GTX 580 OC vs 6870 OC vs 5970 Review [HardwareHeaven.com]

W

WorldExclusive

[H]F Junkie
Joined
Apr 26, 2009
Messages
11,548
[HardwareHeaven.com]

That comparison is the GTX 580, overclocked to the fastest speed we have seen advertised by a manufacturer on standard air cooling against the 6870 in Crossfire with the same highest overclock rule. In addition to that we will also add the 5970, in full, to the article as it shares the same price as the 6870 CF and 580 OC. Each of the cards will be tested in a selection of the latest games including Call of Duty: Black Ops, F1 2010 (DX11 Patch) and Fallout: New Vegas with the latest patch which fixes performance issues.
Click to expand...

Source
http://www.hardwareheaven.com/revie...oc-crossfire-vs-radeon-5970-introduction.html
 
all in all

[QUOTE="Hardwareheaven]Conclusion

As stated earlier in this article the purpose of these tests was to take a look at how one of the top GTX 580s will compare to the older 5970 dual GPU card and the latest 6870 Crossfire (Overclocked) system as all share a similar price point.

If we are looking purely at performance it is fair to say that the dual GPU systems are faster in a number of benchmarks, sometimes by a significant amount. As always though there are caveats. The first and main being the reliance on driver quality and profiles. Looking at Aliens vs. Predator as an example we had a great experience on the dual GPU Radeons however F1 2010 was the complete opposite and basically is unplayable on the AMD cards.

The GTX, like the 6870 Crossfire system, had its highlights too with Call of Duty: Black Ops and HAWX 2 performing well on the latest NVIDIA card.

When deciding which product is best each consumer should take these results AND those from our original launch review which also covers additional features of each card. These include PhysX (LINK), 3D gaming (LINK), 3D Blu-Ray (LINK), GPU Computing (LINK), surround gaming (LINK) and power, noise and temperatures (LINK). As a quick summary the GTX 580 runs quieter, cooler and requires less power than the AMD cards. It also has a vastly superior 3D solution when compared to AMD. The 6870 excels in media use where it can output Dolby TrueHD and DTS Master Audio and has the benefit of being able to power 3 screens from one card.

If we had to pick one of the systems from this or the original launch review as gamers, we would lean towards the GTX 580. Our reasoning? A solid gaming experience at 1920x1080, no reliance on driver profiles, low noise cooling with smooth fan speed increases and the great 3D/PhysX solution.

Everyone has different needs though and we look forward to hearing what you think about this interesting comparison in our forums.[/QUOTE]
 
gamingarena23 said:
Heh again they OC both 580GTX to the max and 6870Xfire but not the 5970 they left it at 720/1000mhz? which actually is the only card from the shootout that can be overclocked easy 20% by default and up to 30% or more...
Click to expand...

I was wondering about that........
 
gamingarena23 said:
Heh again they OC both 580GTX to the max and 6870Xfire but not the 5970 they left it at 720/1000mhz? which actually is the only card from the shootout that can be overclocked easy 20% by default and up to 30% or more...
Click to expand...

I think the 5970 did enough damage at stock, lol. OCing the card would have made the review pointless.
It's still two 5870s on one card downclocked only because of power consumption. I never thought the 580 could beat the 5970 overall.
Some of our expectations are too high i.e. 6970. The rumors point to the 6990 having two 6950 gpus, but if they could down clock two 6970 gpus for the 6990, I would buy one no matter how sketchy the drivers are.
 
Last edited:
I don't know how they can keep a straight face when posting benchmarks from Hawx 2, the only game or benchmark in history where any card from a 460 768 MB on up can beat a 5970.

How can anyone see that and not just toss it out saying "Well this game is screwed up. We're tossing it until Ubisoft figures out how to make a game engine that works right."
 
fattypants said:
I don't know how they can keep a straight face when posting benchmarks from Hawx 2, the only game or benchmark in history where any card from a 460 768 MB on up can beat a 5970.

How can anyone see that and not just toss it out saying "Well this game is screwed up. We're tossing it until Ubisoft figures out how to make a game engine that works right."
Click to expand...

Nvidia usually has review guidelines/recommendations, some reviewers follow them and others don't.
 
WorldExclusive said:
I think the 5970 did enough damage at stock, lol. OCing the card would have made the review pointless.
Click to expand...

