I personally think that its perfectly resonable for TR to compare cards that are $300 each, but what they needed to do was to point out the the 260 was heavily overclocked somwhere and multiple times, cause a HD4870 heavily overclocked would have wooped in this article. End Point.

It appears to me the many of you are arguing the wrong points.

"Overclocked" - Whether factory performed or user performed, is just that, Overclocked. Period. End-of-Story.

Before the other side says, "YAY someone gets my point!!!" You point is well, pointless. This review is between 2 cards of similar price points ~$299. Clocking doesn't mean anything. Comparing it to a different card will also do nothing. Choosing to compare these 2 cards is a price for power comparison, and nothing more. Stop your whining about, "Why didn't he compare similarly clocked cards!!! BLARGH!!!"

Enough, you are all very tiring to listen too, while a majority are completely ignoring this fact. (Some of you thankfully got it)

I appreciate the high quality of TR reviews, and this one was no exception.

One point of curiosity, though:

Why was a reference 4870 1GB compared to a heavily-overclocked 260v2? I can see it from a "best performance at a given price point" perspective, but doesn't this distort the basic premise of the article? It would seem to skew results by how quick secondary manufacturers are to market with aggressively-overclocked cards. When the 4870 1GB OC (why wasn't the 1GB offering called 4875?) cards come to market, a consumer wouldn't be able to use this review to compare offerings.

Perhaps include a reference card for each article flagship product in future reviews?

My 8800 GTS 640MB 112SP was unable to deal with S.T.A.L.K.E.R. Clear Sky in DX 10 mode @ 1920x1200 so I was thinking about upgrading prior to this article. My local dealer had one Core 216 in stock so I picked it up. The Core 216 provides playable frame rates in DX 10 mode @ 1920x1200 with all the settings maxed.

I tweaked it up to AMP²! levels using the supplied OC'ing utility.

These cards are too close in performance in most cases, to actually argue over which one is better, it doesn't make sense really. I'm not going to kill someone over a 5 fps difference
It comes down to which company you prefer, Nvidia or ATI/AMD

The "reference" 4870 only has 512mb of ram, so TR should take some pliers and remove the "over memory" so that these test are fair.....

I'll be labelled a fanboy for saying this but when all other factors are equal I just seem to prefer nvidia cards.

It's hard to qualify but I'll try to give you my opinion:

The Catalyst Control Centre upsets me - slow to load, sometimes the services would crash, and the whole driver installation still seems messier with a Radeon. The skinning of the interface lacks the only option I'd want (OS integration) and I'm loathe to say that pictures of Ruby and adverts in my drivers page are welcome.

My particular sample group of graphics cards also favours nVidia in terms of reliability, but perhaps I've just been lucky with nVidia and unlucky with ATi. I'd like to say that my sample size is bigger than most as I get through maybe 100 graphics cards a year and have experience with about 500 machines which get replaced at a rate of about a dozen per month. I've had more application and game issues with ATi/AMD as a whole but I know AMD have improved their game a lot in the past 12 months.

I'm sure it happens with nVidia cards too, but some ATi board partners (eg PoweColor and Sapphire) have caused me trouble in the past with custom BIOSes. Sure, getting the right drivers to work isn't rocket science but it's the little things like knowing that Windows can identify your board correctly that make a difference, all other things being equal. I'm guessing that nVidia was just historically stricter with it's partners which is why I haven't come across it on nVidia boards to date.

Thanks for this comparison, I've been looking into getting either of these two cards to replace my 8800 GTX dino. It does look like the GTX 260 will be my next purchase even though i am very egar to move back to the ATI front just to be able to have that bit of future proofing via crossfire. That being said though, i've never taken the oppertunity or even considered adding a second card to my system when a single card's performance starts to become an issue in the latest games, so I guess its what some would call marketing hype.

The performance of the GTX 260(+) is really good but... Thats a factory overclocked card and I live in Britian, so the price of that is not going to be the £200 or £220 of the HD 4870 1GB and will be more like £250 just plainly because the card is pre overclocked. Last but least there is also the pain of finding a card that actually has the 216 shaders. So maybe Nvidia will be loosing out just plainly because of very bad europien marketing and exploitation, that seems to be occuring as a result of their unclear naming and pricing scheme over here.

Ok, folks. One last shot here for several of you who keep posting incessantly, being rude to one another, and flame-spamming the discussion.

l33t-g4m3r, Meadows, pmonti80, and sammorris, we've heard from you enough to know where you stand. You need to end your participation here so other voices can get a word in edgewise. If you do not, you will be banned.

Others who are rude or who start going crazy with posts will be banned, as well.

Also, I am going to go nuke some posts where people have violated our rule against personal attacks on one another. If your post has gone missing, that's why.

Well there's plenty of flames going on at Anand over this comparison test and its results...yikes!
Personally I think TR tests are"good enough" to show if one card or another is clearly superior.I don't think that was the case here.
With regard to the use of overclocked cards:Why not overclock the ATI card by a similar percentage as the test Nvidia card is above factory specs?
If the OC version has say,10% higher engine and memory speeds than the standard specs surely it wouldn't be hard to overclock the ATI card by a similar amount?
That would seem to take care of any advantage of an overclocked part against a stock one.
Also TR's poll of the most common resolution used by its members came out at 1680x1050....the 4870 won all but one game test including an absolute thrashing in GRID.
Given that,I'd say the 4870 is the faster card.

I initially had the same problem understanding results in the last video card review for the 4670 when the 9600GT was more or less the same as the 9800GT. Now I know the 8800GT-9800GT isn't all that much faster but the fact they were tied in many benchmarks made me wonder...sure enough turns out the 9600GT was an overclocked version. It didn't get too much attention because those weren't the featured cards in that article though.

