The Fastest 3D Cards Go Head-To-Head

[dimaf1985]

well this article would have been great except you guys neglected to use updated drivers. how professional.

the 4870x2 should have been in there for sure.

and i can't believe the amount of articles that keep mentioning the heat issues of the 4850. ppl, all you have to do is edit the xml file that governs the default CCC profile. the default fan speed is 2%. change it even to 30% and temp won't go above 70 degrees under load. mines at 50% and i still cant hear it and the temps never go above 65 degrees.

 

[bydesign]

This review should have been published a month ago when all the other sites had their reviews. This could have be completed days so stop with apologist crap. The fact is Tom's is no longer a relevant review site. There is nothing that you will learn from this review that wasn't published already in July from other sites. Some of those site also had a preview for the 4870x2. More importantly as many have stated there are better drivers.

 

[Razorbladehaze]

Try to make this relatively quick.

I tend to agree that the attempt to do a comprehensive test is a good idea, especially with all the cards that are in close performance proximity and the different variations (especially on the nvidia side).

Yes I am pro ATI, but do use both brands. Pro Intel and use AMD stuff to.

I tend to agree with many here that the results just don't make a lot of sense. As many on here point out. They contradict some of Tom's own benchs and are also erratic compared to many other review sites. Both SLI and CS tank in almost all tests?

Yes I agree that the drivers are an issue for reviews, not just on toms but with many other review sites. Something tells me that its to shortcut and use previous results in some cases.

Yes i agree that leaving out the 4870x2 while including both the gx2 and 3870x2 is suspicious.

I disagree and think that using a Means test(avg) for all tests combined is actually a pleasant thing to see but may want to eliminate outliers.

Good article idea, mediocre implementation, suspicions abound. Especially about pro-nvidia results (as the one guy said not but a few days after nvision, additionally Tom's tends to have a history of nvidia bias in not just numbers but text as well) that tend to be contradictory to the aggregate of all the other review sites I check (about 5).

Additionally more tests should be run and included different gaming genres.

 

[clintox]

"The Fastest 3D Cards Go Head-To-Head" minus the fastest 3D card. LOL

 

[MushroomMap]

The consensus around the net is this is a poor review.

- Testing was done under 3 ghz, can you say bottleneck.

- What kind of reliable result will say that a single GTX280 outperforms two GTX280's in SLI

- ATI testing results are not on par with other results from multiple sources

 

[jaragon13]

[citation][nom]eklipz330[/nom]i don;t agree with anyone that says they should have used a more powerful cpu, read the title, they're trying to make a pointbesides, there a plenty of benchmarks out there with more powerful cpu's... i don;t need to see another one, this is good to see on how the gpu's do on these not so great procs if any1 understands what im trying to say[/citation]
No.I say it should be done with a VERY low end proc,low end proc,middle processor(dual core),middle processor(quad core) and extreme processor(QX9770,for example).
Also,the LATEST drivers are a NEED.No offense,I am NOT going to read something like this if it's not even up to date.I don't need any "should be ballpark-like estimate",I need "damned accurate".

 

[Guest]

36 pages of nothing :

Old drivers. No HD4870X2. Bad CPU.
Temp reports are wrong.

 

[silversound]

WHERE IS 4870X2?????????????
Tino, you do not wanna see ATi red on top or nVidia provided more subsidice?

 

[dimaf1985]

well this would have been a great article except for the fact that catalyst 8.6 drivers dont do the 4000 series in CF mode any justice, so really all these benchmarks can go straight out the window. anyone who comes across this article who doesnt know about this, is going to make an uninformed decision based on flawed data.

also no 4870x2? this article is unfinished.

and i am just amazed at the amount of articles that are STILL being written about the 4850's temps. ppl, just edit the xml file that governs the default CCC profile. you'll notice that the default fan speed is 2%. obviously it heats up. set it to 30-50%, you won't notice any noise at those levels, and temp won't go above 70 degrees under load

 

[Guest]

Comon tom's where is the 9800 GX2?! Everyone knows that it outperforms the GTX260 and is on par with the GTX280, you can't ignore facts. You shouldn't exclude cards just because they are one generation behind, it still outperforms the current generation!

 

[Roost]

Comon tom's where is the 9800 GX2?! Everyone knows that it outperforms the GTX260 and is on par with the GTX280, you can't ignore facts. You shouldn't exclude cards just because they are one generation behind, it still outperforms the current generation!

 

[chaos23]

This article should have been posted weeks ago.. Should have used the latest ATI drivers.. and you guys should have included the 4870x2..

 

[anonymous x]

what's with all these people begging see a review with a quad core @ 4GHz? Very select few cpus can get that high, even on water. I like to see a q6600 at 4 GHZ without cpu lapping stable all the time with games. I appreciate toms on using test cpus that normal people can afford.

 

[Guest]

Unfortunately you spoiled your otherwise nice review by simply adding the the points of the categories to total points. The result is highly biased, as you rate power consumption twice as high as performance, and price and noise as high as performance. I don't think a lot of people share this rating. You should weight the categories, e.g. performance 75 %, noise 20 %, power consumption 5 %, price 10 %, ....

