Starcraft AA benchmarks: ATi vs nVidia

[Skurge]

If he was smart he would look for the factory overclocked Point of View "Beast" version review. Its a gtx 460 1g @ 850 core from the factory. Hint. Hint.

http://www.fudzilla.com/graphics/graphics/here-comes-the-beast-gtx-460-running-at-850-mhz

Yeah, if he was smart.

Anyway, is that the 1GB version?

 

[Grooveriding]

Can we get this thread cleaned up. It's been derailed and turned into an advertising venue trying to sell particular video cards rather than discussing the topic.

As usual.

 

[Skurge]

Yeah, I'm partly to blame.

I don't even have the game yet.

 

[happy medium]

After reading the reviews and looking at the pictures and video, is the 40/50 percent hit in performance with AA even worth it visually?
It does not look like it to me. I'm talking about both brands.

 

[0roo0roo]

You can't judge aa from stills, like many gpu effects its far more apparent in motion. the clean render from fsaa increases the apparent resolution by quite a bit.

 

[happy medium]

You can't judge aa from stills, like many gpu effects its far more apparent in motion. the clean render from fsaa increases the apparent resolution by quite a bit.

So your saying its worth the performance hit?

I think both Ati and Nvidia need to tweak there drivers a bit for this game with AA enabled.
Comparing no aa to aa benchmarks, the no aa benchmarks seem so much more stable. The aa benches are all over the place. SLi and crossfire are broken also.

here is some benches with no aa ,ultra settings @ 1900x1080.

 

[SlowSpyder]

The difference between the 460 and the 5850 is just a few fps. Even closer when you consider minimum frame rate. (The 460 actually wins a benchmark there).


The 5850 needs to drop down to $199 to be competitive with the 460.

The 460 is a great bang for the buck card. For this game, with AA, the 460 is a better choice than the 5850. The 5850 is faster overall, but I don't think it's advantage is enough to justify it's price difference.

But, the pretty much the same can be said about the 480 and the 5870. The 5870 will give you an identical experience in this game as the more expensive, louder, more power hungry 480.

Even the premium 5970 actually delivers more frames per dollar than the 480. You don't often see that from ultra highend cards.

I guess what you could say is that, going strictly by this bench, there are some winners and losers.

Winner: GTX460 - Almost as fast as the more expensive 5850.
Loser 5850 - Too much of a premium over the 460 for it's very moderate performance edge.

Winner - 5870 - Very nearly as fast as the more expensive, hotter, more power hunger 480.
Loser - GTX480 - All those negatives for not tangible difference over the 5870.

Winner - 5970 - Nothing else is even close.
Loser - Nvidia? Despite coming out months later they can't match this card. Of course when we start looking at multiple cards things get thrown up on the air. But for a single card solution, the 5970 is the king by a decent margin.

 

[Grooveriding]

The 460 is a great bang for the buck card. For this game, with AA, the 460 is a better choice than the 5850. The 5850 is faster overall, but I don't think it's advantage is enough to justify it's price difference.

But, the pretty much the same can be said about the 480 and the 5870. The 5870 will give you an identical experience in this game as the more expensive, louder, more power hungry 480.

Even the premium 5970 actually delivers more frames per dollar than the 480. You don't often see that from ultra highend cards.

I guess what you could say is that, going strictly by this bench, there are some winners and losers.

Winner: GTX460 - Almost as fast as the more expensive 5850.
Loser 5850 - Too much of a premium over the 460 for it's very moderate performance edge.

Winner - 5870 - Very nearly as fast as the more expensive, hotter, more power hunger 480.
Loser - GTX480 - All those negatives for not tangible difference over the 5870.

Winner - 5970 - Nothing else is even close.
Loser - Nvidia? Despite coming out months later they can't match this card. Of course when we start looking at multiple cards things get thrown up on the air. But for a single card solution, the 5970 is the king by a decent margin.

I agree with your assessment, but would quantify it with the statement that the 460 and 5850 are only delivering playable frames with AA at 1680x1050.

If you want to go 1920x1200 or higher, or use SSAA you need a 5870, 5970 or 480. Even looks like the 470 is almost falling short with AA at 1920x1200.

 

[happy medium]

This game is more cpu bound then anything IMO.

The only gpu that cant give playable framerates @ 1900x1080 will 4x aa ULTRA settings is the 5830.

even @2500x1600..Ultra settings.

Well either cpu bound or bad sli scaling

 

[3DVagabond]

I have a bit of a problem following exactly which flavor of 460 they are comparing from these posts. Care to pick one? There's a huge difference in performance between brands/models/memory/shaders/clocks. Just reading these threads would make people think you can buy any 460 and get the same performance. Especially when the pricepoint of $200 gets used. At least give links to where the graphs are from. Even if it was posted on a previous page/post I think we should repeat it

 

[happy medium]

I have a bit of a problem following exactly which flavor of 460 they are comparing from these posts. Care to pick one? There's a huge difference in performance between brands/models/memory/shaders/clocks. Just reading these threads would make people think you can buy any 460 and get the same performance. Especially when the pricepoint of $200 gets used. At least give links to where the graphs are from. Even if it was posted on a previous page/post I think we should repeat it

I would think they were using all reference cards.
http://www.pcper.com/article.php?aid=958

If you want a better price point I allready linked a Asus gtx 460 1gb factory overclocked to 800 core for 210$ shipped. That should easily match or beat a 5850 in this game for 75$ cheaper.

http://www.amazon.com/gp/product/B00...namitedata.com

 

[Grooveriding]

I have a bit of a problem following exactly which flavor of 460 they are comparing from these posts. Care to pick one? There's a huge difference in performance between brands/models/memory/shaders/clocks. Just reading these threads would make people think you can buy any 460 and get the same performance. Especially when the pricepoint of $200 gets used. At least give links to where the graphs are from. Even if it was posted on a previous page/post I think we should repeat it

I linked the results of the benchmark that is the subject of this thread, a benchmarking comparison of ATI vs NV in SC2 with AA. They are using a 1GB 460 in the review.

