[Monstru]

Finnaly, GTX 275 vs HD4890, 2 x GTX 275 vs 2 x 4890 and 3 x GTX 275 vs 3 x 4890, in 1680x1050, 1920x1200 and 2560x1600.

GTX 275 vs HD 4890 @ lab501 (with Google translator)

Conclusion? Buy whichever you like, but no more then one card

[xfrgtr]

NVIDIA GTX 275

[maxpower4891]

hope this test results are real ..

this should bring some cool air to the ati fanboys!

[Alert89]

A benchmark-session with pro-nVidia games (HAWK & CoD4 are the exceptions). What do you except?

[waver_01]

Funny how different sites have different results. I'll stick to anandtech though, which has the HD4890 beating GTX 275. Not even mentioning OC ability.

[Nuker_]

Random sites such as this use to appear sometimes,with flawed becnhmark results (with the R600 release there was this site that came out of nowhere showing it destroying the 8800gtx for instance).

[Monstru]

So I take it that for all the fanboys is really hard to understand that these 2 cards are really the same stuf, huh? Just buy the one that makes you happy, but do not spend money on more than one, that is my opinion after looking at the final graphs

[konstantine]

From a general view of the is review, i can tell that the reviewer is a stupid amateur. You don't test Crysis and Far Cry 2 with 8x aa and then test world in conflict, HAWX and COD 4 with 4 x aa.......

As for the charts, they are clearly fabricated.
I play Left 4 dead at 1920.1200 res. on max setts with 8x "quality" adaptive aa and 16x high quality af...the last time i played Left 4 dead with fraps running, about 2 weeks ago, The FPS never droped under 120fps....and i was averaging around 140fps...how in hell does an oc 4890 gets only 107 fps at 1680 res, with 4xaa....and default MS-aa and AF.?

In Crysis Warhead, they didn't mention on what setts they tested the cards.....And the results are easily to be ditected as fake.

Here's the deal in Crysis, In the dual card test, 2 4890s beat 2
gtx275s at 1920 res. without aa, and lose with 8xaa.In reality, the exact opposite is what happens.I can tell you that in Crysis at 1920 res. on max setts without aa, you get 50% more performance over a single 4870....and unless you eneable 8xaa "stress the cards", you aint gonna get double the performance of a 4870 in crossfire mode.This is coming from a personal experience. Nvidia cards do get double the performance in sli mode in Crysis with and without aa.but they drop significantly more with 8x aa compared to the radeon cards. You're getting over double the performance of the 4890 without aa which is clearly BS.
As for the single card tests in Crysis, a single 4890 is faster than a single gtx275 with and without aa, cuz my 4870 is almost as fast as my friends GTX285..."the difference is less than half a fps in favour of the gtx285".

A romanian site.......hehe...

[Monstru]

Thank you for your kind words adressed to me, an also for the kind remark about my country. They truly represent the balkan spirit and now I am at ease, knowing that the true spirit of the east will not die easily, even in a world of progress and globalisation. I am sad we got rid of that kind of mentality 20 years ago, but I am happy to see that in Russia things are exactly the same as they were back then.... счастливы коллективизации.

As far as the settings are concerned, I allways use the maximum level of Anti Aliasing that can be set in the game for ATI cards. I couldn't understand the point of any review that tests 3SLI configurations and does not use this kind of level. Even so, if a game does not allow for a higher setting to be made in game, I do not take much interest in forcing the AA from the drivers, for the review is meant to help all users make their decision, and most of them just plug&play their cards. Therefor, all games (except for Hawx, which had some serious issues in 2560x1600 with 8X AA in multi-card configurations) have the maximum settings possible for an ATI card (Nvidia allows for a higher level of AA to be set in the game, but I compare the cards using exactly the same methodology, so the same AA is used on Nvidia cards).That is what the consumer will get at home, that is what I test.

The settings used for Crysis are Enthusiast / DX10, as I have mentioned in the review. If there are any other question or concerns about the testing methodology, I would be more than happy to answer them. An regarding what you say about Warhead, please show us the test that you have written so we can all see what it's all about.

[t4ROKuN-]

Originally Posted by konstantine

From a general view of the is review, i can tell that the reviewer is a stupid amateur. You don't test Crysis and Far Cry 2 with 8x aa and then test world in conflict, HAWX and COD 4 with 4 x aa.......

