ethomaz said: disolitude said: They are. Sure it is... most games these days get 95% scaling on the second GPU.  Not true... more close to 50%. ATI: Catalyst 13.1 WHQL HD 7790: 12.101.2.1-130313a March 22, 2013. http://www.techpowerup.com/reviews/AMD/HD_7790_CrossFire/

That's a low end card which as bandwitdh issues. Also that's an average of games some of which don't scale well. The ones that scale do at almost 90% efficiency. However you have games like Batman Arkham Asylum which don't scale at all. Essentially console ports usually suck at scalling

Games like Farcry 3, Crysis 3, Metro 2033 or any powerhouse games that are properly coded for PCs often see 100% scalling with second GPU.

http://www.hardocp.com/article/2012/12/05/far_cry_3_video_card_performance_preview/3#.UXGB17VJM1I

You can go through this list and you will see that every game scales almost 100% with 7970 crossfire

http://www.guru3d.com/articles_pages/asus_radeon_hd_7970_crossfire_review,1.html

Anno 1404 doesn't scale at all and drags the average down. 

disolitude said: That's a low end card which as bandwitdh issues. Also that's an average of games some of which don't scale well. The ones that scale do at almost 90% efficiency. However you have games like Batman Arkham Asylum which don't scale at all. Essentially console ports usually suck at scalling Games like Farcry 3, Crysis 3, Metro 2033 or any powerhouse games that are properly coded for PCs often see 100% scalling with second GPU. http://www.hardocp.com/article/2012/12/05/far_cry_3_video_card_performance_preview/3#.UXGB17VJM1I

Like I said it all dependent on drivers and the result is not always great... there are a lot of graphics issues too... the SLI is a little better but in the same boat.

And 100% scaling is bullshit or anything is holding the single-GPU... if you have a GPU with A specs and mate a Crossfire with 2x A specs the result will never be 100%.

ethomaz said: disolitude said: That's a low end card which as bandwitdh issues. Also that's an average of games some of which don't scale well. The ones that scale do at almost 90% efficiency. However you have games like Batman Arkham Asylum which don't scale at all. Essentially console ports usually suck at scalling Games like Farcry 3, Crysis 3, Metro 2033 or any powerhouse games that are properly coded for PCs often see 100% scalling with second GPU. http://www.hardocp.com/article/2012/12/05/far_cry_3_video_card_performance_preview/3#.UXGB17VJM1I Like I said it all dependent on drivers and the result is not always great... there are a lot of graphics issues too... the SLI is a little better but in the same boat. And 100% scaling is bullshit or anything is holding the single-GPU... if you have a GPU with A specs and mate a Crossfire with 2x A specs the result will never be 100%.

It's not bullshit actually... And this issues are massively overblown. I've ran SLi and crossfire sincle GTX 260 and microstuttering has been all but eliminated.

You probably have CPU bottlenecks or are using 1080p resolutions when you don't see high scalling. Try playing with 5760x1080 in eyefinity surround. GTX 670 SLi has 90-100% scalling compared to single 670 with almost any game I have played. If you're already getting high frame rates and CPU is a bottleneck, a second card isn't going to help you at all. However having something like 3D vision enabled on 3 screens will see amazing scalling even on 3 cards because you're actually pushing the GPU to the max. 

I can send you benchmarks and screenshots to prove it if you'd like to see what kind of scalling 2 high end GPUs get on 3 monitors.

disolitude said: ethomaz said: disolitude said: That's a low end card which as bandwitdh issues. Also that's an average of games some of which don't scale well. The ones that scale do at almost 90% efficiency. However you have games like Batman Arkham Asylum which don't scale at all. Essentially console ports usually suck at scalling Games like Farcry 3, Crysis 3, Metro 2033 or any powerhouse games that are properly coded for PCs often see 100% scalling with second GPU. http://www.hardocp.com/article/2012/12/05/far_cry_3_video_card_performance_preview/3#.UXGB17VJM1I Like I said it all dependent on drivers and the result is not always great... there are a lot of graphics issues too... the SLI is a little better but in the same boat. And 100% scaling is bullshit or anything is holding the single-GPU... if you have a GPU with A specs and mate a Crossfire with 2x A specs the result will never be 100%. It's not bullshit actually... And this issues are massively overblown. I've ran SLi and crossfire sincle GTX 260 and microstuttering has been all but eliminated. You probably have CPU bottlenecks or are using 1080p resolutions when you don't see high scalling. Try playing with 5760x1080 in eyefinity surround. GTX 670 SLi has 90-100% scalling compared to single 670 with almost any game I have played. If you're already getting high frame rates and CPU is a bottleneck, a second card isn't going to help you at all. However having something like 3D vision enabled on 3 screens will see amazing scalling even on 3 cards because you're actually pushing the GPU to the max.  I can send you benchmarks and screenshots to prove it if you'd like to see what kind of scalling 2 high end GPUs get on 3 monitors.

