6870 VS GTX460 Die sizes and price/performance

[Arkaign]

^^ So another site running FRAPS during games contradicts [H] as well? Color me shocked

The 5850 is a premium card, and if I was offered both for free, I'd take the 5850. That said, I think it's a consensus that includes my most trusted source (AT) that the 5850 is in no way shape or form dominantly faster than GTX460 as they show.

If it weren't so damned overwhelming it might make more sense, but to be that far off of everyone else's testing just makes it seem like they botched the test. I still won't claim bias, but I might call them a little wonky.

 

[Larries]

That's interesting, and actually by far the worst I've seen a GTX460 in testing vs. 5850. I'm not gonna call bias, but compare to Anand test :

http://www.anandtech.com/show/3809/nvidias-geforce-gtx-460-the-200-king/18

Unoverclocked, stock 1GB GTX460 (not overpriced Asus TOP card) is very close to stock 5850 performance. Once overclocked, it's solidly between HD5850 and HD5870 stock performance. Yes, you can also overclock 5850s as well, and probably beyond the reasonable range for a GTX460 to match, but the price difference is definitely there.

The article brings up a kind of apples/oranges comparison too, by mentioning rebates on one card but not the other. The 5850 is a great card, but it's soon EOL and it's still a noticeable amount more expensive unless someone flat out overpays for the 460.

Actually, there is only one game common between Hardcop's and Anandtech's review (Battlefield BC2), and Anandtech's review shows that at clock and 1920x1200 max quality, 5850 at clock is 52.6 fps vs 460 at clock of 44.7 fps; while Hardcop is 50 fps for 5858 and 40.3 fps for Asus 460... which is pretty consistent to me (in both case, the fps is very close).

Also, does the Asus 460 come with mail in rebate?

Because hardcop is, afterall, reviewing the Asus ENGTX460 TOP 1GB card, NOT a generic Nvidia 460 card.

Of course, you can always argue that Hardcop is listing a powercolor price vs an Asus price.

But in the end, some vendor's "top" or "super overclocked" model 460 does seem a bit overpriced.

 

[Elixer]

Some like to claim that AMD is going to run Nvidia out with a price war because they get better performance per mm. Some like to claim that Nvidia is in the drivers seat because they use that extra silicon for CUDA and HPC functions. There really isn't a right or wrong, just different. Though I would tend to agree, AMD can probably more easily make money in the video card world. Nvidia is raking in money in the HPC and professional area. So, meh.

They are also both bleeding red ink, so they need to sell them to make a profit, or they will die.

People just don't understand how a business works, so they run for congress instead, and spend other people's money, and when there isn't enough of that, they print more. D:

 

[blastingcap]

Either Anand or [H] screwed the pooch, and you can guess whom I trust more.

Read this to see why you should read both AT and [H]: http://enthusiast.hardocp.com/article/2008/02/11/benchmarking_benchmarks/6

 

[RussianSensation]

Read this to see why you should read both AT and [H]: http://enthusiast.hardocp.com/article/2008/02/11/benchmarking_benchmarks/6

The bottom line is:

1) Review a wide variety of websites (Anandtech, Xbitlabs, HotHardware, HardOCP, Tom'sHardware, Techspot, Legionhardware, Bit-tech.net, HardwareCanucks).

2) Compare the benches based on the resolution YOU game at.

3) Compare the games YOU want to play/are currently playing.

Some games run better on certain brands. For example, at 2560x1600, a GTX460 is worse than an HD5850. HD5850 will also be better in BF:BC2 for example. However, if I you happen to love I dunno say Civilization 5, Cities XL 2011 (random lol), then GTX460 is better.

The same is true for last generation. We generally consider HD4890 ~ GTX275. However HD4890 destroys a GTX275 in Civilization 5 but GTX275 is significantly faster in Medal of Honor.

It just depends on YOUR GAMES

 

[blastingcap]

The bottom line is:

1) Review a wide variety of websites (Anandtech, Xbitlabs, HotHardware, HardOCP, Tom'sHardware, Techspot, Legionhardware, Bit-tech.net, HardwareCanucks).

