The Fastest 3D Cards Go Head-To-Head

[sirrell]

You Say "The loss of performance can only be explained by the lack of CPU horsepower to help facilitate scaling, which can be clearly seen from our overclocking results" Not entirely true...
But you have to realise that your also only using DDR2!!
Why i'll never know, you should have used something like
"790i Ultra mobo & OCZ DDR3 2000MHz"
Well least then the high end cards could really perform..

 

[LarryRF]

Why in gods name would you test a card with drivers not even set up for the cards? Face it, ATI is the top right now...no matter how much you try to slant the tests...I agree with the other guy, "deletes bookmark"
Larry

 

[calamit]

[citation][nom]Haiku214[/nom]Well the main reason why they don't have the 4870x2 and the latest drivers is simply because they made this article a couple of weeks ago. If you could just imagine how long and tedious it is to produce all these data and results. It's just sad that after finally finishing the article, a lot of new stuff has already happened(new drivers and the x2).[/citation]

Normally, I'd let this go, but that fact this is a comprehensive review.
It's still a huge FAIL. If that's the case then there's a huge lack of insight. It's not classified info that ATI's drivers aren't 100% for new cards until 2-3 driver revisions after launch.

 

[phantom93]

Sorry if this is a double post, IE like went wierd on me when i submitted the other one.

The whole 4870x2 thing needs to stop its retarded, no f'ing sh** its faster then the GTX 280, ok its like 2 4870's but combined on one card, actally that IS WHAT IT IS. The GTX 280 is still the FASTEST SINGLE GPU on the market for consumers.

I am not saying the 4870x2 is lower in perf then the 280, it is the fastest card available, but it is not the fastest single GPU option. If nvidia made a GTX 280x2 or sumthin GX2 280 w\e. People would be talking diffrently...

 

[eodeo]

Only complaint I have is that you only go up to 4x anti-aliasing(aa). I don’t play any game without at least using 8x aa. 4x is simply too jaggedy and most games can run fine with 8x aa.

Not to mention that 4870 becomes faster than gtx 280 in most games with 8x aa. Oh wait, i did mention it. My bad.

 

[JerryC]

Where is the 4870x2 ??

 

[3lvis]

Wheres the 9800gx2 sli? I suspect it would have finished near the top. And without the 4870x2 its hardly comprehensive.

Its nice to see that the 8800 Ultra still holds its own in SLI against these newer cards ....and with $199 price tag @ tigerdirect its a better value then even the 4850. It lacks the Nvidia cuda and physics updates, but then the 4850 doesnt have those either.

 

[guusdekler]

PLEASE add the results to the desktop VGA charts .... looks like this is the long awaited update !

 

[gsteacy]

Guys this article was originally posted on August 4th/5th depending on your timezone, and would have taken weeks before that to write. 8.7 might have come out half-way through benching, so they couldn't swap over and still have valid information.

 

[Guest]

Is there a mistake? The single 8800gts 512 does better than the 8800 gts 512 in sli in some cases. Is that a typo or correct?

 

[hurbt]

zzzzZZZZzzz

no real new information.

And SLI vs CF comparisons look terrible...

 

[computerfarmer]

All of us love to see ratings. This helps us choose and helps us feel good about our choices.
GTX 280/260 use driver Forceware 177.39 and all other nVidia cards use Nvidia ForceWare 175.16.
ATI cards, new and not so new all use the same driver version (8.6) in this comparison.
This leads us to try and understand how valid these comparisons are?
Do these cards get better scores on Intel or AMD platforms?

I do understand the amount of work involved in such an article.

 

[bungholio]

Maybe you guys do not realize that the benchmarks for this article were performed a couple of months ago, not yesterday. Updated drivers and 4870x2 were not available then.

The intent of the article was to show at what point the cpu would be limiting performance from various high-end video cards, not "What is the max frame rates possible from a top end system".

 

[rhysee]

Gosh , I,m think TomsHardware is under resourced! no money in the kitty to make decent articles for reviewing the LATEST video cards .. ie 4870X2
I think they use old articles and shish it up and put the latest cards specs and models in to try and get us to read this dribble..

In a real world Toms should should re do all tests with latest drivers , yes it may take longer but for a site of this stature and populatiry its a no-brainer to do it this way ?..

Say 1week max to do tests and results/report compiling ?..

I,ve read all these comments , Are toms gonna comment at all ?.. explain their way outta this ?./ Doubt it

This site and its reports are slowing going backwards in performance and popularity .

Gee maybe another website face change update needed to try and get us readers back ????

These tests dont seem accurate , I know you simply use the older scores from test done way back .. ie Toms never retest the whole suite of cards ,jsut the ones they have on hand ,

Dissapointed and I hope you can get back on the band wagon of decent hardware reviews with real world results and the newest drivers!!

Peace!!