Doesn't it make you wonder what the point of the review was? What did the reviewer want to show us by overclocking everything but the most overclockable card in the review, namely the 5970? :confused:

Edit:
Especially considering that the 5970 can be had for $499.99 at newegg.com ($469.99 with rebate), making it a cheaper alternative then the GTX 580:
newegg 5970
 
Last edited:
So you would just toss a benchmark if you don't like the result? Seems pretty damn shady; glad reviewers tend to have more credibility than your average Joe.
 
kd5ywa said:
So you would just toss a benchmark if you don't like the result? Seems pretty damn shady; glad reviewers tend to have more credibility than your average Joe.
Click to expand...

Doesn't seem like people have questioned or even discussed the results. Its rather the methods thats being reacted upon. :)
 
Tamlin_WSGF said:
Doesn't it make you wonder what the point of the review was? What did the reviewer want to show us by overclocking everything but the most overclockable card in the review, namely the 5970? :confused:

Edit:
Especially considering that the 5970 can be had for $499.99 at newegg.com ($469.99 with rebate), making it a cheaper alternative then the GTX 580:
newegg 5970
Click to expand...

I do wonder why the 5970 was included, I mainly posted it because of the 6870CF OC comparison.
 
WorldExclusive said:
I do wonder why the 5970 was included, I mainly posted it because of the 6870CF OC comparison.
Click to expand...

The reviewer explains that he included it because it costs the same. :)

But why include it in an OC review and not OC the card? Especially since the 5970 is two 5870 GPU's downclocked. That card is made to be OC'd and even advertised so.
 
kd5ywa said:
So you would just toss a benchmark if you don't like the result? Seems pretty damn shady; glad reviewers tend to have more credibility than your average Joe.
Click to expand...

Its actually the first thing i've looked for in the 580 reviews, if they use HAWX2 i move on. As a reviewer i wouldn't even take a chance at credibility by using it.
 
Because AMD said so? Lol, how is boycotting it at AMD's request any better than including it at nVidia's request? Y'all are funny.
 
WTF? they averaged 93 fps in Mafia 2 with AA and Physx on high with a gtx580? how is that even possible that they got that high of fps? there is no way that is possible based on reviews I have seen with a gtx480 getting half of that with those settings.

at the same time how is that as ugly Black Ops game that demanding. that's sad when a console port can only get 60-70fps on the fastest gpus out there.
 
trinibwoy said:
Because AMD said so? Lol, how is boycotting it at AMD's request any better than including it at nVidia's request? Y'all are funny.
Click to expand...

No its because as a reviewer once you are "labeled" you do lose some credibility.. I just would never involve myself as a reviewer in a controversy between the 2 card makers, it has nothing to do with the benchmark itself.
 
H-street said:
No its because as a reviewer once you are "labeled" you do lose some credibility.. I just would never involve myself as a reviewer in a controversy between the 2 card makers, it has nothing to do with the benchmark itself.
Click to expand...

Labelled by whom? In any case, more information is better than less. If the reviewer wants to distance himself from the issue he can put a disclaimer on the result. Competely suppressing it like AMD wants is ridiculous. What's next - boycott any game that doesn't make AMD look good?
 
trinibwoy said:
Labelled by whom? In any case, more information is better than less. If the reviewer wants to distance himself from the issue he can put a disclaimer on the result. Competely suppressing it like AMD wants is ridiculous. What's next - boycott any game that doesn't make AMD look good?
Click to expand...

You can say what you want, but the proof is in the reviews... How many reviews have you read that have used HAWX2?

I counted 2 , This one and legitreviews.com (Review)

it has nothing to do with making "Amd look bad" or "Nvidia look good", its all about using something that may be "tainted" and more specifically not being a "stooge" to any company. Even if AMD didn't say anything about HAWX2, the very fact that Nvidia was directing reviewers to use it should be enough to raise an eyebrow.

If you don't get the issue then there isn't much more to be discussed, but the proof of the review sites and the fact that very few of them have used it even though Nvidia has asked them specifically to us it. Which to me speaks volumes for the integrity of reviewers in general but that is for another discussion.
 
The review was mediocre. Resolution was pretty low for a lot of gamers.

The conclusion is weird...
The 6870 crossfire solution won 5 of 8, and took 2nd place in 3 of 8.
The 580 won 2 of 8, and took 3rd place in 6 of 8

For the same price, it would appear the 6870 CF solution was dominate, yet they recommend to purchase a 580 because it's a single GPU. They obviously didn't have problems downloading profiles to run the games, so it's not that hard. Seems strange they would advise purchasing the lowest performing card of the 3 options.
 
H-street said:
You can say what you want, but the proof is in the reviews... How many reviews have you read that have used HAWX2?

