Confusing Benchmark results...

[railven]

So last night I was taking a peak at some GPUs, and then I saw something that sort of baffled me.

Using the Techpowerup review for the 980 Matrix, I saw the following odd results.
http://www.techpowerup.com/reviews/ASUS/GTX_980_Matrix/25.html

WoW Nvidia performance get's crushed across the board when going from 1440p to 2160p.

The Radeon's get barely touched. I'm wondering if this review was done AFTER 6.2 dropped which brought some NV specific lightning options.

If they just set the slider to Ultra for each the tests, I would assume for the AMD hardware it used the standard OC, but for the NV cards it slotted into the NV specific OC. Let me try to get a link:

http://us.battle.net/wow/en/blog/17944910/engineer’s-workshop-anti-aliasing-lighting-2-12-2015

I can't verify if that is the correct link with the information (firewall at work), pretty sure it is.

I wish Techpowerup gave a little more input for their results rather than just a data dump.

Actually, thinking about it the performance drop isn't out of whack going from the other resolutions to the next up. Just from 1440p to 2160p.

WoW has always favored Nvidia.

 

[Noctifer616]

WoW has always favored Nvidia.

Do you actually have relevant data that supports that, as in raids and battlegrounds with 20 or more players?

I don't really consider benchmarks in the open world or cities as relevant when it comes to WoW.

Also the review doesn't mention what AA options they are using. Graphics settings don't actually modify AA settings, it's a separate option. Bad review all in all.

 

[railven]

Do you actually have relevant data that supports that, as in raids and battlegrounds with 20 or more players?

I don't really consider benchmarks in the open world or cities as relevant when it comes to WoW.

Huh? WoW was a TWIMTBP title. It has Nvidia specific code. And, various reviews have always favored Nvidia by a good portion. I was actually surprised that the Radeon's beat the GeForces at 2160p.

Take it for what it is worth. As both a Radeon and Geforce users, my personal experience has supported the review sites.

 

[PhIlLy ChEeSe]

I never played it but isn't WOW not as demanding on GPU'S?

 

[railven]

I never played it but isn't WOW not as demanding on GPU'S?

It can be. Even my GTX 780 @ 1150/6800 bogs down to the 40-50s during intense raid fights.

And I have to dig up the other post, because turning on SSAAx2+CMAA @1440p drops me into the upper 30s. Someone said they do SSAx2+CMAA 60 FPS no issue.

Because of the draw distance some areas can tackle your FPS.

 

[RussianSensation]

W1zzard was asked if that performance test was an error/mistake or some fluke and in the forums/review sub-forum section he responded that No, he retested all the GPUs at "ultra" several times and it is true that at 4K R9 290 series crushes NV in WoW.

This has been consistent for a long time now but you just haven't noticed because you probably weren't paying attention to TPU's WoW benches at 4K.

http://www.techpowerup.com/reviews/Gigabyte/GeForce_GTX_980_G1_Gaming/24.html

Having said that even the Titan manages 82 fps. Regardless, I think TPU's testing of WoW is not advanced/demanding enough because in raids people can drop to 30-50 fps easily.

 

[railven]

W1zzard was asked if that performance test was an error/mistake or some fluke and in the forums/review sub-forum section he responded that No, he retested all the GPUs at "ultra" several times and it is true that at 4K R9 290 series crushes NV in WoW.

This has been consistent for a long time now but you just haven't noticed because you probably weren't paying attention to TPU's WoW benches at 4K.

<snip>

Having said that even the Titan manages 82 fps. Regardless, I think TPU's testing of WoW is not advanced/demanding enough because in raids people can drop to 30-50 fps easily.

Yeah, I definitely don't focus above my current resolution that often (but will be buying a new monitor with my new GPU this summer).

How long has this been happening? Because I don't recall such a difference during my last few purchase windows.

780 Lightning:

I wonder if the AA is the secret sauce. The 7970 takes a visible hit, where as in the recent test, it took no hit.

290X Using 290X Lightning review:

Yeah, the difference isn't as stark as this recent bench mark.

GTX 980 launch:

Something happened at some point, whether with the recent patch or some AMD optimization. But the 290X was barely faster than the 780 at that resolution yet this recent review, the 780 gets BTFO by the 290X in this most recent review.

When did those AMD Omega drivers drop?
10/21/2014 review:

290X and 780 neck and neck.

12/3/2014:

290X BTFO 780

Damn, AMD must have done something in their drivers. Kudos to them. I might pick up a lightly used 290X just to test this out haha.

 

[Head1985]

103% performance increase not bad:biggrin:
Omega magic drivers.

 

[flexy]

Sorry..WHAT baffles you?

The 970/980s only have 3-3.5GB performant memory plus the R290x have crazy memory bandwidth plus "real" 4GB. Sort of logical that it MIGHT perform better at those extreme resolutions where memory and b/w really count. Just my $0.02

(Nope, not a fanboy, in fact I just got a 970. Just stating facts.)

 

[Techhog]

Sorry..WHAT baffles you?

The 970/980s only have 3-3.5GB performant memory plus the R290x have crazy memory bandwidth plus "real" 4GB. Sort of logical that it MIGHT perform better at those extreme resolutions where memory and b/w really count. Just my $0.02

(Nope, not a fanboy, in fact I just got a 970. Just stating facts.)

The 980 has the full 4GB. Only the 970 has that limitation, which is caused by Nvidia being short-sighted when designing cut down Maxwell chips.

 

[Childs]

Assuming I am reading this right, which I may not since I'm about to fall asleep at my desk, the 290x actually increases its frame rate going from 1440p to 2160p. 290 and 295x2 are basically the same? Something must be wrong with the chart.