GeForce GTX 460 1GB SLI vs. Radeon HD 5870 CFX @ [H]

This is making my decision way harder then it should be.. But the numbers are very impressive...
 
I hear ya man. Just for the record, Im not calling BS on you guys or anyone for that matter. I trust your reviews as God's own words. Just saying that something is amiss somewhere.

I figure that since a single 5870 is much faster than a GTX460, the hardware angle of the ATI cards is obviously squared away. But if putting 2 of them together results in half assed results, then the problem has to lie in the software end......right? :confused:

Im definitely not smart enough to be an engineer so Im just guessing here but I do know its something AMD needs to figure out cause, like I said earlier, there's becoming fewer and fewer reasons to recommend ATI over Nvidia these days. And Im an AMD fanboy!!

Yes, something is amiss somewhere, CFX is scaling terribly, that is what is amiss. This is nothing new, we have been telling you guys for months that GTX 480 SLI kills 5870 CFX, but yet a single 480 and single 5870 come in neck and neck in performance.

12795758727vW3BUbw7C_5_4_l.gif


This issue is not new, it just jumps out at you now since the lowly 460 is looking so good in SLI. Put the single 460 against a single 5870 and the 460 is going to get raped.

We are not writing these articles for funsies but because of what we were seeing pointed out to us in our testing.

As for all of you guys comparing these numbers in articles past, you just cannot do that. You can't compare different editors, machines, maps, drivers, and profiles across months and call bullshit. There is a reason we do ALL THE WORK every time we write an article, because using older data will get you pulling for false conclusions.
 
Thanks Kyle. I know you are compiling a performance review for x16/x8 configurations. For those of us on the fence with P55 boards can you give an estimate of what kind performance loss is expected for a 460 SLI setup? I'm so close to pulling the trigger before these cards become impossible to find.
 
Guys, I have re-run the BC2 test on 5870 CFX, after re-installing latest Cat 10.7a and latest Crossfire Application Profile update again, to verify the results once more, and the numbers are the same, same performance. I have verified that CFX is working in the game as previously stated also. We've done our best to ensure that our test setup is correct, we are using fresh installs of the games, latest drivers, and latest profile update.
 
Last edited:
Guys, I have re-run the BC2 test on 5870 CFX, after re-installing latest Cat 10.7a and latest Crossfire Application Profile update, to verify the results, and the numbers are the same, same performance. I have verified that CFX is working in the game as previously stated also. We've done our best to ensure that our test setup is correct, we are using fresh installs of the games, latest drivers, and latest profile update.

I followed these reviews, and boards closely and have definitely found, even on my single 5850, performance is very nuked anything past it. in some cases my FPS doubled in benchmarks, and in my own run around at least 25% improvement was noticed, again double in other areas.
 
Thanks guys. Really Validated my decision to switch out my setup and stick it out with Nvidia rather than wait and upgrade to 5870CFX. AMD's solution is just goofyfooted from top to bottom. Like in the old days when you had to buy a Master card for crossfire AMD still has most of the work loaded on the primary and the secondary card running along behind like a fitful toddler. Sitting down with Eyefinity and still being forced to use only the primary card made me realize how little AMD's changed their solution.:(

Gonna hold on to my most excellent GTX260's until I run into something unplayable and then jump to GTX460's or whatevers the next step.

Per chance Can anyone tell me if I'd actually get a boost going from the P55 to X58 since I guess I'm only running x8/x8 instead of x16/x16?
 
Well consider that most people are still going to run single card solutions, AMD is doing a great job kicking ass in that department. And they bring multi-display gaming with one card as well.

But surely for those willing to do SLI, this 460 1GB solution is a freaking madman.
 
What about at multimonitor resolutions like 5760x1200? I've read that the 460s start to lose their advantage as resolutions increase and I'm wondering if SLI 460s still beat 5870 CrossFire at ultrawide resolutions or does the 5870 CrossFire setup eventually overtake and win out against SLI.
 
Also, a question for anyone with an nVidia card and a 3D TV, are their drivers able to output 3D to 3D TVs (Samsung for example)? I have a 5870 right now, and even with iz3D's 3D driver I can't get 3D to be outputted correctly.
 
I get about 120 FPS average in BFBC2 at 1920x1080 all high 4x MSAA. Maybe it's just my resolution but I am happy with the results.

I am going to try disabling CFX to see if there's a huge difference.

Without crossfire, same level (Level 1) I get about 85avg. So it's about right as far as crossfire scaling. Using 10.6 drivers.
 
Last edited:
What about at multimonitor resolutions like 5760x1200? I've read that the 460s start to lose their advantage as resolutions increase and I'm wondering if SLI 460s still beat 5870 CrossFire at ultrawide resolutions or does the 5870 CrossFire setup eventually overtake and win out against SLI.

While we have not tested this exactly, here are my feelings on it. The 460 is going to be hampered at 3x1 resolutions to some extent due to textures and levels of AA. If you are a person that lives without AA, I would say 460 1GB SLI is fine for you. If AA is more important then you will likely wan to look at a solution with more RAM.