I counted 2 , This one and legitreviews.com (Review)

it has nothing to do with making "Amd look bad" or "Nvidia look good", its all about using something that may be "tainted" and more specifically not being a "stooge" to any company. Even if AMD didn't say anything about HAWX2, the very fact that Nvidia was directing reviewers to use it should be enough to raise an eyebrow.
Click to expand...

You didn't look very far. Very well respected sites included HAWX 2 in their 580 reviews. You are advocating for reviewers running scared at the drop of a hat. What does "tainted" mean exactly? It's a new benchmark that takes advantage of nVidia's tessellation hardware - of course they will promote it! The only "taint" was AMD bitching about it because Evergreen can't handle those tessellation levels as well as Fermi.

http://techreport.com/articles.x/19934/8
http://www.hardware.fr/articles/806-17/dossier-nvidia-geforce-gtx-580-sli.html
 
It's also worth noting that the one title where Crossfire fell down hard due to a lacking driver profile is actually perfectly playable on a single 6850. In has 45 >min< FPS.

http://www.hardwareheaven.com/revie...70-and-6850-graphics-card-review-f1-2010.html

You could disable Crossfire or turn off CAT AI and get single 6870 performance in that title. That's the Achilles heel of MGPU; driver profiles. Doesn't matter here and it would suck if the profile was missing or broken for a game like Metro.

Honestly, there aren't a lot of titles that aren't playable on a single overclocked 6850 or 460 @ 1920x. I'd say save 300usd and turn a few things down in metro, or run DX9 in that game. If you want to see an interesting thread that shows DX9 vs 11 in metro, try below.

http://www.overclock.net/pc-games/690645-metro-2033-dx9-vs-dx11-without.html
 
Last edited:
gamingarena23 said:
Heh again they OC both 580GTX to the max and 6870Xfire but not the 5970 they left it at 720/1000mhz? which actually is the only card from the shootout that can be overclocked easy 20% by default and up to 30% or more...
Click to expand...
yeah wtf, the review is pointless the 5970 would rip out the other setups' asshole if it was even put to it's "rated speed" :D
OC'd it'd be a massacre for anything but perhaps an nvidia optimized game, and even then it'd be close
 
Hey guys, I wrote the article and just wanted to address a couple of points.

5970 OC vs Not OC: The reason for not overclocking the 5970 was that when checking prices on the day of the article 5970 OC versions were rare and where they were available, overpriced. The 580/6870 are just starting to see OC versions and we were focusing on OC which were available to buy in the price range. If you want an article with 5970 OC vs 6870 OC vs 580 OC then it would probably be a MAX OC article with all running at their top overclock. Thats a completely different article to this one which looks at retail level OC/Price.

For HAWX 2, i couldnt care less what AMD and NVIDIA say. Will the game perform at around the same level? Yes. Will we use the actual game in our reviews, just as we do most big new titles? Yes. So whats the problem with taking an early look at performance? Nothing (unless you are a fanboy)

For the record, when we met with NVIDIA at the 580 meeting they used HAWX2 as an example for tessellation, were very open about AMDs opinions on it and put no pressure on any journalist one way or the other on what they should/shouldnt include.
 
Mr. Bennet says "real world gaming" Test the shit like you play in a real world enviroment with the drivers us slugs use or GTFO. Simple.
 
Veridian3 said:
Hey guys, I wrote the article and just wanted to address a couple of points.

5970 OC vs Not OC: The reason for not overclocking the 5970 was that when checking prices on the day of the article 5970 OC versions were rare and where they were available, overpriced. The 580/6870 are just starting to see OC versions and we were focusing on OC which were available to buy in the price range. If you want an article with 5970 OC vs 6870 OC vs 580 OC then it would probably be a MAX OC article with all running at their top overclock. Thats a completely different article to this one which looks at retail level OC/Price.

For HAWX 2, i couldnt care less what AMD and NVIDIA say. Will the game perform at around the same level? Yes. Will we use the actual game in our reviews, just as we do most big new titles? Yes. So whats the problem with taking an early look at performance? Nothing (unless you are a fanboy)

For the record, when we met with NVIDIA at the 580 meeting they used HAWX2 as an example for tessellation, were very open about AMDs opinions on it and put no pressure on any journalist one way or the other on what they should/shouldnt include.
Click to expand...

What you saying there?

You don't need to buy actual 5970OC version you could of just OC the one you reviewed they are actually made to work at 850/1200 speed just sliding the slider in CCC..
ATI even supply's overvoltage tool for 5970 which you rarely need on them anyways..
Everybody knows that...
 
gamingarena23 said:
What you saying there?

