ASUS ENGTX580 Video Card Review @ [H]

FrgMstr

FrgMstr

Just Plain Mean
Staff member
Joined
May 18, 1997
Messages
55,730
ASUS ENGTX580 Video Card Review - The ASUS ENGTX580 packs NVIDIA's award-winning GeForce GTX 580 GPU inside, and it comes with a small factory overclock free of a price premium. The GTX 580 is certainly faster than the GTX 480 and the Radeon HD 5870, but what about a pair of Radeon HD 6870s in CrossFireX?
 
Nice performance from the CF 6870 cards.

Cant wait to see what the 560 can do.
 
Thnx for the arti, V good comparison - gtx580 and 5970 are the same price here, not to mention 2x 6870s, even less:eek::eek::eek:..HOWEVER?????? has MGPU improved with Win 7?? Random slowdowns?? Microstutta? Freeze-n-go? A few words about this from ed??....
 
Great teaser, but where is the Eyefinity 6970 CF vs Surround 580 SLI article?
 
I just love Asus.

Normally I would have skipped a GTX 580 review but when I saw it was Asus, I had to stop and see what was up.
 
i started hating multi gpu configuration ever since i switched to AMD 5870 crossfire. Im gonna be selling those two POS( theyre awesome as a single card though) once the 580 drops a bit in price in the next few months. i dont think i will multi gpu ever again since im pretty much replacing the GPU once a year.
 
I think I've had just about every combination of GPUs in the 58xx AMD, 480 and 580 in nvidia; surround and EF.

Cards from ASUS, XFX and EVGA as well as MSI.

The 580 is a very nice card.
I had too many small problems with CF.

I think your article reinforces that. Nice reading.:D
 
Thanks for the great review guys. It was really interesting to see the different GPUs excel in their own departments. The 6870 CF had a surplus of shader performance, so you guys used CFAA on it. The 580 had memory to spare, so you turned up SSAA. And it proved something I learned a long time ago from these reviews: Apples to apples comparisons aren't everything.

EDIT: And look at the 6870s destroy the 5970 :O.
 
Good article, I have a question though:

At levels of AA higher than 8x, irrespective of the filtering method or the method of AA, is there truly an image quality difference?

A better question is, even if there is a difference in still image captures at those levels of AA, is it something you can notice in game that makes a difference?

Back a few years ago, one review that did IQ comparisons of AA with several games, one I remember being World in Conflict, proved that going from 0 to 2 or 2 to 4 had large impacts on IQ, but going from 4 to 8 was much lesser and 8 to 16 was practically unnoticeable.

To further expand the question, I would postulate that at 8x AA 2560x1600 has been 'taken to the limit of graphical quality'. There is no point of 16x AA, rather a jump to higher resolution is required. Since there is no standard (read: one monitor) resolution jump after 2560, to me that says we've hit a sort of maximum for graphical display power. Would you agree?
 
Great article! I'm enjoying learning all of the new data points around the 580 cards with respect to multi-gpu setups of various types. I am greatly looking forward to your impressions and experiences with multidisplay setups. Keep up the great work, and keep disseminating the data to the masses!

One spell check item: On page 8, it says "to smell", I believe you mean "too small". Feel free to delete or edit this message upon correction!
 
I too would love to see 580 SLI vs 6970 CF, also, overclocked 580 against overclocked 6970... :)
 
Outstanding article! You took roughly $500 and pitted AMD against Nvidia which is *exactly* what I hoped [H] would do in a review. The insights into the abilities of each video card are quite on the mark and all in all this is an excellent review! I even learned something abuot the edge-detect abilities of my cards I own. Mark and Brent, take a bow, you all are upholding [H] standards pretty damn well and ARE the benchmark for the other tech review sites. Damn fine job gentlemen.

Very much looking forward to the 69xx and 5xx multi-gpu scaling charts. Thanks again!

Update* - I can't help but wonder what happens when you raise the limit by $100-150 and allow for 2x 6950 with their 2Gb of memory per card. Thats why i'm looking forward to the 69xx vs. 5xx multi-gpu battle :)
 
Last edited:
Considering the less-than equal matchup represented here, a $480 crossfire setup versus a $500+ single card you can't actually buy, I think the HD6870s performed admirably. I do think though, this article should really have been done with two HDr6950s. It's good to see how, at sensible levels of AA even the HD6870s can beat the 580, but the 2GB of video memory on the 6950s plus the more 2560x1600-capable architecture would have shown really just what a massive lead they take.

