Guild Wars 2

First: Arenanet if this thread turns out to help people I would sure appreciate a magic conversion of my main character from human to asura.

With that said I have cpu @ 4.8 ghz and a Geforce 680 oc´ed. It should easily be able to handle Gw2 @ 120 fps constantly. The other night when I was in wvw I noticed i had mostly around 25 fps while usually I have 58 (I limit this in afterburner so that thats as it should be). My cpu sat at <20% usage and my gpu was at <50% usage. Being a long time fps player and computer tweaker I wondered. So I ran a script I use to start up fps games. And I turned hpet on and Voila I sit again at 58 fps stabily.

I dont know which one of these fixed this for me. Hpet should be enabled if your motherboard supports it at all. This is also not a magical fix if you have a slow computer. However if you have a decent or good computer, poor fps, and low cpu/gpu usage, there is a very high chance this should work. But I digress.

Heres what to do. Please note that I didnt test extensively what the actual fix was, since I changed 2 things at once, but I put the likely culprits in order. You will also probably be spending the better part of 15 mins or more in order to achieve this depending a bit on how computer savy you are.

Here goes;

1.) Powerprofile. Control Panel > Power Options. Choose “High Performace”. Click Apply. Make sure Core Parking is disabled in the High Performance power profile. It’s a bit tricky to do this but don’t worry I put in the misc section how to do that.

If this works for you, great. Scroll down for how to automate this, unless you always want to be in High performance (which will increase your usage of electricity. We don’t want that we care about the environment).

2.) Turn Hpet on / off. (Hpet = High precision event timer) First check if its enabled or not. Type “bcdedit /enum” in a cmd (Press windows button and type “cmd”). If it says “useplatformclock Yes” its enabled.
Previously it was thought disabling hpet was best for gaming but now it seems its better to enable. So here I would first test to enable it. Make sure its enabled both in windows and in the bios. If you don’t know how to get into bios read your motherboard manual. If you don’t know what motherboard you have go to astra32.com and download the little utility there. Run it and check.
How to enable or disable hpet:

bcdedit /set useplatformclock true (then reboot) is to enable HPET. Enable hpet in bios.
bcdedit /deletevalue useplatformclock (then reboot)is to disable HPET. Disable hpet in bios.
The bcdedit /set commands are both run in the cmd window.

If you have good fps by now good. If you not reverse the hpet process. Ie if you enabled it, disable it. If you disabled it enable it.
Misc:
This is mostly fyi. It´s not really related to the article. Hpet affects dpc latency. Make sure its good with http://www.resplendence.com/latencymon. You should sit at below 200.

I have uploaded my batfiles to box.com.
Please note that you need to make sure yourself that they are correct for you before you use them!
With that said they should be safe and I´m pretty sure should work on Windows 7 64-bit. However please check. In order to do that you need to go to Von Dachs tweak post @ http://forums.electronicarts.co.uk/battlefield-bad-company-2-pc/1273328-performance-tweaks.html. Scroll down to to where it says “Disable Core Parking Windows 7 Tweak may improve performance”. Disable core parking in your powerprofile. Basically it disallows cpus to idle while this powerprofile is used.
The 2 batfiles which are there to change powerprofiles are Highperformance.bat and Balancedpower.bat. If you want to use my gw2.bat to start guild wars 2 you need pskill from http://technet.microsoft.com/en-us/sysinternals/ and nvidia inspector from http://downloads.guru3d.com/NVIDIA-Inspector-1.9.6.6-download-2612.html. Again, please be careful before using this. Only use it if you know what you are doing.

Lastly forgive for this thread is being slightly poorly formatted. I also don’t supply all links directly in it. My health is poor and my concentration wavers. If this gets good response and someone wants to format it please do so and contact me on the forums. Please also let us know in comments if this works for you.

Thank for posting this information, I’m sure some people will benefit from these tips. The HPET trick is a good one. Most current bios support it (there’s usually an option for it in the ACPI menu), and even if they have it disabled with no option to turn it on, windows will use LAPICs, instead of relying on the default TSC counters (you can use WinTimerTester to check the current clock rate precision)

Thanks for the tips! I’m tempted to check it out even though I’m content overall, sieges do dip to 18fps at times… however my system is starting to be a bit dated.

