Latest Guild Wars 2 GPU Performance Data

MrK6

Diamond Member
Aug 9, 2004
4,458
4
81
https://www.guildwars2.com/en/news/bill-freist-talks-optimization-and-performance/

I'm excited that the development team is this focused on performance optimization and is working this closely with the community. I think it shows great promise for the game, among many other things. :thumbsup:

(Not sure if settings were consistent across all cards/accounts as I didn't participate in the beta, but the chart shows mostly everyone should have a good, playable experience).
 

ShintaiDK

Lifer
Apr 22, 2012
20,378
145
106
GF and I used GTX680 and GTX670 in 1920*1200. We had 50-60FPS all the time (Vsync on) with everything maxed. And we used 301.42.

Quite impressive in terms of the upcoming driver tho! 50FPS to 70FPS or 40% increase o_O

The autodetect chart is abit hmm hmm. Since settings can be wastly different. Clearly shown in the HD4000 vs HD4800.
 
Last edited:
Feb 19, 2009
10,457
10
76
Why is the gtx680 only a tiny bit faster than a 580?? Also 560ti vs 580..

Smells like they need more optimizations. Beta is beta afterall.

Edit: Good on AMD for fixing their drivers in this NV game prior to release. Would be interesting if 7970 is faster than 680..
 
Last edited:

ShintaiDK

Lifer
Apr 22, 2012
20,378
145
106
Why is the gtx680 only a tiny bit faster than a 580?? Also 560ti vs 580..

Smells like they need more optimizations. Beta is beta afterall.

They dont use the same settings. So you cant crosscompare. All settings are with auto on. And different GPU means different auto settings.
 

Meekers

Member
Aug 4, 2012
156
1
76
This is why you do not use benchmarks from before optimization started when people were screaming that AMD was going to suck for GW2. Looks like they got everything together before the game was released.

Coupled with the recent price cuts that chart puts AMD in a pretty dominant position for GW2 at the moment. Still need to see if either side is able to make up performance after release with new drivers.
 

ShintaiDK

Lifer
Apr 22, 2012
20,378
145
106
This is why you do not use benchmarks from before optimization started when people were screaming that AMD was going to suck for GW2. Looks like they got everything together before the game was released.

Coupled with the recent price cuts that chart puts AMD in a pretty dominant position for GW2 at the moment. Still need to see if either side is able to make up performance after release with new drivers.

I have to remind you that all is used with auto settings. It can be anything from low settings to all maxed.

So you cant make such a judgement in either way. You cant even compare 2 different GPUs from the same vendor.
 

Meekers

Member
Aug 4, 2012
156
1
76
Well my 4870 which I had when beta started auto detected to all high settings, as did my new 7950. Though neither of them got set to super sampling on auto.

I imagine most of those upper cards are using pretty similar settings.
 

imaheadcase

Diamond Member
May 9, 2005
3,850
7
76
Why is the gtx680 only a tiny bit faster than a 580?? Also 560ti vs 580..

Smells like they need more optimizations. Beta is beta afterall.

Edit: Good on AMD for fixing their drivers in this NV game prior to release. Would be interesting if 7970 is faster than 680..

Like others said its auto settings. I would like to see SLI comparison in this game between the two.

This is the most polished MMO I ever saw fyi.
 

Meekers

Member
Aug 4, 2012
156
1
76
Also in the article is this graph:

Framerate-Comparison-590x289.jpg


A good question would be if the performance graph includes these outdated drivers and if there is a similar graph for older AMD drivers. Still a lot of information to collect after release.
 

ShintaiDK

Lifer
Apr 22, 2012
20,378
145
106
The nVidia graph is most likely 301.42 WHQL vs 304.79 BETA. Thats also what the client recommends when you start it.
 

Arkadrel

Diamond Member
Oct 19, 2010
3,681
2
0
They dont use the same settings. So you cant crosscompare. All settings are with auto on. And different GPU means different auto settings.

Why would a 7970 and a 680 run differnt "auto detected" settings though?
Im gonna go out on a limb and say their probably both useing the highest settings of everything.