It's an excellent review as they are using a realistic demo to benchmark with, that simulates the actual gameplay of a fully built-up base with lots of units on the screen.

http://www.xbitlabs.com/articles/vid...y_6.html#sect1

Nice to see a review that uses real gameplay conditions.

If you are gaming at 1680x1050 and want AA in Starcraft 2 it looks like you can get by with a $200 card. 1920x1200 and above, you need something more powerful.

 

[happy medium]

The Xbitlabs AA results will be out of date faster than ATI and NVIDIA update their "hacks," which should be very soon (for beta).

Best post in this thread ,I totally agree.

 

[GodisanAtheist]

I can never understand why Xbit Labs insists on leaving Catalyst AI off (and by extension a good number of ATI driver optimizations) when Nvidia seemingly doesn't even give the user the option to turn off their driver optimizations.

Also, where the shit are the 4890/4870/GTX260/GTX250 in the "mainstream" category?

Nevertheless, I gotta love Xbit's reviews. Clean and straightforward, aforementioned niggles excluded.

 

[evolucion8]

Catalyst A.I on improves the performance quite a bit, don't understand why do they turn it off.

 

[ronnn]

So looks like an ati win, which is a surprise for meant to be played game. Unfortunately I am a first person shooter and its all a big whoop.

 

[ronnn]

Best post in this thread ,I totally agree.

The catch of course is many will now start accusing nvidia of being slow off the mark and how they mistreated their customers. At least nv pr had a field day accusing ati of such.

 

[happy medium]

I wanna know why in this pc perspective review the 5870 and gtx 460 are much faster then in the xbit review? Yes, they are using the ultra settings also.

 

[3DVagabond]

I linked the results of the benchmark that is the subject of this thread, a benchmarking comparison of ATI vs NV in SC2 with AA. They are using a 1GB 460 in the review.

It's an excellent review as they are using a realistic demo to benchmark with, that simulates the actual gameplay of a fully built-up base with lots of units on the screen.

Nice to see a review that uses real gameplay conditions.

If you are gaming at 1680x1050 and want AA in Starcraft 2 it looks like you can get by with a $200 card. 1920x1200 and above, you need something more powerful.

Thanks, Grooveriding. I appreciate the provided link. It's just (not you but others) that supplying a graph/chart without the link makes it worthless for comparison. I've seen 20 percent difference between different 460's. I wouldn't want someone to look at a graph from a SOC card and to believe that the $199 reference 768 will perform the same way. There are those that would take one chart for one card from one game at one resolution, present it out of context and proudly proclaim that the 460 is faster than a HD 5*** (pick a number).

 

[Athadeus]

Because they used different replays most likely (with the PCPer one obviously less stressful).

 

[happy medium]

Thanks, Grooveriding. I appreciate the provided link. It's just (not you but others) that supplying a graph/chart without the link makes it worthless for comparison. I've seen 20 percent difference between different 460's. I wouldn't want someone to look at a graph from a SOC card and to believe that the $199 reference 768 will perform the same way. There are those that would take one chart for one card from one game at one resolution, present it out of context and proudly proclaim that the 460 is faster than a HD 5*** (pick a number).

I gave you the link to the other site ^^^^^post #61, the original xbit link is the first post?

Thats why I asked the question in post #68. WHy are they so different?

 

[3DVagabond]

I gave you the link to the other site ^^^^^post #61, the original xbit link is the first post?

Thats why I asked the question in post #68. WHy are they so different?

I'm talking in general here. No point in anyone taking it as directed at them. Graphs/charts need links, that's all.

 

[Sind]

I wanna know why in this pc perspective review the 5870 and gtx 460 are much faster then in the xbit review? Yes, they are using the ultra settings also.

I have a 5870, played the campaign and many others vs ai and coop, ultra, 4x and without AA, never have I seen this card at 1920x1200 pass 90fps even for a sec with no AA(looking at the G15 display with afterburner on it) yet they have an average of 103 for something? Same with AA on, average FPS from my experience so far is no where near 60fps with a 5870 on loaded maps with 4xAA on, usually around 45-50ish.

Something is not right with with those numbers from PCP, the Xbit numbers are much more in line with the performance I get with this card. CPU is at 3.8, card is at 875/1250.

 

[happy medium]

I wanna know why in this pc perspective review the 5870 and gtx 460 are much faster then in the xbit review? Yes, they are using the ultra settings also.

I have a 5870, played the campaign and many others vs ai and coop, ultra, 4x and without AA, never have I seen this card at 1920x1200 pass 90fps even for a sec with no AA(looking at the G15 display with afterburner on it) yet they have an average of 103 for something? Same with AA on, average FPS from my experience so far is no where near 60fps with a 5870 on loaded maps with 4xAA on, usually around 45-50ish.

Something is not right with with those numbers from PCP, the Xbit numbers are much more in line with the performance I get with this card. CPU is at 3.8, card is at 875/1250.

Is there a in game benchmark for this game? Kinda like Crysis? Mabe thats what there using.

 

[Sind]

No benchmark in game so far. Everytime I look at the numbers I laugh out loud. Very misleading and totally inaccurate. That PCP review should be deleted.