Even ignoring the current availability crisis, I was never that impressed with the GTX580, it's a lot to pay for the 'fastest single-GPU on the planet' trophy given how little that actually means in the real world. At anything but enormous levels of AA, even a pair of HD6850s reliably match, or even sometimes beat the GTX580. They didn't come out any later, they don't use any more power, and they're less than 3/4 of the price.
Unlike cards like the GTX280 in the past, where you can forgive all their quirks for the extra single-GPU gains, the 580 can't really offer anything but its single GPU scaling. Given not only that nvidia aren't able to offer multi-display without a second card, and that scaling of the new radeons is over 90% now on average, I consider the 580 a bit of a flop, certainly nowhere near as worthwhile an investment as the 570. That too, however, I feel is overpriced, as the same caveats about the HD6850s applies - here in the UK i can buy two 6850s for slightly less than I can buy a GTX570, and in a fair few games they still use less power. Given the considerable performance advantage they offer, I'm having a hard time recommending either of the GTX500 cards.

g-money: This is something HardOCP have covered in several previous articles. I wondered it too, and when I looked, I can see a difference, I'm not going to lie. It's pretty slight though, and given the performance drops it causes, I rarely find myself thinking it's worth it, except if you're running an easy game with loads of surplus power [i.e. you're getting 200+ fps or something silly].

Frankly, my next upgrade's almost definitely going to be either 2x6950 or 2x6970. I'm eagerly awaiting a HardOCP test benching these.

Also, now I'm finally registered on the forums, a BIG thumbs up to HardOCP for not caving to the HAWX 2 debacle. Massively impressed with your integrity there. In the age of biased benchmarks and skewed results, this is one of increasingly few sites I still trust.
It's a pleasure to finally be part of its community.
 
the people asking for the 6900 comparisons in this review need to realize how long it takes them to do the reviews, not to mention how many actual test cards [H] receives to do their 6900 series reviews. if i remember correctly from what kyle said a while back it's something like 2 weeks to do a review + editing. so that would put the start of this review well before the 6900 series review was finished.

nice review mark, glad to see you put the 6870 CF in there. but damn $500+ for a single card that isn't dual gpu is just ridiculous. i guess the limited supply allows them to charge that insane price.
 
Absolutely, and I understand how much time and effort goes into producing a review properly.

The GTX580 came out here in the UK at £390, which removing tax and converting to USD works out at $515, about the typical US price. However, with the lack of supply you can be very lucky and get them for £420 ($555), but most of the time when they trickle into stock they're £450/460 ($595/$610). That's an astronomical amount of money to pay for a card like that, when two HD6850s seem on average 25% faster, and are only £265 ($350), or when two HD6870s are on average 40% faster, and only £380/$500. Even two HD6950s are cheaper here at the moment!
 
samuelmorris said:
Absolutely, and I understand how much time and effort goes into producing a review properly.

The GTX580 came out here in the UK at £390, which removing tax and converting to USD works out at $515, about the typical US price. However, with the lack of supply you can be very lucky and get them for £420 ($555), but most of the time when they trickle into stock they're £450/460 ($595/$610). That's an astronomical amount of money to pay for a card like that, when two HD6850s seem on average 25% faster, and are only £265 ($350), or when two HD6870s are on average 40% faster, and only £380/$500. Even two HD6950s are cheaper here at the moment!
Click to expand...


man i feel bad for you UK people sometimes. especially with that asinine VAT tax you throw on top of the actual price.
 
nice review. its looking like 2gb is going to be the new high end standard.

but at this rate guys your going to have to start testing in 3 monitor setups to be able to show the differences.
 
sirmonkey1985 said:
man i feel bad for you UK people sometimes. especially with that asinine VAT tax you throw on top of the actual price.
Click to expand...

I feel more sorry for those blokes in the UK who don't have Micro Center like many of us in the US do. As a result, they can't buy a Core i7 950 for $199 or an AMD PII 1055T for $179.

Awesome review guys. I never realized that a couple of ATI 6870's in CFX could offer performance so competitive with the mighty GTX580. It is awesome to see that AMD has really taken CFX scaling to a whole new level with the 6000 series.
 
amdgamer said:
Awesome review guys. I never realized that a couple of ATI 6870's in CFX could offer performance so competitive with the mighty GTX580.
Click to expand...

The 5xxx series saw CFX scaling in the 90-95% range with the vast majority of cards and games. Even the 4xxx had very good scaling of its mainstream and enthusiast level cards in both real-world gameplay and synthetic.

What I feel is even more astounding than the scaling levels of both Nvidia and ATI are the quality level of drivers and frequency of updates we are seeing from both manufacturers.
 
sirmonkey: It goes up to 20% [from 17.5] in January :(

Valset: Possibly, but I definitely want to keep eyefinity/3D Surround separate from single monitor setups. I'm glad HardOCP test 2560x1600 as, apart from it being the resolution I personally use, it's the most demanding size you can have on a single screen, and besides the fact that not everybody (myself included) likes gaming on multiple displays, it's also not even possible on single cards from nvidia, so for now it does have be tested in isolation.