If you go back through my GTX 280 SLI article you can check the screen shots and see how much RAM those settings I am using are costing. Just Cause 2 we could run 4XAA and be bumping up to the 1GB RAM buffer, but it still worked. :) Batman could not do 4XAA. Check out the settings and the screen shots, I think there is a lot to be learned from that article if you take some time with it.
 
Somebody should make 5870 in cf 10.5 vs 10.7 . I played BF2 with 5870 in 2560 with 4xaa and supers.. many times and had fps in high 70 or more.. From this review somebody would say cf sucks badly but it is far from true, 10.7 and 10.6 sucks for CF. I would use 10.5 for cf in mine PC like many others with that setup..Very shallow review and first one after long time i dont like .
btw. Spitfire is on gtx470 now and 56c with 820/870 and only noisy like s-flex 1200rpm..
 
Last edited:
The only issue I have with the article is you are comparing a factory over clocked GPU to a reference clocked GPU. From what I have seen and done personally, the HD5870's hit 1000Mhz easily. With that in mind, are the GTX460's able to keep up. My initial conclusion would be that they would be slightly behind, but not enough to justify the $300 difference. This is one reason why I think the GTX470's are the sweet spot for Nvidia. They overclock well and SLI well and also cost less than the HD5870's. However it would be interesting to see how much of a difference overclocking the HD5870's would make.

I am still glad to see reviews like this as they show you do not need to hock your grandmother to get great performance.
 
The GTX 460's are a tremendous value, no doubt about it. I'm still kind of surprised that AMD hasn't lowered their prices across their products to stay in line with NVidia's latest offerrings. I'm considering picking one up for my HTPC since it supports bitstream audio over HDMI for Dolby True HD and DTS-HD Master Audio.
 
I have a question. Are these new nvidia drivers also working miracles for older 200 series cards?
Have sli gtx260s and gtx285s also been improved?
 
What about the temps and acoustics of both these setups? Performance is great and all, but it should be noted if there is a Dustbuster/blast furnace in the mix.
 
lol, OCN really hates this review, but I think I did a somewhat decent job of explaining things.
 
Also, a question for anyone with an nVidia card and a 3D TV, are their drivers able to output 3D to 3D TVs (Samsung for example)? I have a 5870 right now, and even with iz3D's 3D driver I can't get 3D to be outputted correctly.

No. 3D TV's aren't true 120Hz technology. You need to buy a true 120Hz computer monitor for it to work. Nvidia is trying to ensure that the next generation of 3D TV's are true 120Hz sets so they will work with the glasses. Check our displays subforum for a list of compatible 120Hz monitors. There is a thread there that lists every last compatible monitor at the start of it.
 
I understand why they used 10.7 since you always have to use the most recent driver during a review.
If reviewers start using older drivers when new drivers are out, the numbers would be flawed.
It's not [H] that ATI dropped the ball on 10.7, but I would like to see best driver vs best driver.
 
I tried, I really did... :(

They are arguing about resolutions, types of AA used, drivers etc.

I also understand their concern about stock vs overclocked cards. When you look at the price difference between the 5870 and 460, I find it acceptable. The 460 is Nvidia's 4th card in their line up. Two stock 460s will still come about 5fps of those numbers, which is not a huge difference.
 
WoW 460 scales well beyond my imagination. I was goint to spend on Radeon 5870 but I might go the other way around. Good Write up guyz.
 
hhahahahaha, I swear I love you sometimes, this is why i keep coming here. But there is some seriously wrong shit going on, i just reverted back to 10.5a myself, on my single 5850, and noticed rather huge gains in all my games.

Yes this may be true that the old drivers have better framerates. But if you're writing a review to show what competing companies are bringing to the table to make their case for supremacy; wouldn't you use the latest offering from the companies? If Kyle had used old drivers wouldn't 75% of the people reading the review say that he didn't use the latest and greatest from AMD, thus the review is invalid?

He used the newest Nvidia drivers; thus in turn the newest AMD drivers were used. It's up to AMD to get their shit together; not a reviewer. And I'm NOT knocking what you said, please don't take it that way. If AMD wants to show their product in a better light, then stop slacking and get to coding better drivers.

Maybe they are working on their next gen technology and thought these drivers were "good enough". I don't know. All I know is that if the methodology used in the review seemed very valid to me. And the results weren't tainted either.

AMD has ridden the 1st to market success for too long and got caught napping. Good thing about the review is that hopefully they will overhaul the drivers to put themselves in a better light. Kyle already said they are looking at performance issues brought up in the review. And that should be great news for AMD videocard users!
 
I love that point of view.

Kyle's getting paid by Nvidia. Well they should fire him.

HardOCP GTX470 review said:
The HD 5850 is the clear value winner when compared to the GTX 470. We have been telling you that the Radeon 5850 is the best value in enthusiast video cards since last year and the GTX 470 does nothing to change that.

