New GTX260 vs 4870 1GB

[BenSkywalker]

You can find extreme cases on both ends There is a currently shipping game where a 9600GT is throttling every card ATi card atm- I thought this test was more interesting honestly as it showed the 1GB 9600GT smacking the 8800/9800 512MBs around silly.

 

[nRollo]

Originally posted by: Zebo

Originally posted by: apoppin

Originally posted by: Zebo
These cards are not equal why you guys keep saying that? Sure in majority of games they are equals +-5% and very playable - but in two games that I know of -Devil may Cry and Grid - one card is playable (more than 30FPS) at max settings the other is not.

http://www.bit-tech.net/hardwa...i-radeon-hd-4870-1gb/9
http://www.anandtech.com/video/showdoc.aspx?i=3415&p=7

You are just cherry picking benches at ultra high detail/resolutions 99% of us do not use.

do you DMC play at 25x16 with 8xMSAA?
:roll:

Now Grid is Quite Playable - according to your standards - on GTX280 at 25x16 on Ultra with 4xAA
- even 33fps on a 260

and do we see the minimum FPS on either of these tests?

I'm not Cherry picking I am going to multiple review sites who test more than 5 games to get a more complete picture of these cards abilities.

Speaking of abilities - it's immaterial what I play at - both cards would satisfy for now - however if a card can't perform with certain games and high res now, how will it's ability be crippled with future games using similar coding or engines which will then effect everyone who ones that card?

It's always the case that one card will run a few games better than the other.

Hellgate London runs much better on a GTX280 than HD4870
Obviously you'd be better off with a minimum fps of 45 than 27 with the X2.

Same with the 28 vs 22 minimum on Oblivion Cyrodiil bench below.

Or what about the games you need to re-name the executable to get AA working at all with AMD, and lose any patches or optomizations?

Or what about the PhysX games?

I stand by what i said- the cards are largely equal, should come down to vendor preference and feature preference for most people.

 

[Wreckage]

Originally posted by: Zebo
These cards are not equal why you guys keep saying that?

When you look at modern games that people actually play like Crysis then you see that they are not equal. The 4870 falls behind once again.
http://www.bjorn3d.com/read.php?cID=1350&pageID=5370

Actually the most important factor Avg and Min frame rates is where the 4870 seems to often have a problem, with reviewers complaining about slow downs while gaming.

 

[apoppin]

Originally posted by: Wreckage

Originally posted by: Zebo
These cards are not equal why you guys keep saying that?

When you look at modern games that people actually play like Crysis then you see that they are not equal. The 4870 falls behind once again.
http://www.bjorn3d.com/read.php?cID=1350&pageID=5370

Actually the most important factor Avg and Min frame rates is where the 4870 seems to often have a problem, with reviewers complaining about slow downs while gaming.

the 512MB version IS somewhat limited at or above 19x12 .. that is why i do NOT recommend 4870/512 for 19x12
- i don't yet know about the 1GB version of 4870 until this Tuesday but even the 280GTX does not blow away even the 512MB 4870 in Crysis with the latest drivers.

Here are my Crysis benches at very high/19x12 - noAA/AF; e8600@4Ghz

HD4870 - 20.04/16/25
GTX280 - 21.78/19/30

now Bjorn3D only tests up to 16x10; here are my benches:

HD4870 - 24.23/19/30
GTX280 - 26.45/19/34

i don't agree with their "tuning" nonsense spiel:

Please note that on other tests of the BFG GTX-260 OCX we've seen from various sources we've seen scores lower than our reported scores. We assure you that they are entirely due to the lack of proficiency in tuning a system for graphics performance on the part of lesser mortals. When we tune a system at Bjorn3D, we really tune it and the difference in performance shows.

all i see here is fans cherry picking extreme benches to support their view that one card is "superior" to another. That said, the GTX280 offers a better playing experience with Crysis than 4870/512 and the X2 actually makes 19x12 almost playable; i cannot comment on GT260 nor on 4870/1GB ['till tuesday ]

 

[thilanliyan]

Originally posted by: Wreckage

Originally posted by: Zebo
These cards are not equal why you guys keep saying that?

When you look at modern games that people actually play like Crysis then you see that they are not equal. The 4870 falls behind once again.
http://www.bjorn3d.com/read.php?cID=1350&pageID=5370

Actually the most important factor Avg and Min frame rates is where the 4870 seems to often have a problem, with reviewers complaining about slow downs while gaming.

People play Crysis? Isn't it just a benchmark?

I'd also like to point out that the highly OCed BFG 260 you're comparing to a stock 4870 is $320 at Newegg...and is much more expensive than the Sapphire 4870 they're using, which can be had for $230...$90 (!!) less.

