HD 6800 Crossfire Scaling

[exar333]

Mrk6 actually I think the mircostuttering was fixed in one of the last few catalyst drivers.


6870x2 draws less power than 1x480 = lol
6850x2 draws less power than 1x470 = lol

fermi has some serious power issues.... hopefully those will be gone when nvidia move on to 28 nm.

The GTX 460 is MUCH improved with power usage, and is actually pretty competitive with AMD's offerings. The 470 and 480 are higher though.

 

[Arkadrel]

http://www.kitguru.net/components/graphic-cards/zardon/his-hd6870-crossfirex-review/7/

Some more scaling data in this review. This is the first review I've seen showing the 850core 460 FTW models against stock 6870s in Crossfire. The 6870s at stock are still faster in 6 out of the 7 games benched.

So 2x EVGA 460 FTW @850 core in SLI, get beat 6 out of 7 times? vs the 6870s at stock speeds.

Lmao those 6870s just need a little bit of overclocking too and it ll be 7 out of 7 games.

 

[Kenmitch]

So 2x EVGA 460 FTW @850 core in SLI, get beat 6 out of 7 times? vs the 6870s at stock speeds.

Lmao those 6870s just need a little bit of overclocking too and it ll be 7 out of 7 games.

The 6xxx series is looking good so far :thumbsup:

The one game they lost in still had better min fps anyways! I'd think that with another driver update or two and they'll do even better.

 

[cusideabelincoln]

I have to admit doing this chart surprised me.

I didn't know or expect there to be so many games that scale above 85%, let alone so many get in the mid-90%. I would usually just assume "oh hey add 50% as an average and 80% if you get lucky). Turns out I was underestimating both SLI and Crossfire. With the continual improvement of drivers and the fact just about every new game will get support from here on out actually gives SLI/Crossfire some potency.

Of course there is always room for improvement. I hope to see even better consistency in the future.

 

[Arkadrel]

I think AMD deliberatly put extra effort into the Crossfire.. Im sure with new drivers things will keep improveing.

You have to remember the 10.10 beta drivers these cards used in benchmarks are probably no wear near perfect yet. Once some of those kinks get worked out, speeds are likely to improve, both on the singel card performance and crossfire performance.

Im pretty sure we can only expect things to get even better for the 68xx's

 

[WaTaGuMp]

Question for you multi video card folks. Do you take advantage of the higher amount of AA you can use? What do you usually run and how much better do games look? For the 1st time ever xfire actually has me curious.

 

[lavaheadache]

so is there any reason why these scaling improvements wont carry over to the hd5xxx series? I find it hard to believe that it is the hardware that is making the difference

 

[busydude]

Question for you multi video card folks. Do you take advantage of the higher amount of AA you can use? What do you usually run and how much better do games look? For the 1st time ever xfire actually has me curious.

Once you start using AA, you can never go back.

 

[WaTaGuMp]

Once you start using AA, you can never go back.

I use AA already, just cant use higher then 8x.

 

[lavaheadache]

ok just for kicks i just tested my scaling in avp and metro 2033

dx11 1080p


avp 93% single card 58 cfx 108


metro 2033- 80% single card 33 cfx 53

seems like pretty much the same scaling as the 6800's

 

[MrK6]

so is there any reason why these scaling improvements wont carry over to the hd5xxx series? I find it hard to believe that it is the hardware that is making the difference

Actually it most likely is hardware that is making the difference. And it's really not the whole 5xxx series, as 57xx CF scaling is pretty good, but more so the 58xx cards that are problematic. My guess is they rebalanced the 6870's, improved I/O's, and removed whatever other bottlenecks were hampering 58xx CF performance.

 

[happy medium]

but more so the 58xx cards that are problematic.

?
Wow, since when? and why?

 

[MrK6]

?
Wow, since when? and why?

Read the whole post, I already answered that.

 

[lavaheadache]

guess you missed my latest post with actula figures?

 

[happy medium]

Read the whole post, I already answered that.

My point is, if I were to have said that 3 months ago, you would have given me 4 or 5 resons why crossfire was better then sli and how great AMD and crossfire is..

Now it seems you can find all the cases where the 5800 series has done poorly to make the 6800 series look like its spectacular and much improved.

