Latest Guild Wars 2 GPU Performance Data

[SomeoneSimple]

SMAA injector wins that by miles lol 🙂
Thats some type of morphological antialiasing right?

Correct, the post-process portion of SMAA is basically Jimenez MLAA (iirc) with better edge-detection and some additional tweaks so it can handle anti-aliased edges better.

 

[TakeNoPrisoners]

I don't know why people think that when a quad core shows %50 usage but the task manager shows each core being used that the game is actually quad core capable.

It isn't quad core capable until it can use all 100% of a quad core CPU. Windows moves processes around to different cores which makes games appear to use all four cores.

The way to force a game to be CPU limited is by disabling Vsync and lowering all the graphics settings to the minimum while running the game windowed with a low resolution with task manager open beside it.

 

[Crysto]

CPU tasks certainly still seem to be the bottleneck, although the performance overall has been excellent (*mostly* sustaining a vSync'd 60fps at 2560x1440 with maximum settings - just render sampling down from Supersampling to Native) on a 2600k@4.9 & 680 Lightning.

Have seen CPU usage jump up to a balanced threading of 70%+ when in certain areas of the game (often brought on by changing direction quickly), however there does appear to often be one thread being pinned up around 75% in populated areas (usually results in lower GPU utilisation and a framerate drop).

 

[Grooveriding]

Nvidia needs to get it together on the driver front for this game. Getting some really annoying issues I have had in the past with games before nvidia updated drivers.

My cards will randomly downclock the core/memory, resulting in horrible framerates and I need to alt-tab and bring the game back to correct it. Sometimes this doesn't work and I have to fully quit and reload the game. Loading into different areas seems to bring it on most of the time, other times though it just happens randomly.

SLI usage is also really low. I get FPS dips into the 40s with GPU usage showing as 40-60% across both cards.

Need a good driver for this game badly :twisted:

 

[Bryf50]

I must be having some underlying performance problem. I'm running the game on the rig in my sig with a gtx 680@1200/3300 and it's struggling in the 30-40fps range at highest setting and just 1920x1080. Just looking at that early benchmark graph a gtx 680 was barely dropping below 60fps minimum. I'm running the latest release drivers but none of the beta drivers seemed to mention GW2.

 

[Akantus]

I must be having some underlying performance problem. I'm running the game on the rig in my sig with a gtx 680@1200/3300 and it's struggling in the 30-40fps range at highest setting and just 1920x1080. Just looking at that early benchmark graph a gtx 680 was barely dropping below 60fps minimum. I'm running the latest release drivers but none of the beta drivers seemed to mention GW2.

I think, the last beta drivers (>month old at this point) should be used for GW2, at least I read it somewhere.
I'm using them and have zero issues with my OCed 460 at 1920*1200 maxed with just shadows dropped down one or two steps.

 

[Haserath]

Looks like toms has a review just for GW2

www.tomshardware.com/reviews/guild-wars-2-performance-benchmark,3268.html

 

[Arkadrel]

Looks like toms has a review just for GW2

www.tomshardware.com/reviews/guild-wars-2-performance-benchmark,3268.html

Looks like:
a Phenom II @4.0ghz = ~55+ fps
a i5-2500k@4.0 ghz = ~69 fps

A phenom II @4.0ghz should be faster than a bulldozer@4.0ghz.
So this isnt a game where a phenom II will get crushed compaired to a Intel CPU.

A 20%+ fps differnce though is still noticeable.
Time to get your CPU overclocked if it isnt already lol, seems like it likes high clock speeds.


Highest ingame graphic options:

Toms also finds More Cores/Threads seem to help performance, even upto 8+.

 

[ShintaiDK]

CPU tasks certainly still seem to be the bottleneck, although the performance overall has been excellent (*mostly* sustaining a vSync'd 60fps at 2560x1440 with maximum settings - just render sampling down from Supersampling to Native) on a 2600k@4.9 & 680 Lightning.

Have seen CPU usage jump up to a balanced threading of 70%+ when in certain areas of the game (often brought on by changing direction quickly), however there does appear to often be one thread being pinned up around 75% in populated areas (usually results in lower GPU utilisation and a framerate drop).

You either got an i7 with HT or an FX. And that screws data since all cores cant perform 100%.

 

[ShintaiDK]

I don't know why people think that when a quad core shows %50 usage but the task manager shows each core being used that the game is actually quad core capable.

It isn't quad core capable until it can use all 100% of a quad core CPU. Windows moves processes around to different cores which makes games appear to use all four cores.

The way to force a game to be CPU limited is by disabling Vsync and lowering all the graphics settings to the minimum while running the game windowed with a low resolution with task manager open beside it.

Usually I would say 60%+ on a real quadcore (not FX4xxx) before you can as a user say its more than 2 threads. Until then it just jumps around as we see.

 

[xboxist]

Was getting 10-20 FPS with my 2x 5770 in xfire. When I disabled it I went up to 30-50.

Using 12.8 catalyst and latest CAP file. /shrug

 

[Despoiler]

Was getting 10-20 FPS with my 2x 5770 in xfire. When I disabled it I went up to 30-50.

Using 12.8 catalyst and latest CAP file. /shrug

12.8 CAP1 will be out soon. It has the GW2 profile in it.

 

[ShintaiDK]

Since 304.79 was recommended for GW2 by ArenaNet. I guess 306.02 is even better.

 

[xboxist]

12.8 CAP1 will be out soon. It has the GW2 profile in it.

Hot info - thanks.

Do you have any idea if it's going to be "tomorrow soon" or more like "two weeks soon" - ?