EDIT: The 4870 is actually $70 cheaper rather than $90 if you include the rebate on the 260.

 

[SlowSpyder]

Originally posted by: thilan29

Originally posted by: SlowSpyder
I thought this whole thing was silly, so I went and looked at the numbers... I'm suprised to see how well the the 4870x2 does compared to the GTX260/280 in SLI configs.

In CoD4 the 4870x2 and SLI 2xGTX260/280 are all limited by the CPU or something else, except at 2560 res. There the 4870x2 is faster then 3xGTX260 or 2xGTX280, but within ~1FPS, so that'd be margin of error I'd think, right?

In HL2 the 4870x2 is faster then SLI 2xGTX260/280 at all resolutions except 2560. At that res the 4870 is faster then the GTX260 SLI, but slower then the GTX280 SLI.

In Enemy Wars Quake Territory the 4870x2 is faster at all resolutions (then any card) at 1980 res or below. At 2560 the SLI 2x GTX260/280 is faster by a decent amount.

In Crysis the SLI GTX260/280 are faster then the 4870x2.

In Assasains Creed the 4870x2 is faster then the GTX260/280 in 2x or even 3x SLI.

In Race Driver GRID, again the 4870x2 is faster then the GTX260/280 in 2x or even 3x SLI.

So what are we fighting about again? I don't think you can beat the 4870x2 for ~$550.

Which sites were you looking at?

http://www.techreport.com/articles.x/15293/11

That was the link that was being used.

 

[apoppin]

Originally posted by: thilan29

Originally posted by: Wreckage

Originally posted by: Zebo
These cards are not equal why you guys keep saying that?

When you look at modern games that people actually play like Crysis then you see that they are not equal. The 4870 falls behind once again.
http://www.bjorn3d.com/read.php?cID=1350&pageID=5370

Actually the most important factor Avg and Min frame rates is where the 4870 seems to often have a problem, with reviewers complaining about slow downs while gaming.

People play Crysis? Isn't it just a benchmark?

I'd also like to point out that the highly OCed BFG 260 you're comparing to a stock 4870 is $320 at Newegg...and is much more expensive than the Sapphire 4870 they're using, which can be had for $230...$90 (!!) less.

i just got the 1GB ASUS 4870 for $269 after $20 MiR from NewEgg:

http://forums.anandtech.com/me...=2236244&enterthread=y

http://www.newegg.com/Product/...x?Item=N82E16814121277

AND .. BREAKING NEWS!!

contrary to what i was expecting, the 512MB of Vram does NOT appear to hold back the 4870X2 when it is forced to run using half its allotment by being forced into a mismatched X3 CF - 4870/512 + 4870x2

mostly there is a decent increase [considering it IS mismatched] and i would have to say drivers are probably to blame for when X3 is slower than X1 - or once where X2 is slower than 4870

However, i would not recommend using a 4870/512M in a 4x PCIe slot if you can run it in a 16x + 16x PCIe MB .. but then i was surprised that there was not a huge performance delta between them.

one more bench to go .. then i look at my data and summarize my conclusions .. and start working on 1GB 4870 in a more "ideal" CF setting. .. Tuesday night i start over

 

[Zebo]

Originally posted by: nRollo

Originally posted by: Zebo

Originally posted by: apoppin

Originally posted by: Zebo
These cards are not equal why you guys keep saying that? Sure in majority of games they are equals +-5% and very playable - but in two games that I know of -Devil may Cry and Grid - one card is playable (more than 30FPS) at max settings the other is not.

http://www.bit-tech.net/hardwa...i-radeon-hd-4870-1gb/9
http://www.anandtech.com/video/showdoc.aspx?i=3415&p=7

You are just cherry picking benches at ultra high detail/resolutions 99% of us do not use.

do you DMC play at 25x16 with 8xMSAA?
:roll:

Now Grid is Quite Playable - according to your standards - on GTX280 at 25x16 on Ultra with 4xAA
- even 33fps on a 260

and do we see the minimum FPS on either of these tests?

I'm not Cherry picking I am going to multiple review sites who test more than 5 games to get a more complete picture of these cards abilities.

Speaking of abilities - it's immaterial what I play at - both cards would satisfy for now - however if a card can't perform with certain games and high res now, how will it's ability be crippled with future games using similar coding or engines which will then effect everyone who ones that card?

It's always the case that one card will run a few games better than the other.

Hellgate London runs much better on a GTX280 than HD4870
Obviously you'd be better off with a minimum fps of 45 than 27 with the X2.

