Guild Wars 2

perception differs per screen as well fyi, on crappy displays that pad frames i can see how bad 45 is, on low latency displays its impossible to notice.

most (heavier player traffic, events, dungeons and WvWvW) the game suddenly shifts to CPU and leaves the GPU alone? Shouldn’t it be the other way around?

Not if the problem ends up being the inefficiency of D3D9, more dynamic objects on screen means more cpu usage.

try renaming local.dat in the my documents\guild wars 2 folder

are you using an nVidia Optimus laptop? another thread posted indicates its not using the nvidia graphics like it should be.

restart the client

someone else had this as well
https://forum-en.gw2archive.eu/forum/support/tech/60FPS-to-15FPS-in-this-last-patch

well, thats a mighty big oops

keep a close eye on the modem when it happens, particularly the Power / uptime light.

i’ve seen mmo’s cause a bodgy modem-router reset itself because of the load.

your games folder is ntfs compressed, thats why the GW2 setup executable doesn’t see it.

looking at the process monitor activity it actually prevents installatin to a compressed drive or folder, likely for performance reasons.

an issue with ntfs compression is that it causes excessive fragmentation when writing updates, which the guildwars 2 dat file actually keeps itself defragmented by itself fairly well.

open nvidia control panel
search for Guild Wars 2 in the profiles list
select the nvidia performance gpu in the start on device drop down menu.

are you both on desktop pc’s
if his machine starts acting odd like this at your house, you might be on a dirty current and need your house wiring checked.

-edit, saw screenshot-

got a screenshot of the C:\ root with the games folder showing?

no difference at all.

http://msdn.microsoft.com/en-us/library/windows/desktop/aa366912%28v=vs.85%29.aspx

you realise the gw2.exe installs the .dat file to the location it is executed from right?

don’t have to overheat to be running on the fringe of stability. New drivers can drive the hardware harder, in the case of nvidia’s drivers, this has happened more then enough times in the past to keep me far away from factory overclocks that are rather excessive

No, they are shorter then what they were for sure.

Previously it took almost a minute to load from the character select screen into the world, and now its back to around 15-25 seconds (depending on whether i flush the disk cache)

no, i don’t think its that good at all, a minimum of 45 is required atleast for the games i play usually as i otherwise notice it isn’t smooth, If this is a case of compile time optimisations being scaled back in the client, then a post skyliner made elsewhere may hold the key.

those are all the threads currently executing, but i was after the Loaded DLL’s, you can enable this in the Lower pane menu

you got lucky, or a decent brand with quality mosfets :p

the reference design skimped on them and used parts that just barely covered stock clocks.

can’t say i did.

570 users, should rma their cards and get a replacement if they bought it before March 2012.

the VRM is dodgy and cannot sustain the loads guildwars places on the combined GPU and IMC.

iSmack.1768:

@Squall Leonhart – Yes, i’m very ill-informed.

Yes, yes you are, see below.

- GPU utilization is around 80-90% in GW2 for me along with 700-1000MB mem usage on both cards. (@Crawford, i’ll take screenshots when i get home. I use MSI Afterburner since my GPUs are MSI TwinfrozrII GTX 560Ti-s)

GPU Usage alone, is not an indicator for temperature. One thing MSI afterburner does not do, is provide a good indication of Memory Controller usage. The memory controller is a portion of the GPU that also contributes to heat, but often runs at less than 30% usage. Guild Wars 2 is one of the few titles that cause the IMC to be utilised upwards of 80+%. Following what you said in the quote below.

Another question: Do you guys run the game at 2560×1440? I’ve spoken to a few others on /map and they also said they were having heat issues while using the same rez. Is there some kind of memleak / engine has issues (unoptimized) for this rez?

The heat is thus a result of a heavily worked graphics card. You can reduce the load on the IMC by using native sampling instead of super sampling. But this comes at the cost of eye candy.

card should be ok then,
The only possibility for the VGA error beeps is something is hard locking the gpu itself which i already gave some potential reasons for. i don’t see how lack of power can be the cause at this point – unless the cards VRM circuit is failing and that sortof thing isn’t easy to test, you’d have to throw a full 100% load at both cards gpu and memory at the same time.

or, this is another testament to nvidia’s increasingly poor driver support for the 295 cards, which has been getting more and more issues in games as time goes on.

Ban Sidhe.9637:

Squall Leonhart.2075:

overclocking the memory of the system won’t really provide that big an increase in gaming.

the only instruction set that has shown to quickly get bottlenecked by 1600MHz dram so far is AVX, and only when trying to overclock past 5Ghz

When overclocking the i5 3570K, would you recommend doing it solely by multiplier, or FSB as well? And in the case you increase FSB, do you keep the memory linked or do you keep memory unlinked and steady at 1600MHz?

