUp as the rest of the slides. I have calculated its score in to the average performance. Check again the first post, i couldn't post more than 8 pictures (forum settings), so i have just put the link of the slide.

Compute and DX-11 comes together. You need a lot of compute power to calculate all those vertices created with Tessellation and you need compute performance for the rest of the DX-11 features, DX-11 is more than Tessellation.

From the AMD site,
http://sites.amd.com/us/game/technology/Pages/directx-11.aspx

In Batman:AC the HD6950 is almost 50% faster than HD5870. Caymans architecture is more Compute oriented than Evergreen and its Tessellation units have been enhanced over the Evergreen units. The result is HD6970 is 64% faster than HD5870. And again GCN is even more Compute oriented with more Tessellation enhanced units over Cayman and we get another 64% more performance from HD6970 to HD7970.

HD6970 64% faster vs HD5870
HD7970 64% faster vs HD6970

I believe it is getting clearer and clearer that HD5870 lack of Compute and Tessellation performance really shows its age in this game. There is no suspicion about the game performance of GTX480 vs HD5870 in this game, it is clearly an architecture design win for the GTX480 and for the GCN over Cayman.

Same thing with Crysis 2 DX-11, HD7970 is 63% faster than HD6970. But when GTX580 was 30% faster than HD6970, people believed that NVIDIA coded the DX-11 patch in order to hurt AMDs cards. GCN scaling vs Cayman in this game is amazing.

I’m not trying to predict Kepler vs Tahiti using GTX480 vs HD5870. I have only said that, we should not expect the Kepler vs Tahiti performance difference in those games to be like HD5870 vs GTX480 of 2010 with DX-9 games. This time AMD have lost its advantage (VLIW architecture) and they are going to play in Compute/Tessellation courtyard against a Compute/Tessellation heavyweight GK100 Kepler.

What I can do is talk about tessellation more generally. But before we get to that, let's talk a little about history.

For most of the previous generation "big chips" (the first one to come out on the market of a particular architecture), there have been one or more features tied specifically to those big chips. For example, with geForce 3, programmable shaders were added--but they were very limited in what they could do. With geForce FX (the 5000 series), long shaders were added, and for the first time you could perform math operations in your shader before you did a texture lookup. With the 8000 series, Geometry Shaders were added. With the 4xx series, tessellation was added.

Sometimes ATI/AMD released these chips/features a little before NVIDIA--that's not the point here. The point is that with the exception of tessellation, every one of those first generation chips had something in common: the first chips that came out with a new feature were terrible at it. Geometry Shaders were so bad in the 8800 that both NV and AMD literally told developers "please don't use this." This was true for long shaders in the geForce FX series, and even way back in the geForce 3 days developers were told to keep their shaders as simple as possible.

Along came tessellation. As it turned out, NV were good at this. Amazing, actually. Unfortunately, AMD was not. The reasons are somewhat boring, but the truth is the reasons behind the tessellation performance was a glorious accident that turned out in NV's favor.

Now here's the dilemma for tessellation: how do you make comparisons between the vendors fair? Selling visual fidelity is hard. It's hard to tell a user: "look, our framerate is lower but our picture is prettier." Consumers don't listen even when it's true. Consumers long ago indicated that FPS was their one true metric for "I bought card A over card B". That means that when something like this comes along, the developer can either hobble the fair comparison between two cards by significantly reducing the workload on one GPU or the other, or they can make the comparison fair and people will cry that the performance on the losing vendor has been intentionally hobbled.

 

[MrK6]

MrK6 said it was an Nvidia-only optimization, when it reality AMD can just as easily use it too. That was the point I was trying to make.

So if NVIDIA is the only one using it, then it's still an NVIDIA-only optimization, isn't it? Honestly, it's like some people try to make trouble on these boards.

First of all, the Shogun 2 slide is from the same review at ThechPowerUp as the rest of the slides. I have calculated its score in to the average performance. Check again the first post, i couldn't post more than 8 pictures (forum settings), so i have just put the link of the slide.

Compute and DX-11 comes together. You need a lot of compute power to calculate all those vertices created with Tessellation and you need compute performance for the rest of the DX-11 features, DX-11 is more than Tessellation.

From the AMD site,
http://sites.amd.com/us/game/technology/Pages/directx-11.aspx

In Batman:AC the HD6950 is almost 50% faster than HD5870. Caymans architecture is more Compute oriented than Evergreen and its Tessellation units have been enhanced over the Evergreen units. The result is HD6970 is 64% faster than HD5870. And again GCN is even more Compute oriented with more Tessellation enhanced units over Cayman and we get another 64% more performance from HD6970 to HD7970.

HD6970 64% faster vs HD5870
HD7970 64% faster vs HD6970

I believe it is getting clearer and clearer that HD5870 lack of Compute and Tessellation performance really shows its age in this game. There is no suspicion about the game performance of GTX480 vs HD5870 in this game, it is clearly an architecture design win for the GTX480 and for the GCN over Cayman.

Same thing with Crysis 2 DX-11, HD7970 is 63% faster than HD6970. But when GTX580 was 30% faster than HD6970, people believed that NVIDIA coded the DX-11 patch in order to hurt AMDs cards. GCN scaling vs Cayman in this game is amazing.

I’m not trying to predict Kepler vs Tahiti using GTX480 vs HD5870. I have only said that, we should not expect the Kepler vs Tahiti performance difference in those games to be like HD5870 vs GTX480 of 2010 with DX-9 games. This time AMD have lost its advantage (VLIW architecture) and they are going to play in Compute/Tessellation courtyard against a Compute/Tessellation heavyweight GK100 Kepler.