Nope. Many sites are just clones of each other because all they do is follow NV's reviewer's guide. What value is there in that? Look how many sites benched an outdated game nobody plays anymore and whose engine is not used by any game anyone plays: Far Cry 2. Why? Because that game has probably the largest pro-NV skew of any game out there, so NV made sure to put it in their reviewer's guide to try to skew average results. (Remember when Fermi was not yet released, NV had some teaser benches for only two things: Unigine (which NV helped develop, hence the ridiculous extreme tessellation in the updated version of that benchmark) and FC2. Coincidence? I think not.

Anandtech
Bit-Tech
HardOCP

are the only sites who didn't bench Far Cry 2 when reviewing the GTS 450. That speaks to their integrity. That's not to say that other sites that did bench FC2 are biased or anything, just that they have not proven themselves to have the stones to stand up to NV. (Maybe everyone else buckled in order to stay in NV's good graces, else they might not receive a card next time.) AT, Bit-Tech, and [H] are definitely top tier. The jury is out on everyone else.

2) Compare the benches based on the resolution YOU game at.

3) Compare the games YOU want to play/are currently playing.

The rest of this I agree with.

 

[Silverforce11]

Yes, different games favor different GPUs, so the numbers can be skewed towards one or the other depending on which games you use in your benchmarks.

If we cant use game performance as a metric to say how fast a GPU is in relation to each other, what can we use?

 

[blastingcap]

Yes, different games favor different GPUs, so the numbers can be skewed towards one or the other depending on which games you use in your benchmarks.

If we cant use game performance as a metric to say how fast a GPU is in relation to each other, what can we use?

I think AT's solution is reasonable: bench a mixture of relevant games and try not to repeat the same game engine too many times. Relevant meaning popular games that people still play and were probably rated pretty well on metacritic, with an eye towards the more stressful games, since the less-stressful games will run fine on weak cards anyway.

HardOCP's solution is good in theory, but the big drawback is that they don't bench enough games. They try to pick some relevant ones, but if they bench just four games, so if even one of them is an outlier, that can complicate things.

Bit-Tech is like a poor man's AT since they tend to bench fewer games, but at least they are independent.

 

[RussianSensation]

At the same time, the reader should be astute enough to understand that some games run faster on a particular architecture. For example, Far Cry 2 and Civ5 run way faster on NV hardware too (these are not "biased games", just benefit from NV's architecture more). This doesn't mean that reviews which included either of those games are biased towards NV. If so, are reviewers supposed to leave out Lost Planet 2, Metro 2033 and STALKER: CoP (sunshafts) where AMD gets a beating and a half???

I mean come on. I understand you are frustrated that there is no game where AMD is as much ahead of NV as NV is ahead of AMD like in Civ5, FC2, Metro 2033, etc. It's ok man, HD6000 will fix that and then you'll start hearing how the games in reviews are biased because they favour AMD cards!!

With your logic, Xbitlabs was purely biased by including Call of Duty 2 w/AA or FEAR w/AA in its X1900XTX review because 7800 GTX 512mb got absolutely hammered in those games. Similarly, including Chronicles of Riddick was biased in their 2005 Roundup because it "favoured" NV???

I'll give you the benefit of the doubt of not knowing that some of these websites are "ancient" in "Internet age" since you are a new member But trust me, a lot of these websites have been a round for a long, long time. If you review Xbitlabs stuff, the faster generation always wins in their benchmarks (like 9800Pro > 5900 series, X850XT > 6800 U or X1900XT series > 7800/7900, 8800 series > HD3800 series by them).

Many years ago, Rollo would claim that Xbitlabs is ATI biased since NV's cards would always lose in the most shader intensive games hehe. If you review more recent games, however, you'll see that X1900 XT series demolishes the 7900 series. Xbitlabs and Anandtech were some of the few websites from that era that foresaw the key strategic advantage of a major trend of ATI moving towards 3:1 shaders:textures architecture in anticipation of games becoming more shader intensive in the years to come.