When you say 58 fps in wvw, are you talking during sieges or just an overall average?

It will be interesting to see if this helps the other users with monster systems but running poorly!

Yes I didnt see any fps drops at all. If you want to much around with dpc you could also try a little utility called timerresolution which increases windows timer resolution and possibly lowering dpc.

Swedemon (hej : if you have low cpu/gpu-usage I definately recommend it.

Guess my dreams of becoming an asura are shattered. Now I know what it feels like to be a woman trapped in a mans body.

Conditioned.2467:

Guess my dreams of becoming an asura are shattered. Now I know what it feels like to be a woman trapped in a mans body.

I’m your saviour! Are you ready ? Here it is: Roll another char!
There you go, revere me forever. :-)
(thanks for the tips btw)

“And I turned hpet on and Voila I sit again at 58 fps stabily.”

Could u reproduce this effect, by disabling Hpet and gaining lower FPS again and than turning it back on again for the gain?

Can u maybe fraps a video to show the steps and the results?

I’m a bit confused, since i’m a programmer and i know what the HPET timer is used for and what it does and i see no reason at all, why it would effect gaming FPS. So either there must be a “crazy” software bug involved or your claim is simply wrong and the FPS gain is related to some other factors.

AndyPandy I have already adressed your question. I did specifically mention 2 things I did. I rated the likelyhood of which one was the reason for my performance gain. The reason I added the hpet solution in here is cause it has been reproduced by 2 others in another, more technical forum.

Please make a video so we can know if you can reproduce this result.

AndyPandy.3471:

I’m a bit confused, since i’m a programmer and i know what the HPET timer is used for and what it does and i see no reason at all, why it would effect gaming FPS. So either there must be a “crazy” software bug involved or your claim is simply wrong and the FPS gain is related to some other factors.

When you are using the platformclock=true, windows uses only HPET for the whole system, disabling individual CPU per-core TSCs. This means there is no counter de-synchronization related problems, nor windows will need to periodically re-synchronize them. Also, HPET operates at a higher frequency than any other timer, which means it’ll always be the highest precision timer in the system.

https://en.wikipedia.org/wiki/Time_Stamp_Counter
The time stamp counter has, until recently, been an excellent high-resolution, low-overhead way of getting CPU timing information. With the advent of multi-core/hyperthreaded CPUs, systems with multiple CPUs, and “hibernating” operating systems, the TSC cannot be relied on to provide accurate results — unless great care is taken to correct the possible flaws: rate of tick and whether all cores (processors) have identical values in their time-keeping registers. There is no promise that the timestamp counters of multiple CPUs on a single motherboard will be synchronized. In such cases, programmers can only get reliable results by locking their code to a single CPU. Even then, the CPU speed may change due to power-saving measures taken by the OS or BIOS, or the system may be hibernated and later resumed (resetting the time stamp counter). In those latter cases, to stay relevant, the counter must be recalibrated periodically (according to the time resolution your application requires).

For one, Guild Wars doesn’t use the mm-timer (it uses RDTSC) and furthermore the potential gain in precision is redundant given the preemptive environment that is Windows/Mac.

Based on my own analysis, the performance issues with the game are simply due to poor design; no amount of hardware you throw at it, tweaking that you do -is going to compensate for this.

For example, even someone with very little technical knowledge would understand that disk access is an expensive operation — Guild Wars incurs, based on several sample machines, roughly 60 page-faults/s and each page-fault results in disk access which typically have a latency of around 2ms. This alone is having a severe impact on the game’s performance [as is easily demonstrated using a profiler]

Why would GW be page faulting when it uses so little RAM? Just curious. I have 16 GB of RAM on my machine, and task manager claims only 2 to 3 Gig is in use. I know windows is a little whacky about how the page file works, but it’s odd if it’s page faulting that often with sufficient physical RAM.