I agree. I think this goes to show how much leg work AMD did in developing VLIW4, and it was much more than just clipping an extra shader to increase efficiency. Another game that shows remarkable improvements on VLIW4 is Metro 2033. However, I also think the weaker tessellation engine on Cayman gave the biggest performance impact in Crysis 2. The fact that the ocean is tessellated beneath entire levels shows a gross negligence in coding efficiency. One of the big problems with PC gaming today is coders/porters getting lazy because of the excess power available to us, which is no excuse IMO. But I digress.

 

[blckgrffn]

Geometry Shaders were so bad in the 8800 that both us and AMD literally told developers "please don't use this." This was true for long shaders in the geForce FX series, and even way back in the geForce 3 days developers were told to keep their shaders as simple as possible.

Alright Jen-Hsun Huang, the gig is up

 

[96Firebird]

So if NVIDIA is the only one using it, then it's still an NVIDIA-only optimization, isn't it?

If that's they way you think, there are bigger problems... :|

Honestly, it's like some people try to make trouble on these boards.

Yes, can you please tell us why that is? We are really getting sick of it...

 

[MrK6]

If that's they way you think, there are bigger problems... :|
Yes, can you please tell us why that is? We are really getting sick of it...

Clearly you're here to just insult other members or start a flame war, neither of which is worth my time.

 

[tviceman]

Is the problem that people don't understand graphics card architectures? The 5870 was a graphics card architecture from 2008 with a tessellation engine and a few other tweaks slapped on. You'll notice in games with heavy tessellation or an NVIDIA-only optimization (Civ5, etc.), the 5870 is behind. Otherwise, the GTX 480 is only 10-15% ahead, as it was at launch. Drivers have done little to differentiate the two cards, only the architectures have.

Nvidia only optimization. As in, only Nvidia's software guys know how to write a proper dx11 driver. NEWSFLASH: Multi-threaded rendering is an Nvidia-only optimization that Microsoft included in the DX11 specs.

 

[tviceman]

Gotcha. Yeah, Civ 5 wasn't NV only optimization, it just plays better with MT rendering. AFAIK its the only game that has it, I could be wrong though.

Battlefield 3 does as well. https://www.google.com/webhp?source...b&fp=8bfec2db87b5ba25&ion=1&biw=1920&bih=1113 But apparently not to the same extent Civ V does.

 

[blackened23]

Battlefield 3 does as well. https://www.google.com/webhp?source...b&fp=8bfec2db87b5ba25&ion=1&biw=1920&bih=1113 But apparently not to the same extent Civ V does.

Hmm, I think bf3 supports multiple threads from CPU's but not multi threaded rendering. One of the bf3 patches increased HT performance in bf3 a while back, I remember that much.....

I'm not 100% certain on this, but this is what i've heard through the grapevine. (regarding cpu vs gpu)

 

[MrK6]

Nvidia only optimization. As in, only Nvidia's software guys know how to write a proper dx11 driver. NEWSFLASH: Multi-threaded rendering is an Nvidia-only optimization that Microsoft included in the DX11 specs.

So NVIDIA got around to it 20+ months after it was first spec'd as an optional inclusion, and AMD still hasn't. What are you arguing, that NVIDIA sucks less? Or are you just here to argue? How does any of what you said above change the fact that right now, only NVIDIA supports that optimization, and so obviously an AMD part would fall behind in a game that uses it? Again, the NVIDIA white-knighting is getting old. This was a nice discussion until certain people hulked out, green with rage.

 

[tviceman]

This was a nice discussion until certain people hulked out, green with DX11 MULTI-THREADED RENDERING unfair Nvidia-only optimization, rage.

Fixed for ya.

 

[MrK6]

Fixed for ya.

Your rage is still rage and it derails the thread. Let's steer this back to the discussion.

 

[blastingcap]

Newer, pricier, power-hungrier card performs better than older, cheaper, less-power-hungry card. What is the news here?

The extra power draw over the course of the last 2 years just widens the price differential between the HD 5870 and GTX 480 even more.

If you wanted the best of both worlds, you could have just gotten a HD 5870 at launch, then switch to the GTX 580 when it came out.

 

[tviceman]

Your rage is still rage and it derails the thread. Let's steer this back to the discussion.

And your lies are still lies so when you stick to the truth I will quit calling you out on it, oh sorry I mean I will quit derailing the thread by calling your lies out.

I'm not even going to bother trying to clean this up
-ViRGE

 

[Vesku]

Newer, pricier, power-hungrier card performs better than older, cheaper, less-power-hungry card. What is the news here?

The extra power draw over the course of the last 2 years just widens the price differential between the HD 5870 and GTX 480 even more.

If you wanted the best of both worlds, you could have just gotten a HD 5870 at launch, then switch to the GTX 580 when it came out.

What he said +1.

 

[Spjut]

To be fair, 1 game that uses it? And its Civ 5 at that. I'm pretty sure 3 year old hardware can run civ 5 just fine, unless you're really interested in running it at 200 fps.

Personally I find Civ 4 to be the superior game, civ 5 is a big disappointment. Nope, no craps given here about MT rendering.

FWIW, Capcom said they didn't use it in LP2 because AMD and Nvidia didn't support it...
AMD still doesn't support it, so what incentive are there for devs? If anything, they may just get accused of being Nvidia optimised

The neat thing with DX11 multithreaded rendering is that it can be backported to DX10/10.1 cards, so all DX10+ users would benefit...Certainly not a little feature to support...

Hmm, I think bf3 supports multiple threads from CPU's but not multi threaded rendering. One of the bf3 patches increased HT performance in bf3 a while back, I remember that much.....

I'm not 100% certain on this, but this is what i've heard through the grapevine. (regarding cpu vs gpu)

Repi confirmed on Twitter(or if it was beyond3d) that they ended up not using multithreaded rendering, that it was too slow for their "use case"