Same with the 28 vs 22 minimum on Oblivion Cyrodiil bench below.

Or what about the games you need to re-name the executable to get AA working at all with AMD, and lose any patches or optomizations?

Or what about the PhysX games?

I stand by what i said- the cards are largely equal, should come down to vendor preference and feature preference for most people.

Talk about cherry picking (and irrelevant since this tread is dealing with Core 216 vs. 4870 1GB - not X2 vs GTX 280)
But I'll run with this since you bring it up. So the GTX kills the X2 in ONE game while the X2 kills the GTX in every other game showing it to be superior card.

This is not the case with the 4870 1GB vs Core 216, they are for the most part even. The 4870 1GB does not dominate the Core 216 like the X2 does the GTX - Except for a couple games. Note the difference.

Anyway I don't cherry pick (eg: physX which nothing uses). I deal with whats out and from a pure performance perspective the 4870 is a better buy due to a couple more games placing heavily in it's favor. At $250 the core 216 would be.

Of course I would never buy this performance leader as close as it is - w/ bloated resource hoggin drivers necessitating bloated .net combined with foreign owned Al-AMD. That's cherry picking.

 

[Zebo]

Originally posted by: Wreckage

Originally posted by: Zebo
These cards are not equal why you guys keep saying that?

When you look at modern games that people actually play like Crysis then you see that they are not equal. The 4870 falls behind once again.
http://www.bjorn3d.com/read.php?cID=1350&pageID=5370

Actually the most important factor Avg and Min frame rates is where the 4870 seems to often have a problem, with reviewers complaining about slow downs while gaming.

Good find about Crysis with the 512MB card. This tread isn't dealing with that card though - see title.

 

[Wreckage]

Originally posted by: thilan29

I'd also like to point out that the highly OCed BFG 260 you're comparing to a stock 4870 is $320 at Newegg...and is much more expensive than the Sapphire 4870 they're using, which can be had for $230...$90 (!!) less.

Try $284
http://www.newegg.com/Product/...x?Item=N82E16814143156

 

[Wreckage]

Originally posted by: Zebo

Good find about Crysis with the 512MB card. This tread isn't dealing with that card though - see title.

The benchmark in the Title covered this and it was the same result.

I was pointing out an example of cherrypicking....

With so many factory overclocked 260's available, it's easily a better choice than either 4870 especially the 512 version.

 

[SolMiester]

Originally posted by: sourthings
This is the same fanboy who will try to tell you ATI has not retaken the single card performance card crown and nvidia still has it in the 280. Another review from the same site showing the 4870X2 is faster than not just a single GTX 280, but two GTX280s in SLI.

http://www.techreport.com/articles.x/15293/5

Yes that's right, the $500 X2 is faster than two GTX 280s coming in at $900. Now watch the fanboy explain how this is just not so. But still he'll insist the 260 'is far and away better than a 4870 1GB'

yes, sure it is

Eh, that is cpu bottlenecking there!! except for a couple of frames!

 

[Wreckage]

Originally posted by: SolMiester

Originally posted by: sourthings
This is the same fanboy who will try to tell you ATI has not retaken the single card performance card crown and nvidia still has it in the 280. Another review from the same site showing the 4870X2 is faster than not just a single GTX 280, but two GTX280s in SLI.

http://www.techreport.com/articles.x/15293/5

Yes that's right, the $500 X2 is faster than two GTX 280s coming in at $900. Now watch the fanboy explain how this is just not so. But still he'll insist the 260 'is far and away better than a 4870 1GB'

yes, sure it is

Eh, that is cpu bottlenecking there!! except for a couple of frames!

They even say so in his link

"Well, this is a great game, but it's pretty much CPU limited with its quality settings maxed out "

total failure by sourthings

 

[sourthings]

Originally posted by: SolMiester

Originally posted by: sourthings
This is the same fanboy who will try to tell you ATI has not retaken the single card performance card crown and nvidia still has it in the 280. Another review from the same site showing the 4870X2 is faster than not just a single GTX 280, but two GTX280s in SLI.

http://www.techreport.com/articles.x/15293/5

Yes that's right, the $500 X2 is faster than two GTX 280s coming in at $900. Now watch the fanboy explain how this is just not so. But still he'll insist the 260 'is far and away better than a 4870 1GB'

yes, sure it is

Eh, that is cpu bottlenecking there!! except for a couple of frames!

If you take the time to pres the 'next page' buttons you'll continue to see the 4870X2 beating GTX280SLI, not just in that benchmark. And both cards are being bottlenecked. Or if you take the time to look at other review sites comparing them you'll see the same.

