5970 benchmarks

straxusii · 17 Nov 2009 at 11:57

If the 9.12s do give a ~20% boost and the 5970 gets excellent scaling then it might get to twice the speed of the 295

I think it's unlikely though I'm expecting more like 50% faster

magicroundabout · 17 Nov 2009 at 11:57

because it's lines on a graph at present painted Red not Green so he doesn't believe them

TaKeN · 17 Nov 2009 at 12:04

These review graphs are hilarious, where they create bar charts which are rather overly exagerated making the one card look like its running away bigtime..

RavenXXX2 · 17 Nov 2009 at 12:25

It is running away bigtime in most of them games/benchmarks, 295 just got its ass handed to it big style.

mmj_uk · 17 Nov 2009 at 12:50

As with 5870 previews they're selectively using 8xAA which is more of a reflection of NVidia's cards being poor at it than 5970 being awesome, it's no coincidence that all of the least impressive results are done at 4xAA or below.

wakayoda · 17 Nov 2009 at 12:54

why drown us in eye candy just get to the point

Wilderbeast · 17 Nov 2009 at 13:13

I expect the 1920x1200 benchmarks to be quite a bit closer, the 295 gets progressively worse the higher the res. These benchmarks are all about showing the card in optimal light. I think the shift from 1920x1200 to anything higher for mainstream is going to be a very slow process, even for enthusiasts the switch is very expensive.

Can't wait to see if this card can actually handle some 5760x1200 action

Wilderbeast · 17 Nov 2009 at 13:15

mmj_uk said:
As with 5870 previews they're selectively using 8xAA which is more of a reflection of NVidia's cards being poor at it than 5970 being awesome, it's no coincidence that all of the least impressive results are done at 4xAA or below.

Can anyone actually see the difference between 4x and 8x?

straxusii · 17 Nov 2009 at 13:33

Wilderbeast said:
Can't wait to see if this card can actually handle some 5760x1200 action

Checkout the link, they have some benchmarks of just that

Duran · 17 Nov 2009 at 14:00

well at 1920, 4xaa and who knows what else ati used to show the 5970 at the best case scenario expect the 5970 to be about 20-30% faster overall. quite a disappointment really.

TheRealDeal · 17 Nov 2009 at 14:11

Duran said:
well at 1920, 4xaa and who knows what else ati used to show the 5970 at the best case scenario expect the 5970 to be about 20-30% faster overall. quite a disappointment really.

The best case scenario is 20-30% who are you trying to kid, that will most likely be around the lowest case scenario where crossfire is working properly.

R088ieS86 · 17 Nov 2009 at 14:14

Haha, FAIL!

mmj_uk · 17 Nov 2009 at 14:21

Wilderbeast said:
Can anyone actually see the difference between 4x and 8x?

I doubt it unless they run super low resolutions or study blown up screenshots.

Also what it the Tom Clancy "fail" all about? did they just throw that in there to make NVidia look bad or is there a logical reason?

Fenris · 17 Nov 2009 at 14:43

mmj_uk said:
I doubt it unless they run super low resolutions or study blown up screenshots.

Also what it the Tom Clancy "fail" all about? did they just throw that in there to make NVidia look bad or is there a logical reason?

It means the card could not produce sufficient FPS to be playable.

arc@css · 17 Nov 2009 at 14:44

Bars on graph are somewhat misleading but if they do perform like that then it is a big boost.

D.P. · 17 Nov 2009 at 14:50

What a load of rubbish.

Owenb · 17 Nov 2009 at 15:11

TheRealDeal said:
The best case scenario is 20-30% who are you trying to kid, that will most likely be around the lowest case scenario where crossfire is working properly.

The 295 is some 10% faster than a 5870, so
GTX295 = 100
5870 = 90

Agreed?

When comparing the 5970 to 5870 the core speed is down 17%(lets say 15%). That give 90*.85= 76.5.

On average Xfire delivers 60% more peformance(when it works)- agreed?

76.5*1.6=122.4.

So it's very possible that the stock 5970 will only end up some 22% faster than a GTX295- in Dx9/10.

For the 5970 to truly hammer the GTX295 it would need one of two things to happen. For the 5870 core to become much faster than a 295, or for ATi to make huge improvements to Xfire.

LoadsaMoney · 17 Nov 2009 at 15:18

Rofl at that graph

Mr Paul · 17 Nov 2009 at 15:21

http://www.anandtech.com/mb/showdoc.aspx?i=3649&p=2

TheRealDeal · 17 Nov 2009 at 15:28

Owenb said:
The 295 is some 10% faster than a 5870, so
GTX295 = 100
5870 = 90

Agreed?

When comparing the 5970 to 5870 the core speed is down 17%(lets say 15%). That give 90*.85= 76.5.

On average Xfire delivers 60% more peformance(when it works)- agreed?

76.5*1.6=122.4.

So it's very possible that the stock 5970 will only end up some 22% faster than a GTX295- in Dx9/10.

For the 5970 to truly hammer the GTX295 it would need one of two things to happen. For the 5870 core to become much faster than a 295, or for ATi to make huge improvements to Xfire.

You are forgetting that the new ati drivers are supposed to give a good performance boost, in shift one 5870 is now faster than a gtx295. The other thing you may not be taking into consideration is the gtx295 running out of memory at certain resolutions. In some situations especially at really high resolutions i have seen crossfire giving over 100% boost in speed.