 

[Grooveriding]

Looks like:
a Phenom II @4.0ghz = ~55+ fps
a i5-2500k@4.0 ghz = ~69 fps

A phenom II @4.0ghz should be faster than a bulldozer@4.0ghz.
So this isnt a game where a phenom II will get crushed compaired to a Intel CPU.

A 20%+ fps differnce though is still noticeable.
Time to get your CPU overclocked if it isnt already lol, seems like it likes high clock speeds.


Highest ingame graphic options:

Toms also finds More Cores/Threads seem to help performance, even upto 8+.

Something is wrong with Tom's results as they just don't make sense. Never put much stock in any of their reviews though.

The 670 is barely faster than the 580 there, 14% or so faster. Also, the results for the 670 and 7970 are practically identical, which doesn't match up to real world performance in other games, normally would make you suspect a CPU bottleneck, but his CPU tests show no bottleneck...

I know from my experience nvidia has some major driver work to do for this game. SLI scaling is terrible, performance is all over the place and not consistent at all. There is no mention of any GW2 improvements in the 306.02 driver they released today as well :|

 

[railven]

Why is there no GTX 680 or even more importantly now the GTX 660 in that chart?

I can sort of vouch for what groove said, girlfriend has a GTX 680 and the few times she wanted to show me something, she had stuttering issues. When I asked about it she said it does it here and there.

Her system is like diseased with shit, so I wouldn't take her gaming experience more than just her system throw throwing gang signs, but on my end and my system - it was smooth as hell, and I was actually impressed since I got into a giant skirmish with like 15+ players, almost as many enemies and I don't think I dropped under 50 FPS.

Regardless of performance, GW2 is an insanely awesome game.

EDIT: Just saw they used:
"Catalyst 12.7 Beta, Nvidia 301.42 WHQL"

Didn't 12.8 Beta have a performance increase for this game? And isn't the latest nvidia beta something like 304 (well prior to the recently released 306s)?

 

[RussianSensation]

Also, the results for the 670 and 7970 are practically identical, which doesn't match up to real world performance in other games, normally would make you suspect a CPU bottleneck, but his CPU tests show no bottleneck...|

Actually it does now. That's because since June 22nd due to Cats 12.7 Betas, HD7970 = GTX670 at 1080P and = GTX680 at 1600P. People keep posting BF3 over and over so you probably still had the impression that 670 > 7970, but it's not overall. AMD's driver team has clawed back almost all of the performance deficit where NV led substantially. So it wouldn't be surprising that 670 = 7970 now that AMD has a new driver for GW2 final game. Computerbase's numbers mimic TPU's almost exactly, confirming that GTX670 ~ 7970 at 1080P and 7970 > 670 at 1600P.

 

[Anubis]

CPU tasks certainly still seem to be the bottleneck, although the performance overall has been excellent (*mostly* sustaining a vSync'd 60fps at 2560x1440 with maximum settings - just render sampling down from Supersampling to Native) on a 2600k@4.9 & 680 Lightning.

Have seen CPU usage jump up to a balanced threading of 70%+ when in certain areas of the game (often brought on by changing direction quickly), however there does appear to often be one thread being pinned up around 75% in populated areas (usually results in lower GPU utilisation and a framerate drop).

im running an I7 920 @ 4GHz w/ HT on with a 7950 with the same game settings and getting the same results at 1440p, Everything on high/ultra and sampling on native, vsynced at 60hz without any issues

 

[Meekers]

Toms obviously spent a lot of time testing this, but they used a beta client and (now) old drivers. I know they had to do this to get the article out so fast but it hardly seems worth it.

I hope someone does something similar with the launch client and new drivers from both sides.

 

[lopri]

Could be one of those titles that thread-hopping causes performance drop. May see a better picture by disabling HT and whatever the equivalent of BD.

 

[Meekers]

Could be one of those titles that thread-hopping causes performance drop. May see a better picture by disabling HT and whatever the equivalent of BD.

I have been told from several people that GW2 uses HT to process audio and is a performance increase over having it disabled. Anyone is free to correct me if that is wrong.

 

[RussianSensation]

There is something useful at Tom's though --> Quad-core Intel i5 is the sweet spot for this game CPU wise, dual core i3 / Pentiums will outperform an FX8150.

 

[Riek]

Something is wrong with Tom's results as they just don't make sense. Never put much stock in any of their reviews though.

The 670 is barely faster than the 580 there, 14% or so faster. Also, the results for the 670 and 7970 are practically identical, which doesn't match up to real world performance in other games, normally would make you suspect a CPU bottleneck, but his CPU tests show no bottleneck...

there are serious bottlenecks in that game..

No framerate difference between 3Ghz and 4Ghz on SB and no performance difference between a 7970 @ 1280 or 1920... Basically its neither cpu nor gpu that bottlenecks it at those points.

 

[Riek]

There is something useful at Tom's though --> Quad-core Intel i5 is the sweet spot for this game CPU wise, dual core i3 / Pentiums will outperform an FX8150.

err, you do realize all those cpu's run at 3GHz?

the 8150 will gain close to 30% where the i3 will gain a few %.

HT doesn't seem to make a big impact, whereass AMD module does.. (except that the baseline for AMD is again way to low to be competitive against i5 or anything higher than that.. which includes the i5 S series.).

 

[Danik]

Anyone get a chance to test the 306.02 drivers in Guild Wars 2? I'll be upgrading the drivers tonight to see if there is any difference. I'm also thinking about overclocking my 670 to improve the minimum frames I'm getting.