I stand by it, and it's proven in many benchmarks, the 4870X2 is almost always faster than GTX280SLI, and costs $400 less.

GTX280SLI is a pointless video card solution unless the only game that matters to you is Crysis.

The raving fanboy in here that always feels the need to try to promote nvidia is welcome to it, if all he cares about is a brand name that's his business, and it's obvious from his posts. I care about performance, I'll support whoever performs better. At this point, the pinnacle of video card performance is the X2 and by a decent margin.

Debating a 4870 vs 260 is just foolish as any one with some sense and lack of bias can see you can't differentiate the two in real world usage. At this point imo the 4870 1GB edges the 260 out, because where it outperforms it, it's by a good margin and even beats the 280. Where they are close or the 260 has an edge, it's by a slim margin.

Unlike when you compare a 280, or 280SLI to a 4870X2, where the X2 has big gains over the 280 and slim gains over 280SLI, but also costs $400 less, which makes 280SLI a ridiculous purchase, moreso because you have to use a crappy nvidia mb to make use of it.

 

[thilanliyan]

Originally posted by: Wreckage

Originally posted by: thilan29

I'd also like to point out that the highly OCed BFG 260 you're comparing to a stock 4870 is $320 at Newegg...and is much more expensive than the Sapphire 4870 they're using, which can be had for $230...$90 (!!) less.

Try $284
http://www.newegg.com/Product/...x?Item=N82E16814143156

Actually try $300 (WITH rebate) (sorry I misstated in my original quote when I said $320).

This is the correct card:
http://www.newegg.com/Product/...x?Item=N82E16814143157

The one you linked has slower core and memory speeds than the one used in the Bjorn review you linked. So the 4870 used in that review is still $70 cheaper. Please try harder next time.

 

[Wreckage]

Originally posted by: thilan29

The one you linked has slower core and memory speeds than the one used in the Bjorn review you linked. So the 4870 used in that review is still $70 cheaper. Please try harder next time.

It's faster than the one in the TR review that stomped all over the 1GB 4870.

 

[BenSkywalker]

At this point, the pinnacle of video card performance is the X2 and by a decent margin.

Age of Conan and Grid the 4870x2 beats the 260SLI setup by a decent amount, 260 takes Crysis, ESIV and AC with the two being pretty much tied in the other two titles benched in this review. That isn't the 280 in SLI, not the 260 216 or even the OCd 260s, just the stock clocked ones. With the base 260 at $200 after MIR and $530 the 4870x2 it seems to me you can get directly comparable performance for $130 less going with 260 SLI setup. Of course, it sounds like you only really care about what vendor it comes from, not actual performance which is of course your right. I'm far more interested in performance and value.

 

[thilanliyan]

Originally posted by: Wreckage

Originally posted by: thilan29

The one you linked has slower core and memory speeds than the one used in the Bjorn review you linked. So the 4870 used in that review is still $70 cheaper. Please try harder next time.

It's faster than the one in the TR review that stomped all over the 1GB 4870.

In the AT review the 1GB 4870 doesn't get stomped on by the 216. As many have said before it depends on which site you look at. Also, $70 is a lot and doesn't justify the minimal increase you'll get in some games.

 

[Wreckage]

Originally posted by: thilan29


In the AT review the 1GB 4870 doesn't get stomped on by the 216. As many have said before it depends on which site you look at. Also, $70 is a lot and doesn't justify the minimal increase you'll get in some games.

The AT review did not use a factory overclocked 260 like we are discussing here.


Hardocp used a stock clocked card and showed that they are very close. However, there are so many factory overclocked cards available the TR review gives a better view of performance.

 

[BenSkywalker]

In the AT review the 1GB 4870 doesn't get stomped on by the 216. As many have said before it depends on which site you look at.

The difference in sites largely depends on which board they are testing. AT, as usual, tests the slowest clocked part that is available. The 650MHZ 216 is a whopping $10 more then the slowest clocked 216. I don't think many people would pay $290 for a 575MHZ 216 when the 650MHZ model is $300.

 

[apoppin]

Originally posted by: sourthings


I stand by it, and it's proven in many benchmarks, the 4870X2 is almost always faster than GTX280SLI, and costs $400 less.

GTX280SLI is a pointless video card solution unless the only game that matters to you is Crysis.

The raving fanboy in here that always feels the need to try to promote nvidia is welcome to it, if all he cares about is a brand name that's his business, and it's obvious from his posts. I care about performance, I'll support whoever performs better. At this point, the pinnacle of video card performance is the X2 and by a decent margin.

