Civ V oddity with video cards.....

[AtenRa]

The point was that Civ 5 is unbalanced more than the difference between cards should be when looking at 80% or so of other tests/comparisons. The % difference in Civ 5 is abnormal so as a comparison, it is a crap test.

So by your logic, we also have to dismiss DE3 as a crap test because HD6970 is as fast as GTX580 in that game, when we all know that GTX580 is 15% faster in 80% of the other tests, same applies for Formula 1 2010 where HD6970 again is equal or faster than GTX580 and in Alien vs Predator and so on and so on. I haven’t seen anyone mention that before for those games and all three of the above are AMD Gaming Evolve. :whiste:

But we don’t use Civ V, HAWX2, Lost Planet 2, Crysis 2 DX11 and more because Fermi cards have better performance than the competition so the tests are crap.

This is getting serious, people should understand that HD69xx and GTX5xx are different in design and some games will run better with one hardware and other games will run better with the other hardware.

One game engine will take advantage of NVIDIA's design characteristics like Hawx2, Civ V etc etc and other game engines will take AMD's design strengths like DE3, F1 2010, Alien vs Predator do.

That’s why we need to see the whole picture and Review sites need to bench more than 3-4 games.

 

[tviceman]

That’s why we need to see the whole picture and Review sites need to bench more than 3-4 games.

Apoppin is just screaming right now for people to look at his reviews at alienbabeltech. The guy runs 25+ benchmarks (all except for 2-3 are games) with every gpu review.

 

[bryanW1995]

Well, I do care about Civ V, so I bought last night an eVGA Superclocked 560 Ti as a substitute card. Would have liked one of those Sapphire Toxic 6950 2GB cards, but someone keeps buying them!


And, as it turns out, the 6950 that's in the computer right now, isn't bad. It only acted up in Civ V. Running Furmark got no faults from it......so probably the tess./whatever Civ has running that the AMD card doesn't like was the culprit. The card also was perfect running Starcraft 2 and anything else I threw at it.

Go back and follow the link that happymedium posted at the beginning of this thread. If AMD would ever add driver support for multi-threaded rendering then we'd see a very large/immediate improvement in performance. However, in this instance nvidia beat them to the punch and amd is just ignoring this game in general and multi-threaded rendering in particular for some reason.

Here's the link again if you don't want to search for it.

http://forums.anandtech.com/showpost.php?p=31520674&postcount=28

edit: @tviceman: I agree, AlienBabelTech is always on my list of review sites when a new card comes out. Apoppin does a great job over there.

 

[tviceman]

However, in this instance nvidia beat them to the punch

Hahahaha beating someone to the punch makes it sound like a close race. This isn't even as "close" as the six month head start AMD got with DX11 GPU's. I think this zeroes back to the tired (and often obsolete) argument that Nvidia has better drivers. While no longer "better," and despite their tardiness with next gen hardware lately, Nvidia does always seem to catch on to or embrace new features within their hardware faster (historically) than AMD/ATI.

 

[arredondo]

So is the primary fault with the developer for favoring one GPU company, or with AMD for not releasing better drivers addressing the tech in this game months after it launched? Does AMD need permission from the Civ V team to make a good driver for it?

 

[AtenRa]

ok guys here we go,

Civilization 5 CPU performance analysis DX-11

I have used the on-game "LateGameView" Benchmark with 60 secs on FRAPS.

Resolution 1920x1200 MAX settings, 4x MSAA, Tessellation High

My signature system with a single HD6950 default at 810MHz core 1250MHz mem (Second vga will be available in week end). AMD Catalyst 11.5 V-sync OFF.

2.66GHz No HT = Min: 4, Max: 31, Avg: 27.40
2.66GHz + HT = Min: 4, Max: 30, Avg: 27.11
3.36GHz No HT = Min: 6, Max: 36, Avg: 33.06
3.36GHz + HT = Min: 8, Max: 35, Avg: 32.50
4GHz + No HT(edit) = Min: 14, Max: 41, Avg: 36.93
4GHz + HT = Min: 15, Max: 41, Avg: 37.11

njoy

Edit: correction in 4GHz

 

[happy medium]

ok guys here we go,

Civilization 5 CPU performance analysis DX-11

I have used the on-game "LateGameView" Benchmark with 60 secs on FRAPS.