It looks like the 7970 is slightly ahead of the 680.
Like 76 vs 72 fps.


This is why you do not use benchmarks from before optimization started when people were screaming that AMD was going to suck for GW2. Looks like they got everything together before the game was released.

Yeah AMD probably realised that GW was a big title and looking bad in it wouldnt be a good idea :p
 

ShintaiDK

Lifer
Apr 22, 2012
20,378
145
106
Why would a 7970 and a 680 run differnt "auto detected" settings though?
Im gonna go out on a limb and say their probably both useing the highest settings of everything.

It looks like the 7970 is slightly ahead of the 680.
Like 76 vs 72 fps.




Yeah AMD probably realised that GW was a big title and looking bad in it wouldnt be a good idea :p

If I recall right, the GTX680 and GTX670 here used FXAA on auto.

And if meekers is right....

Plus you see the 560ti basicly perform like a 660ti would. But we all know that wont be true.

I´m not saying anything for or against as such. Without proper information its just not very useful.
 
Last edited:

Meekers

Member
Aug 4, 2012
156
1
76
My 7950 had FXAA selected on auto. Do not remember if my 4870 picked it up on auto or if I selected it though.
 

WhoBeDaPlaya

Diamond Member
Sep 15, 2000
7,414
402
126
Hope it performs a lot better. One of the earlier beta weekends got pretty choppy for me during the events with lots of players.
Pretty sure it's not a CPU limitation (i7 920 @ 4.1GHz) - was using a 6950 2GB @ 6970, 1080p.
 
Feb 19, 2009
10,457
10
76
Assume their chart for the NV results alone would mean the same settings (who knows), it suggests some kind of CPU bottleneck is happening as they are all so close to each other, when we all know the 680 is heaps faster.

BUT, as MMOs are well known for being CPU limited, its not surprising at all.
 

Arkadrel

Diamond Member
Oct 19, 2010
3,681
2
0
Last edited:

flopper

Senior member
Dec 16, 2005
739
19
76
Feb 19, 2009
10,457
10
76
I smell a Nvidia put an ironbar in the wheel (driver) for amd cards when detected drop fps rate in release version. No way nvidia accept amd has better fps in their own game at release.

I dont know how they would cripple it, since even the crazy tessellation in Crysis 2 isn't working out vs GCN. They can throw any DX11 feature at it and GCN will chew it for breakfast.

GW2's not using physx either.
 

RussianSensation

Elite Member
Sep 5, 2003
19,458
765
126
Edit: Good on AMD for fixing their drivers in this NV game prior to release. Would be interesting if 7970 is faster than 680..

:thumbsup: Ya, looks like AMD fixed their awful performance. Their driver team has been pulling wonders off this generation, reclaiming Dirt 3, Skyrim, Batman AC and clawing back most of the performance nV had in Crysis 2 / BF3. At the same time NV hasn't done anything to fix Anno 2070, Bulletstorm, Serious Sam 3, Alan Wake, Metro 2033, Sniper Elite, Dirt Showdown. Gotta give credit to AMD's driver team as being the leading team this year imo. They are fixing everything under the sun.

They went from this:
gw2%20vhq%201920x1080.png


To this:
Average-Framerate-by-GPU-590x781.jpg


Looks like this game will run smoothly now on even budget GPUs after GW2 optimizations and latest drivers. Cross-fire scaling for 6990 doesn't seem to be working though.
 
Last edited:

MrK6

Diamond Member
Aug 9, 2004
4,458
4
81
To play devil's advocate, are the framerates better because auto-detect gimps settings on AMD hardware due to previous performance? This is just comparing the two graphs when RussianSensation posted them together. I don't know, but I'll find out this weekend :p.
 

VisceralM

Member
Feb 1, 2005
92
0
0
Surprising, and a nice boost for AMD, especially if it's true Nvidia has access to the game while ATI has to wait until launch. Be interesting to see if a new Nvidia driver in a day or so does anything to improve performance.