Right now, Xbitlabs is foreshadowing that GF100 is the faster architecture for DX11 games. That's not biased.

 

[blastingcap]

Don't put words in my mouth RS, I am not "frustrated that AMD runs FC2 poorly." I'm frustrated at the sheep who follow review guides and review games that nobody cares about anymore, if they ever did, out of fear or whatever it is.

Civ V is a popular modern game. Metro 2033 is okay on metacritic, though I don't know if it's actually popular. It's stressful on hardware so there is a case for benching it, too. Ditto with STALKER CoP. Benching any of these would be legit.

Lost Planet 2 has bad reviews on metacritic and I wouldn't mind seeing it not benched... why bench crappy games nobody plays?

Far Cry 2--do you know anyone who still plays it? Do you know any good, popular game that uses its engine?

P.S. I thought I was clear but maybe not. To be ABSOLUTELY clear: just because a reviewer benched FC2 for a GTS 450 doesn't necessarily mean it's biased. (E.g., it could be that the reviewer is slow to update its battery of games tested, or some other benign reason why FC2 is still tested.) But if a reviewer did NOT bench FC2 for its GTS 450, that is evidence of the reviewer's integrity.

At the same time, the reader should be astute enough to understand that Far Cry 2 runs faster on NV hardware just like Civilization 5 runs way faster on NV hardware too. Doesn't mean reviews that included either of those games are biased towards NV. Then are reviewers supposed to leave out Lost Planet 2, Metro 2033 and STALKER: CoP (sunshafts) where AMD gets a crazy beating??

I mean come on. I understand you are frustrated that AMD runs FC2 poorly.

With your logic, Xbitlabs was purely biased by including Call of Duty 2 in its X1900XTX review because 7800 GTX 512mb got absolutely hammered.

I'll give you the benefit of the doubt of not knowing that some of these websites are "ancient" in Internet age since you are a new member But trust me, a lot of these websites have been a round for a long long time. If you review Xbitlabs stuff, the faster generation always wins in their benchmarks (like X1900XT series > 7800/7900 series by them).

Many years ago, Rollo would claim that Xbitlabs is ATI biased since NV cards would always lose their in the most shader intensive games hehe.

 

[sandorski]

ATI is selling 90% of the DX11 market, and they can't turn a profit. It isn't that they dont want to lower prices, they really can't. Stockholders come before customers.

They could start a Price War if they wanted to. It just doesn't make any sense for them to do so though.

 

[RussianSensation]

Lost Planet 2 has bad reviews on metacritic and I wouldn't mind seeing it not benched... why bench crappy games nobody plays?

Far Cry 2--do you know anyone who still plays it? Do you know any good, popular game that uses its engine?

LP2 is another DX11 game. Since there are so few DX11 games, it helps us gauge which cards will be better for future DX11 games.

BFG and I still play FC2. It's a dated bench, and should probably be replaced by Call of Duty: Black Ops once it ships. Doesn't mean it's biased (just exclude it in your own mind if u don't play a particular game!!)

 

[blastingcap]

LP2 is another DX11 game. Since there are so few DX11 games, it helps us gauge which cards will be better for future DX11 games.

BFG and I still play FC2. It's a dated bench, and should probably be replaced by Call of Duty: Black Ops once it ships. Doesn't mean it's biased (just exclude it in your own mind if u don't play a particular game!!)

I don't think I've ever heard anyone else admit to still playing FC2 but okay, I guess it has more than zero people who still play. Still unpopular though.

There are enough DX11 games out there that I don't think it's a big deal if we leave out a crappily-rated one. Back when there were hardly any, AvP got benched a lot. Now? Not as much and rightly so. Why? Because it got mediocre reviews and there are other options for benching DX11 now. No need to use mediocre stuff like AvP or LP2.