Dricoust: Thanks for your input. I think the performance problems are probably both to poor design and poor implementation.

However, still awaiting for others to show reproducability, I would claim there are tweaks you can do to improve performance. It still not performing as well as it should beon my machine but its somewhat acceptable now.

Dricoust.4592:

For one, Guild Wars doesn’t use the mm-timer (it uses RDTSC) and furthermore the potential gain in precision is redundant given the preemptive environment that is Windows/Mac.

Unless you explicitly enable platformclock=true in the BCD, windows will use a combination of (RD)TSC+HPET.

Windows only disables per core TSCs when platformclock is set on the BCD, reverting to LAPICs (on Windows 7, because Win 8 is a different story it seems…) when HPET is off or using HPET when it is on in the ACPI.

Anyhow, TSCs are a real implementation mess, glad there is a way to be rid of them.

VirtualBs: I´m pretty sure thats default though.

It’s not, you have to set it manually using bcdedit. It’s easy to test the differente: Using WinTimerTester, the QueryPerformanceFrequency will be:

• TSC+HPET = ~3.8Mhz
• HPET only = ~14.3MHz

I checked in two different installs of win 7 and its enabled per default in this version.

Strange… All my work PCs and home PCs had it disabled… I had to use the bcdedit /set useplatformclock true trick to enable it.

Why would GW be page faulting when it uses so little RAM? Just curious. I have 16 GB of RAM on my machine, and task manager claims only 2 to 3 Gig is in use. I know windows is a little whacky about how the page file works, but it’s odd if it’s page faulting that often with sufficient physical RAM.

The first restriction is the process working set size which establishes an upper bound on the total amount of comitted (physically mapped) memory. The working set size is not immutable however by default it’s only a few MB (the actual value is a function of the particular OS and available memory). Secondly, the memory manager aggressively trims the current working set size in an opportunistic fashion in an attempt to keep as much physical memory free at all times incase of a peak in the allocation request stream.

Also, the working set size of GW2 is responsible for those out-of-memory errors you may occasionally receive; I reported it awhile back yet it has yet to be fixed.

Unless you explicitly enable platformclock=true in the BCD, windows will use a combination of (RD)TSC+HPET.

Windows only disables per core TSCs when platformclock is set on the BCD, reverting to LAPICs (on Windows 7, because Win 8 is a different story it seems…) when HPET is off or using HPET when it is on in the ACPI.

Anyhow, TSCs are a real implementation mess, glad there is a way to be rid of them.

At present, QPC still relies entirely on the current processors TSC; the HAL callback is merely a stub. To overcome the associated problems with this implementation usually you would adjust for the time drift by synchronizing with the system timer which is indeed typically derived from the most accurate available source (e.g, a PIT) however even this is further [intentionally] diminished in resolution to prevent excessive interrupt servicing. The GW2 developers do not account for this.

Dricoust.4592:

At present, QPC still relies entirely on the current processors TSC; the HAL callback is merely a stub. To overcome the associated problems with this implementation usually you would adjust for the time drift by synchronizing with the system timer which is indeed typically derived from the most accurate available source (e.g, a PIT) however even this is further [intentionally] diminished in resolution to prevent excessive interrupt servicing. The GW2 developers do not account for this.

Yes, but QPC relies on QPF precision, and with the platformclock flag set, QPF is derived only from the HPET timer. On my system, with HPET on + platformclock on, QPF = 14.31818MHz, which by disabling TSC will disable the need for synchronization. If I leave HPET on, but disable platformclock, QPF = ~3.8MHz.

Yes, but QPC relies on QPF precision, and with the platformclock flag set, QPF is derived only from the HPET timer. On my system, with HPET on + platformclock on, QPF = 14.31818MHz, which by disabling TSC will disable the need for synchronization. If I leave HPET on, but disable platformclock, QPF = ~3.8MHz.

Again, QPC does nothing other than return the current TSC of the current processor as can be discovered with a debugger. This is why you are warned about setting the platformclock flag as it’s only intended for debugging purposes — why this hasn’t been mentioned here I’m not sure.