 

[Spider D]

All of you who are complaining about wanting to see "real peoples" CPUs, or wanting to see the effects of CPU bottlenecking, please stop. This article only tested 1 CPU. It can never give the kind of information that you are looking for. Actually I'd be very curious about that type of article too, but it MUST have multiple GPUs AND CPUs. No matter what tom's said, this article is simply incapable of showing comprehensive information about that topic. CPU comparisons are beyond the scope of this article!
As for those going on about PhysX, I just want to point out three things.
1. ATI have a version of HAVOC physics.
2. Any physics performance increases will show on the benchmarks (which I assume would be turned on since Tom's claims some of the tests to have "max settings")
3. If you hate HAVOC and just can't live without PhysX (though they're both almost the same in performance), the project to run PhysX on ATI boards is progressing nicely, with backing from Nvidia itself!

 

[Rescator]

A bit simpler way to view the relative performance of the cards can be seen at NPR GPU Chart and various cards entries are based on tests/benchmarks from more than one site.

 

[randomizer]

[citation][nom]demonhorde665[/nom]This article is perfect BEACAUSE it shows what gains us normal gamer consumers can expect with said cards on our current systems, i just got anew pc like 5 months ago , and im nto rich so yeah i'm on a dual core chip over clocked to 3 ghz and only 2 gigs of ddr2-800 ram , so an article that gives me a closer representation of what i shwould expect with an upgrade is right up my ally , sorry we can't all be spoiled rich kids like you [/citation]
Actually this article gives you an indication of what you would expect 1-2 months ago (the article was published on the 4th/5th August on the German TH site as Ape said). You would actually get better performance with current drivers, but because the article would have been started around mid July, it is only using what was available then.

 

[Guest]

It's kind of lame that the last 3 pages aren't in a single table sorted by price. I mean, we're all looking for the best video card we can afford. We want to pay, say, $300. Which card gives us the best bang/buck for that much money? I would paste them into excel myself except the tables are filled with typos and whatnot so you can't sort them perfectly the same way.

 

[sbuckler]

No min fps

 

[ZootyGray]

Ambiguity. I found myself re-reading to find clarity - clarity was not available. Misleading. So I quit reading.

If Tom's has a delay to publish problem - then THG is going to be gone, because no one wants out of date info.

Re Bias - well it's so widespread and accepted that most people don't even notice.
The bottleneck cpu reveals info and hides other info.
The spintel system reacts differently to an AMD system - and MANY of you would be SHOCKED at what AMD can do on a LEVEL PLAYING FIELD - i.e. without prejudice or bias. I can only tell you that - until it appears, you won't believe it.

Published on German site weeks ago? = pathetic. and possible bias.

Power supply info was useful but this stuff needs more clarity cos people don't understand - me too - and psu info was ambiguous also.

Compare using spintel systems and AMD systems - 1 fast and 1 slow of each = total of 4 systems.
I echo someone else said we don't need every card tested. We need good testing parameters and clear translation of results with NO bias. A better task would have been to match AMD and spintel systems at high and low end (that in itself would be a good review). And then run a cross-section of all cards on all 4 systems.

I read 3 pages of this - and it was sad. I wanted a good review but it seems the workload was too big an undertaking and the tester got tired.

I like to think that THG is over the garbage test reports and articles. I think Chris Angelini has been a good addition - but this junk was probably created even before his time here.

Accurate, unbiased testing, and clear reporting will always transcend fanboy bullsh*t. And if we could have that, you would be shocked at the lies of the spintel machine.

Benchmarks can be chosen to favour any system! And too many sites are basically bribed to LIE TO YOU - surprise!

It's about an7i7rus7.

 

[bf2142-rules-ok]

Jesus could write an article about 'Miracles in the Bible' on Tomshardware and the 'clever dick commentators' would flame it for lack of knowledge base. I suggest they all band together, spend umpteen weeks of hard work actually PRODUCING something ,publish it here if they dare, then we can all spend 10 nano seconds writing a comment about how much better we could have done it.

 

[rhysee]

Wouldnt be hard if someone was keen to write a proper up to date article , I think thats the main point to be made here..

 

[randomizer]

[citation][nom]Rhysee[/nom]Wouldnt be hard if someone was keen to write a proper up to date article , I think thats the main point to be made here.. [/citation]
It was up to date when originally published.

 

[benpot]

I expected more from toms hardware:

Why wasn't the catalyst 8.8 used ?
Crossfire was improved in this version and all HD 4800 series performance was boosted

Why wasn't the HD 4870 X2 tested ?
This is a single card, although it's expensive

 

[randomizer]

[citation][nom]benpot[/nom]I expected more from toms hardware:Why wasn't the catalyst 8.8 used ?Crossfire was improved in this version and all HD 4800 series performance was boostedWhy wasn't the HD 4870 X2 tested ?This is a single card, although it's expensive[/citation]
Maybe if you bothered to read the comments you'd actually know the answers to most of those questions.