You are the only one who apparently believes his own fantasy [and i am using your benchmarks - besides Crysis, where GTX sli easily wins, there are two more games balancing Grid - where even GT260 sli beats X2; we already know there is no difference in CoD4 except at 8xMSAA @ 25x16]:

at the high resolutions you normally cherrypick, the X2 gets slaughtered by 280GTX sli
http://www.techreport.com/articles.x/15293/6
108 with 280GTX SLI to 85 X2 in HL2 at 25x16
- even GT260 SLi beats X2 - 94 to 85

http://www.techreport.com/articles.x/15293/7
112 SLI to 101 X2 in ET-Q4 at 25x16
heck, GT260 sli beats X2 again - 102 to 101 [dead heat]

http://www.xbitlabs.com/articl...tx280amp_12.html#sect1
in hellgate a *single* GTX280 beats a X2

ditto in some Oblivion maps with a X2 losing to a single GT280

now lets see something from you other than GRID; i am tired of your unsupported nonsense that "it's proven" that X2 beats 280GTX SLi

x2 is not faster than 280GTX SLI except in rare situations - like when 2900xt used to beat 8800GTX. otoh X2's performance is rather close to GT260 SLI instead - which far cheaper than a X2!!;
.. i would prefer X2 to be faster than it is, as i have one .. and i definitely prefer 280GTX to 4870/512 when you compare single gpus.



as to price, X2 is currently at $550 with very rare sales and 280GTX SLi will cost you about $760; $800 if you eschew the rebates
.. just over $200 difference, not $400 as you so wildly exaggerate
:roll:

and GT260+ SLi - which is damn close to an X2 in performance, is $100 cheaper - if you have a SLi MB

 

[thilanliyan]

Originally posted by: BenSkywalker

In the AT review the 1GB 4870 doesn't get stomped on by the 216. As many have said before it depends on which site you look at.

The difference in sites largely depends on which board they are testing. AT, as usual, tests the slowest clocked part that is available. The 650MHZ 216 is a whopping $10 more then the slowest clocked 216. I don't think many people would pay $290 for a 575MHZ 216 when the 650MHZ model is $300.

I think it's a $25 difference:
http://www.newegg.com/Product/...x?Item=N82E16814143155

Also, I don't think the difference is attributed purely to the model of the card they use. Take a look at the 2 original 4870 reviews from AT and TR:
http://techreport.com/articles.x/14990/9
http://www.anandtech.com/video/showdoc.aspx?i=3341&p=14

Take CoD as an example...notice at 1920, the 4870 is ahead of the 280 and 260 in the TR review but in the AT review it is behind the 280 and ahead of the 260. Now at that point they're using reference cards so why the difference in results? As I said it depends on which site you look at. I'm pretty certain not all sites have the same testing methodology in all games.

Originally posted by: Wreckage
Hardocp used a stock clocked card and showed that they are very close. However, there are so many factory overclocked cards available the TR review gives a better view of performance.

See response to BenSkyWalker. My point was it depends on which site you look at.

 

[BenSkywalker]

I think it's a $25 difference:

Nope, that isn't a stock clocked board. You have to spend more to get one as slow as what AT uses

Take CoD as an example...notice at 1920, the 4870 is ahead of the 280 and 260 in the TR review but in the AT review it is behind the 280 and ahead of the 260. Now at that point they're using reference cards so why the difference in results? As I said it depends on which site you look at. I'm pretty certain not all sites have the same testing methodology in all games.

TR is using AF, AT isn't. Nothing tricky there

 

[thilanliyan]

Originally posted by: BenSkywalker

I think it's a $25 difference:

Nope, that isn't a stock clocked board. You have to spend more to get one as slow as what AT uses

Okay $20 difference then:
http://www.newegg.com/Product/...x?Item=N82E16814130398

Take CoD as an example...notice at 1920, the 4870 is ahead of the 280 and 260 in the TR review but in the AT review it is behind the 280 and ahead of the 260. Now at that point they're using reference cards so why the difference in results? As I said it depends on which site you look at. I'm pretty certain not all sites have the same testing methodology in all games.

TR is using AF, AT isn't. Nothing tricky there

So you're saying the 4870 WITH AF is faster than the 280 without (EDIT: meant both with AF or both without AF) (at least in CoD)?

Also, you know for sure that the testing methodology is the same between the 2 sites?

And you're sure AT doesn't use AF? I know it doesn't say it in the chart but AFAIK that's a setting they use for their vid card reviews isn't it?

 

[Wreckage]

Originally posted by: thilan29
Now at that point they're using reference cards so why the difference in results?

Are they using the same settings, same CPU, same motherboard, same drivers......

There's more to benchmarking a video card than just the GPU and the game.