I might play around with this anyways, when I get the hardware, but still I’m interested in other people’s experiences.

You can’t really adjust the FSB on sandybridge+, as the clock generator for all devices which use one are linked to the FSB. The multiplier will usually give you more then enough clock speed.

DxDiag will only display the gpu presented to the system, unless you start it using the run on nvidia performance graphics context menu item

the nvidia gpu is otherwise hidden from the system.

never mind, i just spotted it in the original post

is that the Ti version. GTX 660 and 660Ti use different cores, 660 is a GK106, 660Ti is GK104.

You’re not the first person i’ve seen with this

the problem is your power target is causing the gpu to throttle back invisibly (a feature introduced with the 500 series cards)

with evga precision you should have access to a slider to control the turbo / power target, slide it back a bit to stabilise the core.

Your gpu looks to me like it comes from the vendor pre-overclocked, can you post the name of the vendor and i’ll see if theres a Video bios update available.

Both of these are valid possibilities, extremely valid.

i hope you followed a guide for it don’t want to risk getting it wrong

Did you change the device at the bottom and test the second card?, this test is not sli compatible, it has to do each one individually.

Personal observations aside, mmo’s like GW2 scale almost directly with cpu clock rate when many effects and pc’s are on screen.

Not sure how Anet handled d3d9 multithreading, but it can be a double edged sword even if implemented right.

time better spent on multithreaded D3D11 using feature levels for downlevel support, which unlike 9/10, supports thread safe multithreading.

alot more than you, it seems.

cpu-z and hwmonitor are both from the same group, so yes.

What you do have though, is a graphics card that adapts the fan speed based on the temperature sensor reading, which can throw a hissy if hwmon accesses the sensor at the same time.

Its not my CPU either

Says who?

i get almost constant 30fps (capped to) on a 550ti, i assume your cpu is holding you back, which you failed to disclose

oh, i see what you mean. as in, unlike Gamebryo, the engine isn’t tangled up with the video api so it makes it easier to port to alternative api’s.

when you get a chance, reproduce the issue with the OSD server running and find out what the gpu load is when this slowdown occurs.

that could very well be the case

Ok, that confirms the Shaders are ok.

now give Video Memory Stress Test 1.7.116 a go

The only other thing i can think of, is that you’re running a combination of applications that use Local mutex’s instead of Global, (or was it the other way… can’t recall atm) but anyway, mixing seveal applications that acess the graphics hardware directly can cause it to enter a state that only a full power down can resolve.

So, you shouldn’t mix software like Ntune and aida64, with evga precision, hwmonitor, or GPU-Z (though iirc, atleast aida64, precision and gpu-z use a sharable safe method these days)

actually, i missed out on the part where you said sound keeps working. That definitely sounds like its a mutex crash.

should be fixed now

in game i mean, sub sampled, native or super sampled?

what do you have selected in the sample size menu?

im fairly sure they are using Opengl 3.0 on the mac port

thats probably msi afterburners cpu stats

Can you load up Aida64 or EVGA Precision, enable the OSD Server and take a look at what the clock rates are.

now try running Luxmark

Yes, i read it, and discarded it as completely inconclusive and a waste of the time you could have been testing various system configuration changes.

And actually, i just noticed you have a GTX 570….. I hope its a recent one, otherwise you’ve got the kittened Mosfets to factor in. (570’s have one less than the 580’s and on games that tax the Memory and GPU heavily, this can make the card very unstable.

Of the games you listed, NONE of them work the memory controller as hard as GW2.

if you have a PS2 handy, you can simulate GW2’s load by using PCSX2 with GSDX set to an extremely high internal resolution.

screenshots of all the modules loaded by the GW2.exe in processor explorer would help a bit as well, to rule out any foreign applications injecting dll hooks.

it means your password does not meet the minimum strength recommended

solution, use more symbols and numbers in your password

also

ANet announced that they would be requesting all user’s change passwords over the next few weeks or so. They are doing this as a part of trying to combat account compromises, by forcing players to change passwords from an old password – any password that they detect has been used in an account compromise, used in an attempted account compromise, or used prior to forced change is being blacklisted. Blacklisting password means these passwords can never be used again on any account – which they are obviously hoping will help aid vs account compromises.

compressed air or a static filtered vacuum cleaner on low

Benek, make a screenshot of your video card control panel config