$200 after MIR is quite a big caveat. I hear lots of things about people never getting rebates, and there being no law to protect them if the manufacturers ignore them. I'm very glad we don't have rebates in the UK (IIRC They're illegal in a lot of european countries)
 
Is there a mix up in the overclocking results? When overclocked it got lower minimum FPS in Battlefield Bad Company 2 game? Or the results reversed for minimum?
 
Nope - that's what the graph is for. The minimum FPS is indeed lower, albeit only really in one place.
 
Nice review, can't wait for you guys to post the 69xx crossfire article. There some impressive value for everyone from both camps this generation.
 
Nice to see the 68xx's standing up well in crossfire. Makes me super happy with my 2x6850 for $345. Just goes to show that the 580 is a nice card, but very overpriced.
 
3DChipset said:
Is there a mix up in the overclocking results? When overclocked it got lower minimum FPS in Battlefield Bad Company 2 game? Or the results reversed for minimum?
Click to expand...
unless its a significant drop(didnt notice it and to lazy to go back and check it) its probably just a variable in the AI doing something different from the main run through mark did in the review.
 
i'm guessing this review was done or atleast started before the 6970 was in your hands ?
 
Error on page 8 under Overclocking Impressions "The buttons and sliders are way to smell" ....so what exactly do those sliders smell like? :p
 
Im curious why 2 GTX 470 in sli aren't included in the comparison charts for any of these 580 reviews. they fall in the same price range as all the options in this review and I imagine their average performance would be way higher than any of the above.

it's not a big deal but it strikes me as odd that no one seems to wan't to mention dual 470's when they clearly a contender for best value in the $400-$500 price range at the moment.
 
I also would have liked to see AMD's top single GPU and also SLI GTX460s 1GB in the reviews, but I know these tests take a long time so just food for thought. This isn't a complaint though! Great informative review, and really shows the strength of multi-GPU setups when drivers are there.
 
I am amazed out how much better regarded the 580 is over the 480. The 480s were hotter and louder but with my setup I really haven't noticed the difference. The 580s are a bit faster and did smooth out performance in places it was lagging in 3D Surround especially however.
 
Asus video cards norm come with 3 year warranty as well (why i norm buy them grate for folding on if it dies no fuss just get an replacement), what's listed on trade site i buy my bits from all the other cards have 2 normally

need to sell my gtx480 and get an GTX580 before this card is not worth much (ATI cards not an option untill they put more work into there drivers and Drop .Net CAT as its shit and very unpredictable, like the V2 fah tracker updated build does not work)
 
Can someone please explain to me why the same card has to be reviewed from every manufacturer?

"Hey look, the new ASUS GTX 580 is out! - We totally have to review it and see how it fares against ALL THE OTHER GTX 580 card!!!"
 
Have you ever thought that part of the purpose of the article was to show how well a pair of 6870s - for less money - perform against a GTX 580?

Taer said:
Can someone please explain to me why the same card has to be reviewed from every manufacturer?

"Hey look, the new ASUS GTX 580 is out! - We totally have to review it and see how it fares against ALL THE OTHER GTX 580 card!!!"
Click to expand...

Thanks for the review [H]. I know what I'll be spending my money on and it certainly won't be a GTX 580.
 
AceGoober said:
Have you ever thought that part of the purpose of the article was to show how well a pair of 6870s - for less money - perform against a GTX 580?



Thanks for the review [H]. I know what I'll be spending my money on and it certainly won't be a GTX 580.
Click to expand...

Like I said - they've already done that with the other GTX 580.

Why do it again with the same exact card - just by a different manufacturer?
 
g-money said:
Good article, I have a question though:

At levels of AA higher than 8x, irrespective of the filtering method or the method of AA, is there truly an image quality difference?

A better question is, even if there is a difference in still image captures at those levels of AA, is it something you can notice in game that makes a difference?

Back a few years ago, one review that did IQ comparisons of AA with several games, one I remember being World in Conflict, proved that going from 0 to 2 or 2 to 4 had large impacts on IQ, but going from 4 to 8 was much lesser and 8 to 16 was practically unnoticeable.

To further expand the question, I would postulate that at 8x AA 2560x1600 has been 'taken to the limit of graphical quality'. There is no point of 16x AA, rather a jump to higher resolution is required. Since there is no standard (read: one monitor) resolution jump after 2560, to me that says we've hit a sort of maximum for graphical display power. Would you agree?
Click to expand...

I've asked this a few times but never seen a reasonable answer. I run at 1600x1200 and find anything more than 4xAA overkill. Seems folks like to overstress their cards for marginal benefit.
 
You must log in or register to reply here.
Back
Top