Cheers,

MJ

 

[fshaharyar]

yeah TM everyone has a point that the ATI Catalyst 8.7 wasn't included. it would have made a far better comparison for the CF against the SLI coz performance of CF increase at about 5%-10% and single mode it gives the boost of 10%-15% over ATI Catalayst 8.6. I guess i will thank u for this article just for giving some light to lower end cards like 86XX and 38XX series. There is a performance comparison of 8.6 and 8.7 on Tweaktown and here is the link
http://www.tweaktown.com/articles/1514/ati_catalyst_8_7_analysis_xp_vista/index.html

 

[enewmen]

Like others:
I hoped to see a 4870 X2 and 4870 X2 CF.
Just repost the same article with some card and driver updates.
Please...

 

[Guest]

...and the 4850X2 will be out shortly!

 

[Guest]

You know... AMD sells an entire platform called "Spider"(and by platform, I mean parts designed to work together), although I only see ATI cards being benchmarked on Intel CPUs. Of course, I know that 10 Intel fanboys will jump on and say "because AMD cpus are teh suck, they are (supposedly) 10% slower clock-for-clock, which makes them unworthy." Granted, typical OC of a Phenom is like 3 to 3.5ghz, and typical OC of a Q6600 is 3 to 4ghz, but apparently because there are a minority of people who get that extra 500mhz, it makes Phenom a "bad overclocker". Then come the complaints that Phenom uses too much power, although when you subtract 20w that the IMC is using(and consequently the chipset is not using), then it's right on par with Intel.

 

[craig hallworth]

Great article! It totally reinforces that these new cards are a total waste of money when you can get a 8800gt SLI setup for a couple hundred and have comparable results. Because, really, who has a 4ghz quad core chip?

 

[Spider D]

This article has AWESOME potential! You guys at Tom's just need to retest with a few changes.
1. Overclock or upgrade your CPU (push those GPUs so we can see the graphics bottle neck)
Yes most people do not have extreme quad cores, however it is unfair to ignore the full power of high end cards. For weaker cards to get similar scores to the most powerful is simply a lie. Nobody wants to upgrade their PC based upon lies. I want to know what is the most powerful card for my money, not just what will work well with my current CPU. I upgrade individual parts reguarly, so I need to know what will work well with my next CPU.
2. Use newer Nvidia and ATI drivers
Catalyst 8.6 does not show the power of the 4850 or 4870. All your ATI results are invalid, end of story.
Both companies have much better optimized drivers now. They're not mature drivers yet (atleast not for the GTX260/280 or HD4850/4870), but they're good enough that neither company is likely to be able to make big improvements in the next few months.
3. Include the 4870x2 in both single form and CF form.
You cannot have a "fastest graphics card" article which does not include the fastest graphic card.
4. Use percentages for your overall performance and value comparisons.
Overall fps is misleading as you mentioned in the article. A game that pulls 100fps is worth 3 times a hardcore graphical game like Crysis which can barely scrape 30fps. That is totally unfair.

If you do these things, this article will be the greatest Graphics card article in existence! Please, please, please retest.

 

[Guest]

Simply outrageous that the latest Catalysts weren't used. (note: My current and previous 5 gcards have been NV)

 

[GavinLeigh]

Great article. I love to be able to come back to this kind of review and look at information which is relevant to me. I am impressed by the total coverage of cards, even down to 8600GT. Personally I see this as a fair review. I am a long time Nvidia user who is now looking at ATI due to price/performance. Then there is that whole substrate issue and worries about Nvidia durability. This gives me the information I need. Thanks for putting so much work into this.

 

[joshuasgto]

[citation][nom]Duncan NZ[/nom] Its a GPU card comparason, not a complain about not having a powerful enough CPU story.

Oh? And please get a native english speaker to give it the once over for spelling and grammar errors, although this one had far less then many articles posted lately.[/citation]

You have to love the irony of someone complaining about spelling and grammar errors when they spell comparison "comparason" and write "not a complain" instead of "not a complaint".

And to keep on track with the article, though I am an admitted Nvidia fan - I am also a bit disappointed with the lack of inclusion of the 4870x2. I did not realize, until others had posted, that this review was performed several weeks ago.

 

[jerreece]

I was quite shocked to see how my 8800 GTS (512mb) card rated compared to these current generation cards. I would have expected to see the modern nVidia cards hold out more against my card. From the different reviews and benchmarks I've seen, the 9 series GeForce cards are nearly a joke.

There's a lot of resentment in the comments to this article that Tom's didn't use the most up to date ATI drivers. Personally, I agree with this, but I'll note to everybody that the 4850 still won! Even with drivers that didn't officially support the video card, it still won overall title. So kudos to AMD/ATI!!

As for nVidia, maybe I'll upgrade my PSU and get a second 8800GTS (512) and hang tight with SLI. I see no point in getting a 9 series card.

 

[gwolfman]

I congratulate you on not over-exaggerating the power requirements! *me applauding you* It's good to see the truth. All that 1KW PSU promotional crap really confuses people who don't "truly" know. Thank you for a good article!