You don't need to buy actual 5970OC version you could of just OC the one you reviewed they are actually made to work at 850/1200 speed just sliding the slider in CCC..
ATI even supply's overvoltage tool for 5970 which you rarely need on them anyways..
Everybody knows that...
Click to expand...

Of course I know that however the point of the article was to look at the 580 and 6870CF at retail OC and the price point they offer (which happens to be the same when it comes to HIS, Palit/POV and the stock 5970). If you want to move out of the price point (and cooler used) then sure, your point is valid but this was comparing the products at the same price which happened to work well when 2 were OC models and one wasnt. No ulterior motive, just a simple comparison which, for the majority of consumers, is interesting.

If you start OC the 5970 which speed do you go for? Which price? Which model? Like i said, different article. (Which no doubt people would complain about because we didnt use the ASUS ARES or XFX 5970 LE BE)
 
p3sty said:
Mr. Bennet says "real world gaming" Test the shit like you play in a real world enviroment with the drivers us slugs use or GTFO. Simple.
Click to expand...

Ah yes, like 99% of our tests for past however many years :rolleyes: It's just one test which I thought was interesting. Next week it will move to the actual game/gameplay (when HAWX2 is available to buy).
 
Veridian3 said:
Thank-you. Clearly you have excellent taste. :D
Click to expand...

I also appreciate using new game titles for your review. Here at [H] Kyle does it and more reviewers need to do the same also. I'm tired of seeing L4D2, Far Cry 2 and Resident Evil 5 benchmarks.
Once I see those games in a review, I yawn and close the tab. More DX11 titles need to be compared to show these cards true performance.
 
My 5970 is rock-stable 24/7 at 1000/1200. That's 38% OC on the 2 cores... On stock voltage, I could do 825/1200 easily. And this with a plain vanilla Asus 5970.

Why not at least try to OC it on stock voltage? Would have take 1 minute to do!

And 2560X1600 anyone? :confused:
 
WorldExclusive said:
I also appreciate using new game titles for your review. Here at [H] Kyle does it and more reviewers need to do the same also. I'm tired of seeing L4D2, Far Cry 2 and Resident Evil 5 benchmarks.
Once I see those games in a review, I yawn and close the tab. More DX11 titles need to be compared to show these cards true performance.
Click to expand...

Yeah, its down to two things really. 1) Lazyness. 2) People who dont move to new games can recycle their results. With us we retest every product for every article with the latest drivers/patches.

Drives me mad when I see articles which use results from months ago, mixed with results from the last week.
 
LEVESQUE said:
And 2560X1600 anyone? :confused:
Click to expand...

We went through a period of a couple of years using that but have moved to 5760x1080 now (see the full 580 review as an example). 2560x1600 is far less important than it was a year or so ago and given the number of people who use it, I removed those tests from our articles.

EDIT: I previously used 2560x1600 as my main display and very much preferred it to 2x1920x???? which other people use so was in the same position as you. I can say though that I have been completely won over by 3x 1920x1080, if you get the chance you should give it a shot.
 
Veridian3 said:
Hey guys, I wrote the article and just wanted to address a couple of points.

5970 OC vs Not OC: The reason for not overclocking the 5970 was that when checking prices on the day of the article 5970 OC versions were rare and where they were available, overpriced. The 580/6870 are just starting to see OC versions and we were focusing on OC which were available to buy in the price range. If you want an article with 5970 OC vs 6870 OC vs 580 OC then it would probably be a MAX OC article with all running at their top overclock. Thats a completely different article to this one which looks at retail level OC/Price.
Click to expand...

Thank you. :)
I had to figure out the purpose of not OC'ing the 5970. Your review haven't explained why you overclocked all the cards but one card. All it said was the rule you went by:

That comparison is the GTX 580, overclocked to the fastest speed we have seen advertised by a manufacturer on standard air cooling against the 6870 in Crossfire with the same highest overclock rule
Click to expand...

When you write that, people expect you to use the same conditions with all the cards. You should change it to "whatever card you found available at the moment when you wrote the review" or something like that. As it stands, I find it misleading. :)
 
Tamlin_WSGF said:
Thank you. :)
I had to figure out the purpose of not OC'ing the 5970. Your review haven't explained why you overclocked all the cards but one card. All it said was the rule you went by:



When you write that, people expect you to use the same conditions with all the cards. You should change it to "whatever card you found available at the moment when you wrote the review" or something like that. As it stands, I find it misleading. :)
Click to expand...

OK, will look to tweak the wording later today when updating the F1 results (currently retesting with 10.10e which was released a few hours ago).
 
You must log in or register to reply here.
Back
Top