As for the charts, they are clearly fabricated.
I play Left 4 dead at 1920.1200 res. on max setts with 8x "quality" adaptive aa and 16x high quality af...the last time i played Left 4 dead with fraps running, about 2 weeks ago, The FPS never droped under 120fps....and i was averaging around 140fps...how in hell does an oc 4890 gets only 107 fps at 1680 res, with 4xaa....and default MS-aa and AF.?

In Crysis Warhead, they didn't mention on what setts they tested the cards.....And the results are easily to be ditected as fake.

Here's the deal in Crysis, In the dual card test, 2 4890s beat 2
gtx275s at 1920 res. without aa, and lose with 8xaa.In reality, the exact opposite is what happens.I can tell you that in Crysis at 1920 res. on max setts without aa, you get 50% more performance over a single 4870....and unless you eneable 8xaa "stress the cards", you aint gonna get double the performance of a 4870 in crossfire mode.This is coming from a personal experience. Nvidia cards do get double the performance in sli mode in Crysis with and without aa.but they drop significantly more with 8x aa compared to the radeon cards. You're getting over double the performance of the 4890 without aa which is clearly BS.
As for the single card tests in Crysis, a single 4890 is faster than a single gtx275 with and without aa, cuz my 4870 is almost as fast as my friends GTX285..."the difference is less than half a fps in favour of the gtx285".

A romanian site.......hehe...

I guess what the review site did was just swap the results for the 4890 and 275. I guess they are more than just stupid, lazy as well.

New age of marketing strategy, if you can't convince, you "Con".

[eEnzo0]

Konstantine if I were you I would stick only with the vcards and leave other countries alone.
I am a romanian also so?Last time I watched the exchange rate the the rubla was under the RON.
This doesen;t mean that only you the russian kid can afford a good videocard.Maybe your mom and dad gave you money for it.That doesen;t justify your stupid acts.
Bring us the proofs with Crysis.You said 4870 stays only half of fps behind GTX 285.
I dare you to show us the pics or movie.
If not... phuck off.

[konstantine]

Originally Posted by Monstru

Thank you for your kind words adressed to me, an also for the kind remark about my country. They truly represent the balkan spirit and now I am at ease, knowing that the true spirit of the east will not die easily, even in a world of progress and globalisation. I am sad we got rid of that kind of mentality 20 years ago, but I am happy to see that in Russia things are exactly the same as they were back then.... счастливы коллективизации.

So your the reviewer.
I said nothing regarding Romania.....I love romania as a beautful country and I admire Romanians as great people.

Originally Posted by Monstru

As far as the settings are concerned, I allways use the maximum level of Anti Aliasing that can be set in the game for ATI cards. I couldn't understand the point of any review that tests 3SLI configurations and does not use this kind of level. Even so, if a game does not allow for a higher setting to be made in game, I do not take much interest in forcing the AA from the drivers, for the review is meant to help all users make their decision, and most of them just plug&play their cards. Therefor, all games (except for Hawx, which had some serious issues in 2560x1600 with 8X AA in multi-card configurations) have the maximum settings possible for an ATI card (Nvidia allows for a higher level of AA to be set in the game, but I compare the cards using exactly the same methodology, so the same AA is used on Nvidia cards).That is what the consumer will get at home, that is what I test.

First of all you can enable more than 8x aa for ATI card by enabling
Custom Filtering AA "CF-AA" from the CCC.

As for the setts, I'll agree with you that you can't enable more than 4x aa in Cod 4 and world in conflict from the game...but you can enable
8x-AA in HAWX.....

Originally Posted by Monstru

The settings used for Crysis are Enthusiast / DX10, as I have mentioned in the review. If there are any other question or concerns about the testing methodology, I would be more than happy to answer them. An regarding what you say about Warhead, please show us the test that you have written so we can all see what it's all about.

Yes. how did you get Crossfire to scale with twice the performance of a single 4890 without AA? It scales better than SLI.....!!!

And how does a single 4890, without AA, lose against a single 275gtx?
I can tell, from a persoanl experience, that since the 9.2 driver, my 4870 Toxic on stock clocks is as fast as my friends gtx285 on stock clocks in Crysis warhead, on Enthusiast setts at 1920 res. with and without aa. So there's no way a 4890 could lose against a gtx275 in Crysis W. without AA.