Also, going back through TR reviews, the original 9600GT review was on an oc'd Palit card but stock clocks were also included so including both is not something that hasn't been done before.

Damage, we know that which particular card you review may be beyond your control but it's clear you knew to do 'stock' clocks before, don't go changing on us now! :)
=====
One more note about user user oc'ing wrt ATi cards....they include overclocking in the drivers for gosh sakes. Sure it's not 'factory oc'd' (the silliest sham on reference cards ever) but it's right. there. in the drivers....come on people :p

You should seriously stop comparing pre overclocked cards from Nvidia (or ATI) to non overclocked normal cards ATI (or Nvidia).

Edit: Upon reading this again I realize that I've been a bit rude. I've posted on #70 what I should have really posted here.

When performance results at anything but the highest of resolutions are so very close it's getting to minor things that go into choosing a card in this range. I don't like crossfire or SLi, so I'm going with either the 4870 (1GB because it's pretty close to the same price, and is just that little bit better) and the GTX 260 216. I ended up going with the 260 because I was able to find one cheaper, and shipped for free. I had trouble finding a 4870 I liked in stock for cheap. All the cheap ones (at the time) were out of stock. I'm betting most people would be happy with either option really. No game out right now pushes these cards until you get to the maximum resolution they are capable of. Unless you play at the highest res. of the card you won't even be able to see the difference...I like that there is options for anyone who is a fan of either company to the point that both are happy. Guys like me who have no favorite company are happy to have a lot of options! Everybody can be happy!

For what it's worth, Charlie at the Inq says nVidia is saving the 270 and 290 models names for the die-shrunk versions of the current chips.

I wonder how the GPU temperatures are being measured. The consistent difference of ATI GPUs in being hotter than Nvidia GPUs might as well be only a difference of how (or where) its measured by each maker.

This difference gives ATI quite a disadvantage and Nvidia an advantage. I personally would in doubt always choose a card with a low temperature because it reflects its power consumption and the MTBF, too (MTBF = mean time between failures).

Damage (or anyone), we have a situation. Both gl6350 and bx450 have started spamming after they've done their "private message" business on the forum side, their messages are similar on both parts of the site.

Time to nuke.

No surprise really as the standard 260 was beating the standard 4870.

Given the lack of features on the 4870 (not to mention the heat issues) it was hard to recommend anyways.

It's unfair to compare a 17% overclocked card to a stock clocked 4870.You could've used a DIAMOND xoc 4870 or a powercolor pcx one.

Penn and Teller FTW!
(doh! misplaced post = response to #102)

"The handsomely stickered card you see above is the GeForce GTX 260 AMP²! Edition from Zotac. (I would like to thank Zotac for making me type AMP² !, since I could use the exercise.)"

lol, I just spit water onto my monitor, literally.

You can get the BFG GTX260 for even less than $300 (AR)
http://www.newegg.com/Product/Product.aspx?Item=N82E16814143148

Clocked even higher than the Zotac (oh noes!).

What a hilarious discussion. These cards are more or less equivalent in performance. How can some people get so heated up when you can pretty much guarantee that if you blind-tested them with these two cards (or any under/over-clocked variant for that matter) in their own systems at the resolutions they play at, they wouldn't be able to tell the difference? Just choose the one you like best or is cheapest where you live, and save the angst for something important.

In any case, value-wise 4850 Crossfire pwns them both, anyway ;-)

Note to Scott: add 'testing a factory oc'd card with no standard speed card in the same test' to the list of topics that are good for lots of comments and page views :)

you put 9600 gso as 512, isent it only 384?

I realize there is a lot of complaining going on here but I've got to say that it would have been nice to see 4850x2 in there which also goes for about $300. I think that combo is probably the clear winner at that price. (But with the obvious disadvantage of requiring crossfire.)

1GiB of video memory = pointless for the most part (GRID seems to require more than 512MiB of VRAM at higher resolutions+AA).

The GPU will be obsolete long before games "required" that amount for optimal performance.

These days it is silly to spend more than $299 on a video card when 4850, 9800GTX+, 4670, 9500 and all yield sufficient performance. That is unless you are feeding a 30" monitor. If that is the case then the $249-299 category is quite sufficient.

I find flame wars on factory-overclocked products to be very silly. Here is how I envision the people on both ends.
http://img.photobucket.com/albums/v248/Krogoth255/RAGED.jpg

FYI, the normal 260 Mk. II performs like 5-10% better than the 260. It still manages to tailgate the 280 in most benches and outrun the 4870.

Ok fair enough, the argument that the two products tested here are at the same price point is a very valid reason for their comparison and I accept that wholeheartadly.

However, to say that Diamond wouldn't send you a faster sample whereas nVidia did is playing into PR departments' hands a little too much. I'd wager that I could find overclocked versions of both cards here for a similar street price, so if consumer value is what TR is trying to highlight then they should really endeavour to find the best card for $300 (or whatever), rather than the best one that someone would actually send them. Whether that is a failing of AMD or TR though I'm not sure.

This is the last I'll say on the matter!

Since nobody is being honest about this snafu, let me explain it exactly how I see it.

step1: sell limited supply overclocked card at newegg for 299$
never mind 299$ is the price point of the regular cards, and the other OC'd cards are much higher, or that there is a 75/100 mhz difference.

step2: give OC'd card to site for review. enjoy higher numbers.

step3: run out of OC'd cards supply 1 week after review, or raise price through the roof.

step4: everybody buys normal 260's at 299$ thinking they perform like TR's review.

step5: big paycheck for everyone involved.

I think the point most people are missing is.....

Battle royale at 300 bucks

Even though it's in a larger font right in the title.