Resolution 1920x1200 MAX settings, 4x MSAA, Tessellation High

My signature system with a single HD6950 default at 810MHz core 1250MHz mem (Second vga will be available in week end). AMD Catalyst 11.5 V-sync OFF.

2.66GHz No HT = Min: 4, Max: 31, Avg: 27.40
2.66GHz + HT = Min: 4, Max: 30, Avg: 27.11
3.36GHz No HT = Min: 6, Max: 36, Avg: 33.06
3.36GHz + HT = Min: 8, Max: 35, Avg: 32.50
4GHz + HT = Min: 14, Max: 41, Avg: 36.93
4GHz + HT = Min: 15, Max: 41, Avg: 37.11

njoy

It looks like you are getting about the same averages as the Anandtech graph I posted at post#4 with a 6950.

Take your benchmarks and add about 30% and you have a gtx560ti late game benchmarks. And remember Tviceman also has his gtx560ti overclocked 20%+ on top of that. Tviceman should have near 60fps avg, just like a gtx570.

Thanks for the mini review.

 

[tviceman]

I was going to pop up a comparison graph for a gtx560ti, but then I realized I would need to redownload fraps. And then I realized I would need to download openoffice to read/convert the csv file. And then my motivation flew out the window.

Take my word for it Atena, it runs smooth with an overclocked gtx560ti in the late mid/late game with 9 AI's, 16 city states, and on emperor difficulty (which means the AI is producing a crazy amount of units due to it's production advantages).

 

[Spjut]

BTW, I'm embarrassed for AMD that they STILL haven't solved this issue. Apparently nv gave Ryan permission to discuss it in that thread b/c they knew that either amd wouldn't or couldn't add support for that anytime soon. In 5 months civ5 has gone from a very popular/current game to old news, by the time amd gets their support up for it the game will be well over a year old.

I agree....Anandtech's DX11 article also mentions that DX10/10.1 cards will be able to use the multithreading(though they need driver work to work to the fullest extent, haven't read anything if Nvidia has done so for their DX10 hw), so enabling support for it would give a large amount of systems better performance, not only the HD5000/HD6000 series(they probably don't give a damn about their older series now though)

Other devs have mentioned they didn't include DX11 MTR because of lack of driver support

Well, I at least hope AMD gets it in in time for the Battlefield 3 beta

 

[bryanW1995]

So is the primary fault with the developer for favoring one GPU company, or with AMD for not releasing better drivers addressing the tech in this game months after it launched? Does AMD need permission from the Civ V team to make a good driver for it?

IIRC, neither company supported that feature when civ 5 released. Nvidia got a big jump after they updated their drivers. Clearly this is an outlier, it just surprises me that after all of this time amd still hasn't done it. I think that I commented in ryan smith's thread that I thought it likely that it would take a while for amd to catch up b/c nvidia let him post exactly how they implemented the feature. My guess is that it's enough of a PITA on amd's 40nm architectures that they're waiting for 28nm to come out before they implement it. Call it whatever you want, but this is the first instance that I've personally experienced in which nvidia clearly outflanked amd in the driver dept. They deserve kudos for it.

edit: if BF3 has MTR then amd really needs to get their act together and get that driver out ASAP.

 

[greenhawk]

So is the primary fault with the developer for favoring one GPU company,

depends on what you read. Some like to point out that nVidia paid the developers some "assistance" money so they "over did" the features that nVidia's card would do better with vs what was expected from AMD.

ie: nVidia played dirty.

 

[greenhawk]

So by your logic

if you want to take what I typed and run off into the woods with it, that is up to you. I am just saying that Civ 5 is bias in how it works with GPU's. You can either live with it or point it out.

no matter what is said, I will always judge a GPU based on price difference vs performance difference in games I play.

if AMD cards perform the same as nVidia cards at a far cheaper price, then all the better for the consumer.

As to formual 1, no idea about that one as I do not play racing games.

As to Aliens vs Predator, I have no issues with AMD vs nVidia in that type of game. Not because of the current cards, but more to do with the developer of that game seeing is it is a console port. If a current gen GPU can not max out a console port made for hardware several years old, then I would be very very displeased with the video card.

If someone wanted to make a point of that sort of "being equal" and used console ports as the reason for taking one card over another, then I will call that person a fanboy.

To me, using console ports as the reason for choosing one GPU over another is like compairing a ford car vs a Ferarri because they both can get you from point A to point B.