We see no reason to purchase a GeForce GTX 470. It provides no gameplay advantages compared to the competition, and will actually end up costing you more power and dollars for the exact same performance you can get with the Radeon HD 5850. Factor in the power consumption, and it doesn’t seem worth it. If you have an HD 5850, stick with it, the GTX 470 is not an upgrade

HardOCP GTX480 review said:
The GeForce GTX 480 consumes more power than the HD 5870, produces more heat, and costs more money (which might also include the cost of a new PSU). It doesn’t seem like the best value with all this factored in.

Factor in the cost and power, and include the ability to run Eyefinity on a single GPU, the Radeon HD 5870, to us, seems like the better value for the gamer right now.

Seems to me, [H] is just calling 'em like they see 'em.
 
Those guys at OCN complaining are not smart enough to figure out that comparing frame rate numbers from real world gameplay using different drivers, different MAPS, different run-thoughs, different levels and methods of AA, can give you wildly different overall numbers when cross-comparing content. This is why we use NEW comparisons for nearly every review. We very rarely reuse content due to driver, game patch changes, and we do not always use the same maps if we decide a different map gives us better GPU challenges.

Then their gripe is that we used factory overclocked cards that are $20 more than standard 460s, but still $300 cheaper than what we compared them to?

I would highly suggest you not hang out with folks that are not smart enough to figure out the differences between simple data sets. According to those dufuses we need to pick and choose the best driver for every game we test then clock the GPUs back to what NVIDIA or AMD tell us is stock. Oh but wait, then we use displays that are too big for these massively expensive video cards. Those guys can't go bigger than 1080P.

I see a lot of whining but little proving.

FWIW we have been in contact with AMD today on this issue and BF2 never came up. AMD is concerned with AvP numbers.....which we have already verified on our end.

I am 100% confident in what Brent and I have published today. Should we be PROVEN incorrect we will certainly address that.
 
Last edited:
Yeah that was funny. A guy with two 5870s said that 2560x1600 was an unrealistic benchmark.
People are running Eyefinity setups that makes 2560 look low res. AMD is getting killed in AvP.
The 460s shows that the tide is shifting. I need some official info on Southern Islands before I make my final decision on switching from ATI.
 
FWIW we have been in contact with AMD today on this issue and BF2 never came up. AMD is concerned with AvP numbers

Thats what I like to hear. As an AMD fanboy, I would hope that they would see this and realize theyve got a bug or two to figure out. I was afraid theyd either not care or just dismiss the review.
 
I was just thinking... how much older is the 5xxx series compared to the GF100s and GF104s? I'm just thinking Nvidia saw ATI's hand for this long, of course they're going to come out with better performing parts.

Also, I agree with Kyle... Crossfire scaling sucks.
 
Any chance of seeing a review of the 768mb version of the card in SLi at a lower rez like 1920x1200? Seemed to perform really well in your original single card review when compared to the 1gb version at that rez. Wondering how big of a difference SLi would make seeing as how its huge with the 1gb version.
 
Yeah that was funny. A guy with two 5870s said that 2560x1600 was an unrealistic benchmark.
People are running Eyefinity setups that makes 2560 look low res. AMD is getting killed in AvP.
The 460s shows that the tide is shifting. I need some official info on Southern Islands before I make my final decision on switching from ATI.

Yea I read that too man. Why would you spend all that money to run at low resolutions? I think one guy was saying that and talking about adding a 3rd 5870. That thread was so full of QQ, I thought a flash flood of tears might overcome their servers.
 
That is some funny shit there. Now CFX is only good for less than 2560x1600.

2500x1600 scales poorly on ATI CF cards from what I have seen. 1920x1200, CF scaling is much better.

For some reason, 2500x1600 + 4xAA slows down ATI CF a lot more. It might be a driver problem at that resolution.

I will have to let the company that brought multi-display high resolution gaming to the masses that its sweet spot is around 1080P.

AMD has a great product that has been with us for 9 months or so now and we are seeing where some of its shortcomings are since we now have the next gen out from its competition. AMD still owns the single card space and owns it well. Dual card is working out differently. I think this is all a great thing for the enthusiast.

Which one of you posted about those dunderheads in our forum? I feel a poofing coming on.
 
Yea I read that too man. Why would you spend all that money to run at low resolutions? I think one guy was saying that and talking about adding a 3rd 5870. That thread was so full of QQ, I thought a flash flood of tears might overcome their servers.

It was the same guy. 5870 Tri-Fire at 1080p. I would buy two more monitors instead of a third 5870, that's how [H] members think.
 
I still reckon its a driver thing because GTX460 SLI is only a bit faster than a single 5870 as the previous [H] review showed, I do find it difficult to believe that 5870 CF barely catches up any ground at all.
 
I really hope its a driver problem because if its not alot of ati owners will shit a brick....lol
 
OK, the proof is in the $$$$.

I can sell my 5870 for $300, and buy 2 MSI 1G OC Cyclones for $576.28 (tax + ship incl).

For my setup in my sig. Should I pull the trigger?
 
Back
Top