As for Far Cry 2, how did you bench the cards?
Did you use the included benchmarking tool...?
Regarding the results in Far Cry 2, I can tell that at 1920 res. with
4x aa on Ultra-High setts, My 4870 is slighlty slower
than my friends gtx285, "43.2+ fps vs 46.7+ fps", "one week old results", using the benchmarking tool v1.0.0.1, But i don't know how things turn out when enabling 8x-aa......

Doesn't Far Cry 2 support the AA enhancement included in the DX10.1 API...like HAWX does?

[konstantine]

Originally Posted by eEnzo0

Konstantine if I were you I would stick only with the vcards and leave other countries alone.
I am a romanian also so?Last time I watched the exchange rate the the rubla was under the RON.
This doesen;t mean that only you the russian kid can afford a good videocard.Maybe your mom and dad gave you money for it.That doesen;t justify your stupid acts.
Bring us the proofs with Crysis.You said 4870 stays only half of fps behind GTX 285.
I dare you to show us the pics or movie.
If not... phuck off.

What i meant by a "Romanian site".....wasn't anything against Romania or Romanian people.I was refering to the possibility of the site being romanian only in language.....not in identity....You've got completely a different point.

As for the proof, Im willing to buy a digital Cam soon with good quality to upload some vids for you guys.But if you want a proof take a look at this chart from xbitlabs:



And here's the Far Cry 2 chart from xbitlabs, the 4890 is losing in this test due to enabling QUALITY adaptive aa for ATI charts which is more demanding than the default MS-AA.And And in Far Cry 2, ATI's qualtiy Adaptive aa does take effect unlike in Crysis.




And here's the HAWX test from xbitlabs,


You can see how the Adpative aa "on quality" took effect in this test too.In the recent UBISOFT games, the adaptive aa takes effect and affects the performance of ATI cards.But this is how they tested the cards at xbitlabs.In order to allow all HDR efects to be rendered with AA, you need to enable ATI's adaptive aa...But you can still put on "performance" instead of "quality".....

[BLOODY-D]

Originally Posted by Monstru

So I take it that for all the fanboys is really hard to understand that these 2 cards are really the same stuf, huh? Just buy the one that makes you happy, but do not spend money on more than one, that is my opinion after looking at the final graphs

No they are not the same 4890 is the fastest single card right now with less power consumption and DX10.1 support.

and your review is 100% Fake.

[Monstru]

konstantine, I will stop my implication in the debate at this point, as I am not interested in this type of conversation. However, I owe you an explination about the game settings, so I will answer you regarding this part.

First of all you can enable more than 8x aa for ATI card by enabling
Custom Filtering AA "CF-AA" from the CCC.

As for the setts, I'll agree with you that you can't enable more than 4x aa in Cod 4 and world in conflict from the game...but you can enable
8x-AA in HAWX.....

I will NEVER force settings from drivers while testing video cards in real life aplications (games). There are a lot of average Joes out there which simply buy the card, plug it in, install the drivers and play, and I personally do not consider it fair to them to mess around with the CCC or Nvidia Control Panel settings. They should see what they might be able to get as an experience after buying that card I am testing. On the other hand, an experienced user will be able to make a fair judgement seeing the settings that I have used. I know, I am a little bit unfair to non-regular readers, as our regular readers know out methodolosy and I do not explain it in every review, but I also do not have the habbit of showing the reviews around the web, I only do this in cases like this one, where something like this does not exist on the international sites at this moment.

Yes. how did you get Crossfire to scale with twice the performance of a single 4890 without AA? It scales better than SLI.....!!!

And how does a single 4890, without AA, lose against a single 275gtx?
I can tell, from a persoanl experience, that since the 9.2 driver, my 4870 Toxic on stock clocks is as fast as my friends gtx285 on stock clocks in Crysis warhead, on Enthusiast setts at 1920 res. with and without aa. So there's no way a 4890 could lose against a gtx275 in Crysis W. without AA.

I do not "get" any card to do anything. I just install a clean OS, the latest version of the applications I use for testing, the latest drivers, and then I test. If that is how they behave (good scaling in CFX) that is something the driver coders should get "the credit" for, not me. As much as they deserve to be tared and feathered for the poor scaling of 3 GFX in some games.

How does a single 4890 without AA loose against a single GTX 275? I don't know exactly how, maybe we should ask ATI engineers about it. It just does. Do not compare your personal experience with another platform, as results are only comparable on the same testbed. And also....do not compare 9.2 driver with 9.4. The last drivers brings some very good performance impprovements in some applications and some lesser performance in others.