But we don’t use Civ V, HAWX2, Lost Planet 2, Crysis 2 DX11 and more because Fermi cards have better performance than the competition so the tests are crap.

if you like those games, then the comparisons are not crap.

If you know WHY someone performs as it does, then you can better match what you want with what a product can offer.

This is getting serious, people should understand that HD69xx and GTX5xx are different in design and some games will run better with one hardware and other games will run better with the other hardware.

/clap clap

it is what I have been saying. You buy for what you want to use it for, not for any other reason.

The OP wants to play Civ 5 at it's best. Civ 5 favours nVidia's current card offerings. Therefore, he needs a nVidia card. Noting is being mentioned about nVidia being better than AMD or vis versa.

it is 1 + 1 = 2.

Nothing else of reasonable worth can be pulled from comparing GPUs with Civ 5 as a bentchmark .

One game engine will take advantage of NVIDIA's design characteristics like Hawx2, Civ V etc etc and other game engines will take AMD's design strengths like DE3, F1 2010, Alien vs Predator do.

Exactly. And the end user has the choice of deciding which of those games they perfer more / play more and buy accordingly.

 

[AtenRa]

no matter what is said, I will always judge a GPU based on price difference vs performance difference in games I play.

Correct, but just because YOU dont play Civ V doesnt make it a crap benchmark.

if AMD cards perform the same as nVidia cards at a far cheaper price, then all the better for the consumer.

Correct, but if you only bench 3 AMD Game Evolve games, the consumer may come to a wrong conclusion, unless they only play those 3 games.

If the consumer want to play for example HAWX2 or Civ V, then NV cards have better performance/price.

As to Aliens vs Predator, I have no issues with AMD vs nVidia in that type of game. Not because of the current cards, but more to do with the developer of that game seeing is it is a console port. If a current gen GPU can not max out a console port made for hardware several years old, then I would be very very displeased with the video card.

If someone wanted to make a point of that sort of "being equal" and used console ports as the reason for taking one card over another, then I will call that person a fanboy.

To me, using console ports as the reason for choosing one GPU over another is like compairing a ford car vs a Ferarri because they both can get you from point A to point B.

Aliens vs. Predator (PC) supposed to kick consoles‘ ass - at least technology wise

PCGH: You announced that your game is developed for PC, Xbox 360 and PS3. Will the engine be a pure cross-platform product or will there be an optimized version for the PC? If there is a special built for the PC what technical feature can't be realized with the console version or in other words are there any features the PC is the ideal platform to develop for?

Kevin Floyer Lea: The PC version of AvP is fully able to take advantage of the features of all the latest video cards and is optimized to make efficient use of multicore threading if your PC supports it. We will also be one of the first games to really show off how good DirectX 11 games can look, with hardware tessellation of the Aliens using displacement mapping, and Shader Model 5.0 Compute Shaders for post-processing effects.

PCGH: Are there any differences between the Console and the PC Version as far as technical as well as visual aspects are concerned?

Kevin Floyer Lea: All the art assets are authored at higher resolutions than can be squeezed onto current console hardware so, while AvP looks fantastic on Xbox 360 and PS3, it looks truly state-of-the-art on PC. Similarly effects such as shadows and full-screen passes like ambient occlusion are more complex and detailed on PC.

AvP clearly is not a Console Port

if you like those games, then the comparisons are not crap.

So why did you say Civ V is crap in the first place??

/clap clap

it is what I have been saying. You buy for what you want to use it for, not for any other reason.

NO, you said Civ V is crap and Biased as a benchmark

The OP wants to play Civ 5 at it's best. Civ 5 favours nVidia's current card offerings. Therefore, he needs a nVidia card. Noting is being mentioned about nVidia being better than AMD or vis versa.

it is 1 + 1 = 2.

Nothing else of reasonable worth can be pulled from comparing GPUs with Civ 5 as a bentchmark .



Exactly. And the end user has the choice of deciding which of those games they perfer more / play more and buy accordingly.

i rest my case

 

[AtenRa]

It seams Catalyst 11.8 increase performance :thumbsup:

4GHz + HT = Min: 15, Max: 41, Avg: 37.11
4GHz + HT Cat 11.8 = Min: 21, Max: 43, Avg: 39.60

 

[Meghan54]

Well, after using both cards in the same game at the same resolution and settings, I can definitely say the 560 Ti is a little faster but the graphics it produces, esp. the buttons, etc., are not as well produced as the 6950's graphics output. The 560Ti's images in some areas are very much more cartoonish looking compared to the 6950.