I don't need proof. But I'm all for glorious pics!

disolitude said: It's not bullshit actually... And this issues are massively overblown. I've ran SLi and crossfire sincle GTX 260 and microstuttering has been all but eliminated. You probably have CPU bottlenecks or are using 1080p resolutions when you don't see high scalling. Try playing with 5760x1080 in eyefinity surround. GTX 670 SLi has 90-100% scalling compared to single 670 with almost any game I have played. If you're already getting high frame rates and CPU is a bottleneck, a second card isn't going to help you at all. However having something like 3D vision enabled on 3 screens will see amazing scalling even on 3 cards because you're actually pushing the GPU to the max.  I can send you benchmarks and screenshots to prove it if you'd like to see what kind of scalling 2 high end GPUs get on 3 monitors.

Something is holding the performance of your single GPU... the SLI/Crossfire scale is not that good... that the point a single GPU with the same specs than a Dual-GPU will always perform better.

If AMD make a single 4096 SPs / 256 TMUS / 64 ROPs GPU to compete with that HD 7990 then you will see what the word moster means.

ethomaz said: disolitude said: It's not bullshit actually... And this issues are massively overblown. I've ran SLi and crossfire sincle GTX 260 and microstuttering has been all but eliminated. You probably have CPU bottlenecks or are using 1080p resolutions when you don't see high scalling. Try playing with 5760x1080 in eyefinity surround. GTX 670 SLi has 90-100% scalling compared to single 670 with almost any game I have played. If you're already getting high frame rates and CPU is a bottleneck, a second card isn't going to help you at all. However having something like 3D vision enabled on 3 screens will see amazing scalling even on 3 cards because you're actually pushing the GPU to the max.  I can send you benchmarks and screenshots to prove it if you'd like to see what kind of scalling 2 high end GPUs get on 3 monitors. Something is holding the performance of your single GPU... the SLI/Crossfire scale is not that good... that the point a single GPU with the same specs than a Dual-GPU will always perform better. If AMD make a single 4096 SPs / 256 TMUS / 64 ROPs GPU to compete with that HD 7990 then you will see what the word moster means.

You're so wrong... So something is holding back my single GPU while dual GPUs are firing on all cylinders? I don't think so.

And I agree that single powerful GPU is better than 2 low end GPUs with similar power. But same money rarely gets you 1 GPU that is as powerful as 2 low end ones. GTX 660 in SLI destroys a single GTX 680 and costs less.

Here are some scaling charts with higher resolutions since you seem to like those. AMD was getting 192% performance in crossfire 2 years ago with their high end cards.

http://www.tomshardware.com/reviews/crossfire-sli-3-way-scaling,2865-10.html

Hynad said: disolitude said: ethomaz said: disolitude said: That's a low end card which as bandwitdh issues. Also that's an average of games some of which don't scale well. The ones that scale do at almost 90% efficiency. However you have games like Batman Arkham Asylum which don't scale at all. Essentially console ports usually suck at scalling Games like Farcry 3, Crysis 3, Metro 2033 or any powerhouse games that are properly coded for PCs often see 100% scalling with second GPU. http://www.hardocp.com/article/2012/12/05/far_cry_3_video_card_performance_preview/3#.UXGB17VJM1I Like I said it all dependent on drivers and the result is not always great... there are a lot of graphics issues too... the SLI is a little better but in the same boat. And 100% scaling is bullshit or anything is holding the single-GPU... if you have a GPU with A specs and mate a Crossfire with 2x A specs the result will never be 100%. It's not bullshit actually... And this issues are massively overblown. I've ran SLi and crossfire sincle GTX 260 and microstuttering has been all but eliminated. You probably have CPU bottlenecks or are using 1080p resolutions when you don't see high scalling. Try playing with 5760x1080 in eyefinity surround. GTX 670 SLi has 90-100% scalling compared to single 670 with almost any game I have played. If you're already getting high frame rates and CPU is a bottleneck, a second card isn't going to help you at all. However having something like 3D vision enabled on 3 screens will see amazing scalling even on 3 cards because you're actually pushing the GPU to the max.  I can send you benchmarks and screenshots to prove it if you'd like to see what kind of scalling 2 high end GPUs get on 3 monitors. I don't need proof. But I'm all for glorious pics!