I usually like reading your posts, but lately you sound ridiculous. :|


Personal attacks are not acceptable.

Moderator Idontcare

 

[MrK6]

My point is, if I were to have said that 3 months ago, you would have given me 4 or 5 resons why crossfire was better then sli and how great AMD and crossfire is..

Now it seems you can find all the cases where the 5800 series has done poorly to make the 6800 series look like its spectacular and much improved.

I usually like reading your posts, but lately you sound ridiculous. :|

Where did I say CF wasn't better than SLI in that post? Where did SLI even come into this? I said that the 58xx cards had more trouble scaling than the 57xx cards and hence were problematic. Furthermore, how does any of that green light a personal attack on me?

 

[cusideabelincoln]

ok just for kicks i just tested my scaling in avp and metro 2033

dx11 1080p


avp 93% single card 58 cfx 108


metro 2033- 80% single card 33 cfx 53

seems like pretty much the same scaling as the 6800's

That's good, but those are only two games. Also you can't accurately compare your numbers to everyone else's because you're using different resolutions, different hardware, and probably different settings. What you could do is get a couple of HD 6800 cards and see how they do in your setup, though.

But that's why the chart is separated by review site. Everyone uses a different methodology. There are some consistencies, though. Battleforge, Bad Company 2, Alien v. Predator, Hawx, Unigine, and Crysis show similar results from review to review.

Metro, though, shows different scaling from review to review.

 

[Grooveriding]

ok just for kicks i just tested my scaling in avp and metro 2033

dx11 1080p


avp 93% single card 58 cfx 108


metro 2033- 80% single card 33 cfx 53

seems like pretty much the same scaling as the 6800's

Test some more if you have time. It would be interesting to see how much of the 68XX Crossfire scaling improvements have been carried over the 58XX series.

This is the first time I have seen Crossfire overall outperform SLI. Part of my switch from 5870CF to 480SLI was the better scaling of SLI. This is going to make 6970 CF all the sweeter to switch back to.

 

[Arkadrel]

Yeah my impression was that with the 5xxx cards, the counterparts SLI was better than the Crossfire scaleing.

Which is why the 68xx crossfire scaleing is getting so much attention, its beating SLI for once, something AMD/ATI has never really had before overall.

 

[happy medium]

how does any of that green light a personal attack on me?

Personal attack? You gotta be kidding. You freaking crybaby.

Edit : If you read it the way I meant it, it was compliment to your posting style that I hate to se go down hill, jeeeeez.


Personal attacks are not acceptable.

Re: "You freaking crybaby."

Moderator Idontcare

 

[betasub]

Personal attack? You gotta be kidding. You freaking crybaby.

Oh, a smilie - I guess that absolves you of name-calling too.

 

[Grooveriding]

Question for you multi video card folks. Do you take advantage of the higher amount of AA you can use? What do you usually run and how much better do games look? For the 1st time ever xfire actually has me curious.

Always use AA, as well as 16x AF and I set texture filtering to the highest settings globally via drivers.

I find after 8XAA there is not a difference I can see with my eyes. Super-Sampling AA really makes a nice impact on IQ. It's reasonable to run it in older games but on newer games it is a serious performance hit and makes performance tank.

 

[Silverforce11]

It's a hardware issue, 6K series was designed to be much better at CF scaling ground up. The same trend is present in Cayman. This means Antilles is going to blow your mind .. and budget.

 

[3DVagabond]

It seems to make sense that it's a hardware issue. I've pointed out before, physically a 5770 is 1/2 of a 5870. It's not 1/2 the performance though. So, when you crossfire it you aren't going to get 2x the performance. At best, if it scales perfectly, you are going to get back to 5870 performance. Pretty much, that's exactly what you get, too.

 

[Idontcare]

Oh, a smilie - I guess that absolves you of name-calling too.

No it doesn't. Folks need to dial back the level of rhetoric currently being used in their tit-for-tat escalating personal flame-battles.

Infractions will continue to be meted out until morale improves, people WILL be happy....or else!



Moderator Idontcare

(seriously though, way too many personal attacks across the threads in the forum these past couple of days, it needs to stop)