C'est la vie.

And to top it off, I started getting random crashes of the Nvidia driver, their latest version......the 280.26 package. Put the next older driver package in.....275.33 and the crashes seem to have stopped.

Never ever had a crash of the ATi drivers. And I used Driver Cleaner after uninstalling the AMD card and before installing the Nvidia card and drivers.

 

[bryanW1995]

The OP wants to play Civ 5 at it's best. Civ 5 favours nVidia's current card offerings. Therefore, he needs a nVidia card. Noting is being mentioned about nVidia being better than AMD or vis versa.

Actually, we don't know if civ5 is inherently better on current-gen nvidia cards or not. All that we know is that nvidia's driver dept has been on the ball vis a vis this issue while amd's driver dept has been resting on their laurels and/or playing too much minesweeper. Several of the other games that you talk about about do have an architectural advantage for nvidia, however, and many other games have that same advantage for amd.

 

[bryanW1995]

Well, after using both cards in the same game at the same resolution and settings, I can definitely say the 560 Ti is a little faster but the graphics it produces, esp. the buttons, etc., are not as well produced as the 6950's graphics output. The 560Ti's images in some areas are very much more cartoonish looking compared to the 6950.

C'est la vie.

And to top it off, I started getting random crashes of the Nvidia driver, their latest version......the 280.26 package. Put the next older driver package in.....275.33 and the crashes seem to have stopped.

Never ever had a crash of the ATi drivers. And I used Driver Cleaner after uninstalling the AMD card and before installing the Nvidia card and drivers.

I've had several issues with the leaderheads causing crashes. To fix it you have to turn them on low resolution leaderheads or just no animation for them at all. That is annoying, but even on low res they still look really good.

 

[Meghan54]

Actually, the crashes were while the computer was idling on the desktop, not during gaming. Never got to playing the game before the driver began crashing randomly. An hour or two between driver failures, but definitely the Nvidia driver causing it.

I put the older driver package on and the problem seems to have gone away....for now, at least. And hasn't crashed in the game yet.....cross my fingers.


Honestly, I'm about to put the 6950 back in and dump the 560Ti at a loss.

 

[bryanW1995]

If that's the case let me know, I wouldn't mind getting a 560 Ti.

 

[AtenRa]

One more run with HD6950 1GB CossFire setup at 4GHz + HT Cat 11.8 and Power Tune at +20% (Power Tune was at +20% in the previous runs as well)

Tessellation in CCC was set to "use Application Settings" in all runs

4GHz + HT Cat 11.8 = Min: 21, Max: 43, Avg: 39.60
4GHz + HT Cat 11.8 = Min: 09, Max: 63, Avg: 55.06

 

[tviceman]

I became unlazy and downloaded the necessary stuff to benchmark and convert it to a graph. Here are my results:

i7 2600k @ 4.5 ghz HT disabled
gtx560ti @ 1000mhz core, 2300mhz memory
8 gigs ram
Anti-aliasing is set in game.

min, max, average reported by fraps: 62 min, 76 max, 65.433 average

 

[happy medium]

I became unlazy and downloaded the necessary stuff to benchmark and convert it to a graph. Here are my results:

i7 2600k @ 4.5 ghz HT disabled
gtx560ti @ 1000mhz core, 2300mhz memory
8 gigs ram
Anti-aliasing is set in game.

min, max, average reported by fraps: 62 min, 76 max, 65.433 average

Thats gotta be the most consistant benchmark I've ever seen, that line is allmost straight. :thumbsup:

 

[ch33kym0use]

No problem with Civ V here for me with nvidia card.

 

[Spjut]

No problem with Civ V here for me with nvidia card.

What card is it?

 

[tviceman]

For sh*ts and grins, I also ran the benchmark on my notebook.

i7 2630QM @ stock (2.0 ghz HT on)
gtx560m @ stock
8 gigs ram
Anti-aliasing set to 1x (off)

With a significantly slower quad core processor and mobile graphics processor with performance about equal to a gts450 desktop GPU, it still averaged 33.783 fps at 1080p and detail settings maxed out. Not bad!