I think it's misleading if a site only benches 5-6 games and one or more of them is some old game hardly anyone plays anymore. If it's out of 5-6 games, a lone outlier can skew conclusions. Although I can and do dismiss games I'd never play, it does not change the fact that the review site is misleading by including a strong outlier in a small sample of games.

I get your point that the more info the better, but in a time-constrained situation in which I can't read 20 reviews, I start with AT, then read [H], and if I am still curious for some reason, I will read others (e.g., TPU because their charts make it easy to see which cards scale better as you go up in resolution). But I always start here and at [H]. I believe that other people don't have time to read 20 reviews either and would recommend to them to start at AT and [H] as well.

 

[Dark Shroud]

The only other site that even comes to my thoughts besides AT & [H] for reviews is Tom's Hardware. And even then I'm not too big into them anymore.

 

[Will Robinson]

Anandtech
Bit-Tech
HardOCP

are the only sites who didn't bench Far Cry 2 when reviewing the GTS 450. That speaks to their integrity. That's not to say that other sites that did bench FC2 are biased or anything, just that they have not proven themselves to have the stones to stand up to NV. (Maybe everyone else buckled in order to stay in NV's good graces, else they might not receive a card next time.) AT, Bit-Tech, and [H] are definitely top tier. The jury is out on everyone else.

The jury also recognizes The Tech Report as a top tier,unbiased review site that also(FYI) didn't test Far Cry2 during their GTS450 review.

 

[blastingcap]

The jury also recognizes The Tech Report as a top tier,unbiased review site that also(FYI) didn't test Far Cry2 during their GTS450 review.

I don't consider TechReport to be a serious review site, especially after their fouled-up SLI vs CF fiasco where they tested only at one setting and resolution for some games, plus they used oc'd GTX 480s against stock 5870s without even bothering to downclock them. (Their explanation was ridiculous. Downclock to stock or manually oc the 5870s, come ON.) Basically, I have zero respect for Damage as a reviewer.

That said, I don't think TechReport is NV biased (if they were, they'd definitely have used FC2), just incompetent. The amateurs at TR should leave GPU reviews to pros.

 

[Dark Shroud]

Wasn't it TechReport that also used out dated AMD drivers for on of their reviews as well?

Edit: All I remember is one of the sites that wasn't AT or [H] used outdated drivers for testing at a launch for FERMI. I'm not sure if it was Tech Report or not.

 

[Will Robinson]

You couldn't be more wrong about A.The Tech Report
or
B. Scott Wasson

You must have very limited/only recent knowledge of the tech site scene.

 

[blastingcap]

You couldn't be more wrong about A.The Tech Report
or
B. Scott Wasson

You must have very limited/only recent knowledge of the tech site scene.

Read the review yourself and draw your own conclusions. It was amateurish.

...wow they took down the article. I guess they realized how terrible it was.

It *was* at: http://techreport.com/articles.x/19404/ I think.. that's what shows up on Google when I try to search for that article. Currently the link isn't working. EDIT: now the link works. Weird, maybe my ISP had a hiccup.

They did stunts like test at one setting and resolution for HALF of the games they reviewed (e.g., Just Cause 2, BFBC2, Metro 2033). ONE setting and resolution combo. I mean, come ON, do it right or just stop trying to review GPUs at all. They used a ZOTAC AMP! to compare against stock 5870s and excused themselves by saying that it was the hardware that they were sent, rather than downclock the ZOTAC to stock or manually overclock the 5870. They benched only a handful of games, at a time when Crossfire scaling was clearly broken for some of them, and they didn't acknowledge the new CF profiles after they came out, unlike HardOCP. (That doesn't excuse AMD for broken CF but TR didn't even mention it.) Furthermore, they tend to come out with GPU reviews several days after everyone else, so despite having MORE time to polish their reviews than everyone else, they come out with a worse product.

Need I go on? I'm sorry but that's just amateurish compared to other sites *cough* Anandtech *cough* that keep drivers up to date, don't do crap like benching stock vs oc without clearly labeling it as such, bench more than six games, and bench games at more than a single settings/resolution combination.