Here is proof that god exists, and his name is Windows! :)

disolitude said: You're so wrong... So something is holding back my single GPU while dual GPUs are firing on all cylinders? I don't think so. And I agree that single powerful GPU is better than 2 low end GPUs with similar power. But same money rarely gets you 1 GPU that is as powerful as 2 low end ones. GTX 660 in SLI destroys a single GTX 680 and costs less. Here are some scaling charts with higher resolutions since you seem to like those. AMD was getting 192% performance in crossfire 2 years ago with their high end cards. http://www.tomshardware.com/reviews/crossfire-sli-3-way-scaling,2865-10.html

This test just proved my point... five games selection... I can choose five to invert the results.

Crossfire and SLI are all driver dependent and have a lot of issues... some games scales great, other nothing and other in middle... but if you can make the SLI/Crossfire scale near or over 100% then there are something wrong with your single tests... the single-gpu is not using it power... the scale will be always below 100% and AMD/nVidia make drivers optimizations to try to reach more close to 100% possible.

SLI/Crossfire is driver dependent and there are one limitation for the scale... the hardware is only 100% more powerful... so AMD/nVidia do optimizations for specifics games to try to reach more close possible to 100% but that is impossible.

I can believe in close to 90% scaling when the drivers are optimized like a hell for a specific game (there are no optimization for all games in driver level, just for specific games) but if you have a Crossfire/SLI scaling close or over 100% then you have either issues with your single-gpu bench or the quality settings are different (a lot of settings didn't work in Crossfire/SLI mode and I already saw SLI tricks that downgrade the image quality to get better results in performance... that didn't happen with single-gpu).

That was my point about Crossfire / SLI... the only cons is the price/cost of two mid GPU is better than one single high GPU... for everything else (power, consume, noise, etc) the single GPU destroy the Crossfire/SLI... in fact it is better to make a single-gpu with twice the units then make the Crossfire/SLI in a technical term.

ethomaz said: disolitude said: You're so wrong... So something is holding back my single GPU while dual GPUs are firing on all cylinders? I don't think so. And I agree that single powerful GPU is better than 2 low end GPUs with similar power. But same money rarely gets you 1 GPU that is as powerful as 2 low end ones. GTX 660 in SLI destroys a single GTX 680 and costs less. Here are some scaling charts with higher resolutions since you seem to like those. AMD was getting 192% performance in crossfire 2 years ago with their high end cards. http://www.tomshardware.com/reviews/crossfire-sli-3-way-scaling,2865-10.html This test just proved my point... five games selection... I can choose five to invert the results. Crossfire and SLI are all driver dependent and have a lot of issues... some games scales great, other nothing and other in middle... but if you can make the SLI/Crossfire scale near or over 100% then there are something wrong with your single tests... the single-gpu is not using it power... the scale will be always below 100% and AMD/nVidia make drivers optimizations to try to reach more close to 100% possible. SLI/Crossfire is driver dependent and there are one limitation for the scale... the hardware is only 100% more powerful... so AMD/nVidia do optimizations for specifics games to try to reach more close possible to 100% but that is impossible. I can believe in close to 90% scaling when the drivers are optimized like a hell for a specific game (there are no optimization for all games in driver level, just for specific games) but if you have a Crossfire/SLI scaling close or over 100% then you have either issues with your single-gpu bench or the quality settings are different (a lot of settings didn't work in Crossfire/SLI mode and I already saw SLI tricks that downgrade the image quality to get better results in performance... that didn't happen with single-gpu). That was my point about Crossfire / SLI... the only cons is the price/cost of two mid GPU is better than one single high GPU... for everything else (power, consume, noise, etc) the single GPU destroy the Crossfire/SLI... in fact it is better to make a single-gpu with twice the units then make the Crossfire/SLI in a technical term.

No this test didn't prove any of your points but actually proved that if you cherry pick games you can find 1 in 5 that doesn't scale well that brings down the average.

Drivers optimization is necessary yes, but most game engines are properly supported in 2013 and when a game for PC comes out using a new engine, it usually is designed with crossfire and SLI scaling in mind. Name me a high profile PC game that came out in the last 12 months that doesn't scale well. No console ports...

As far as heat, power and all that...sure. I'm not here to argue that SLI/crossfire is better on power and heat, just that it gets up to 100% scalling and 90% on most PC games. 

Also, regarding heat, 2 MSI Twin Frozr 7970s produce much less heat than a single reference 7970 and a power difference comes down to around 20 dollars per year if it runs 24/7. So yeah... Those are not major problems to me in order to get up to 100% boost in performance.

@ethomaz

i can't believe how much time you put in reading all that stuff but when i asked you why you aren't a pc gamer since you care so much about graphics you said something like "i don't want to put so much time in it with all the drivers and stuff". seriously, even 10% of the time you use to read all the tech stuff about drivers would be enough to handle all the problems you can have with drivers as pc gamer! you talk about great graphics all the time but don't take advantage of the systems which have them.

please start to be a pc gamer lol