As for Far Cry 2, how did you bench the cards?
Did you use the included benchmarking tool...?

Yes, I use the benchmark tool, as I do not trust no benchmark done with Fraps. These are the settings for 1920x1200 for instance:

1.

1920x1200 (60Hz), D3D10, Fixed Time Step(No), Disable Artificial Intelligence(No), Full Screen, Anti-Aliasing(8x), VSync(No), Overall Quality(Ultra High), Vegetation(Very High), Shading(Ultra High), Terrain(Ultra High), Geometry(Ultra High), Post FX(High), Texture(Ultra High), Shadow(Ultra High), Ambient(High), Hdr(Yes), Bloom(Yes), Fire(Very High), Physics(Very High), RealTrees(Very High)

2.

1920x1200 (60Hz), D3D10, Fixed Time Step(No), Disable Artificial Intelligence(No), Full Screen, Anti-Aliasing(None), VSync(No), Overall Quality(Ultra High), Vegetation(Very High), Shading(Ultra High), Terrain(Ultra High), Geometry(Ultra High), Post FX(High), Texture(Ultra High), Shadow(Ultra High), Ambient(High), Hdr(Yes), Bloom(Yes), Fire(Very High), Physics(Very High), RealTrees(Very High)

Regarding the results in Far Cry 2, I can tell that at 1920 res. with
4x aa on Ultra-High setts, My 4870 is slighlty slower
than my friends gtx285, "43.2+ fps vs 46.7+ fps", "one week old results", using the benchmarking tool v1.0.0.1, But i don't know how things turn out when enabling 8x-aa......

I have no ideea of what kind of hardware each of you guys has, what drivers, settings and so on. But on the same testbed with the exact same settings, this would be really hard to believe.

But if you want a proof take a look at this chart from xbitlabs:

Oh...the poor guys must have gotten something terribly wrong, as I have teste the GTX 285 vs 4850X2, and 4850X2 is a MUCH powerfull videocard then what they are showing, 4870 isn't even in the same league with it.

I was refering to the possibility of the site being romanian only in language.....not in identity....AKA a n-vidia paid out site

It's funny that when there are very well known sites out-there that say that GTX 275 is clearly better then 4890, while I say that these boards are the same, you must think we are a paid out nvidia site. Well, everybody has a right to have an opinion, and I suppose that yours lays with the "ATI paid out sites" that say the 4890 is the second best thing after Christ's ressurection.


My final words on this subject.... THESE TWO VIDEOCARDS OVERALL HAVE A SIMILAR PERFORMANCE!!! You can show me what reviews you like, you can show me what graphs you like, I know what I have tested with my own hand and I'd rather have faith in that. And if that means that you would consider me an amateur or an idiot, so be it, it is a free world and everyoneis entitled to an opinion. I take pride in beeing FAIR and saying what I find out, not twisting the results like other sites out there.


You really want to know my opinion (as I have stated in the conclusion of the test, but I see that Translator doesn;t do a great job)? I hate both ATI and Nvidia right now. I hate them and their marketing strategy. I hate the fact that I have spent 2 weeks trying to see which board is better, just to discover that these products are not the work of some great engineers, but of some very inventive accountants and marketing personel. They really took care to be right "there", to have the performance right like the competition, to give nothing more for this price. And I hate them for this type of strategy and for beeing so hungry for cas$$. ATI bragged about cutting down the price of the 4850X2 for 3 months, but we all knew that they wouldn't do that because of the manufacturing costs. But wouldn't that really have been a great thing? To have such a powerfull board at that price? Nvidia gives us performance with the eye-dropper, instead fo really going beyond the limits and cutting down the price for the GTX 285. But nooo.....why give a few bits of bus extra and a few MB of GDDR for the same price, when you cand do something that can make more money for you....Hence...GTX 275 ....4890....2 boards that bring nothing new, just a new price tag (and a very good overclocking potential for 4890, I have to admit this). I am sick of G200 and 4800. Enough allready, bring out the serious guns and show us the next generation of GPU's, and some serious progress.

BLOODY-D - Sorry mate, you are compeltly right, 4890 is the best card ever. And not only single GPU, just simply the best card. Ok, let's do something different. You tell which side's fan you are, and I tell you what you want to hear, so you can feel good about what you have at home


Sorry guys, the level of this comments is beyond my power of understanding, I will leave you guys to carry on this conversation. Good luck!