Compare what TR did to what HardOCP did, too. HardOCP doesn't bench as many games but the ones they do, they actually PLAY the games because benches don't always match up with actual gameplay fps. They took AMD to task for broken CF and got it fixed, even getting Terry Makedon to show up at the forums to soak up forum complains so AMD could better diagnose what was going wrong. And HardOCP went back and rebenched as new drivers and CF profiles came out.

Sorry but Tech Report is a second-rate GPU review site. I like their other reviews sometimes though, like for the Sansa Clip+. Maybe Scott doesn't have as many resources as AT or [H] but in that case they should either get more efficient and do it right, or stop. The fact that they are almost always late to review new GPUs (look at their review dates for GPUs at launch--they are almost always several days behind everyone else) should tell you that they are either a) understaffed and/or b) they don't place a high importance on GPU reviews, which should make you wonder why you're reading a half-hearted review there rather than a full-bore review at Anandtech or [H].

 

[Tsavo]

Tech Report definitely has some pretty lightweight reviews. They have been around forever though, but they've never gone from being a tiny website which is a reflection of the lack of quality reviews. At least their video card reviews aren't an utter joke like the crap that HardOCP pinches out.

 

[3DVagabond]

Wasn't it TechReport that also used out dated AMD drivers for on of their reviews as well?

Edit: All I remember is one of the sites that wasn't AT or [H] used outdated drivers for testing at a launch for FERMI. I'm not sure if it was Tech Report or not.

TechPowerUp (TPU) is pretty infamous for using outdated drivers. I believe that's who you are refering to.

 

[tviceman]

Civ V is a popular modern game. Metro 2033 is okay on metacritic, though I don't know if it's actually popular. It's stressful on hardware so there is a case for benching it, too. Ditto with STALKER CoP. Benching any of these would be legit.

Lost Planet 2 has bad reviews on metacritic and I wouldn't mind seeing it not benched... why bench crappy games nobody plays?

Far Cry 2--do you know anyone who still plays it? Do you know any good, popular game that uses its engine?

So should reviewers only benchmark highly popular and/or games with whored out graphics engines? If that is the case, lets just stick with Unreal Tournament 3 and Counter Strike: Source. That'll cover about 9/10 of all players either through the UT3's graphics engine or CS's player numbers.

I'll gladly take benchmarks for the less popular games if it's a good indicator of how the GPU will perform under certain tasks/ levels of stress. LP2 looks like it is a great indicator of how current cards can perform under future titles that will be utilizing tessellation. I don't see why it should not be included due to what metacritic says. I've played some really fun, great games that metacritic says I should never have wasted my time with and I've also played a few games I wished I would never have spent money on that metacritic says I should absolutely revere.

 

[bryanW1995]

Having looked over your three responses to my posts, I find myself in agreement with part of that sentiment.

The part that involves you continuing to post in this thread.

no need for personal attacks. russiansensation has owned many many cards over the years from both daamit and nvidia, plus he has shown over thousands of posts here that he doesn't need to attack other members to get his point across.

I suggest you follow the same approach.

 

[Idontcare]

Thread is long and well off-topic.

Can anyone tell me whether or not the numbers in this BSN article are considered "accurate"?

AMD's Cayman [HD 6900] die is exactly 380mm2, i.e. 43mm2 more than previous gen Cypress [HD 5850/5870/5970], while Barts is 230mm2, significantly more than 166mm2 of its predecessor [Juniper].

http://www.brightsideofnews.com/new...lands-vs-nvidia-fermi-die-sizes-compared.aspx

 

[railven]

Thread is long and well off-topic.

Can anyone tell me whether or not the numbers in this BSN article are considered "accurate"?


http://www.brightsideofnews.com/new...lands-vs-nvidia-fermi-die-sizes-compared.aspx

Wouldn't that be considered off topic too since Cayman is 69xx and this thread is about Barts 68xx vs GTX 460?

EDIT: Oh, I see the other thread was locked. Well guess the merging of threads has begun.