Dricoust.4592:

Again, QPC does nothing other than return the current TSC of the current processor as can be discovered with a debugger. This is why you are warned about setting the platformclock flag as it’s only intended for debugging purposes — why this hasn’t been mentioned here I’m not sure.

Thanks for the input. Do you know why does the QPF change whether useplatformclock is on or off?

If QPF is only based in the TSC shouldn’t it always be the same, independent of useplatformclock (reflecting timer resolution)?

Thanks for the input. Do you know why does the QPF change whether useplatformclock is on or off?

If QPF is only based in the TSC shouldn’t it always be the same, independent of useplatformclock (reflecting timer resolution)?

It should be yes, however as has been the case for over a decade the implementation is broken (and hence why the mm-timers are the recommended method). I assume the HPT is simply not expected to be used when debugging.

In any case, when using useplatformclock=true, QPF is higher than without it, so that is always a benefit in timer resolution.

I still don’t understand why using a different timer will notable improve game FPS? So while changing the timer u get higher precision and the timer call itself may cost less cycles, but those timers are used like a couple of dozen times per frame only, so there should no measurable FPS gain?

AndyPandy make a video!

Conditioned.2467:

AndyPandy make a video!

A video of what?

Btw “The other night when I was in wvw I noticed”, so basically u can’t reproduce this situation and ofc its not possible that the restarting of GW2 could be the reason for the FPS gain? Did u get the same bad FPS after a relog/reboot or did u directly jump to the conclusion of your HPET script?

Is your “test setup” in any way reproduce-able, if so how? What happens atm if u enable disable your script on your PC running GW2, does it still affect your GW2 FPS?

I simply wonder that u put all the effort into explaining what HPET is and how u enable it, but u seem to skip the part trying to reproducible validate that it actually effects GW2 FPS?

AndyPandy.3471:

I still don’t understand why using a different timer will notable improve game FPS? So while changing the timer u get higher precision and the timer call itself may cost less cycles, but those timers are used like a couple of dozen times per frame only, so there should no measurable FPS gain?

http://www.nvidia.com/object/timer_function_performance.html

The time required to return from the various timeGetTime(), GetTickCount(), QuerryPerformanceCounter(), and special assembly _emit 0×0F timing methods can vary by a factor of 150.

It all depends on the used function. For all we know, they can be used thousands of time per frame. At 60Hz, each frame is ~16.6ms (1.66×10^-2 s) — most of those timer functions have precisions around 10^-8 to 10^-7. In the worst case (10^-7), they can be called 166.000 times per frame if need be.

AndyPandy.3471:

Conditioned.2467:

AndyPandy make a video!

A video of what?

Btw “The other night when I was in wvw I noticed”, so basically u can’t reproduce this situation and ofc its not possible that the restarting of GW2 could be the reason for the FPS gain? Did u get the same bad FPS after a relog/reboot or did u directly jump to the conclusion of your HPET script?

Is your “test setup” in any way reproduce-able, if so how? What happens atm if u enable disable your script on your PC running GW2, does it still affect your GW2 FPS?

I simply wonder that u put all the effort into explaining what HPET is and how u enable it, but u seem to skip the part trying to reproducible validate that it actually effects GW2 FPS?

No, I simply refuse to spend more time with this because you are too lazy to try it out for yourself.

why not? so you’re saying this HPET thing is all myth and waste of OUR time?

VirtualBS.3165:

It all depends on the used function. For all we know, they can be used thousands of time per frame. At 60Hz, each frame is ~16.6ms (1.66×10^-2 s) — most of those timer functions have precisions around 10^-8 to 10^-7. In the worst case (10^-7), they can be called 166.000 times per frame if need be.

Um u seem to misunderstand what they are talking about, first of timing code is used a dozen of times per frame not " thousands of time". Its also common practice to actually only update your internal game timer only once per frame, so game calls get all the same cached timing result within a frame, thats needed since otherwise strange timing errors would happen.

Than we talk about a change from one function call that takes 3000 cycles and replacing it with a 100 cycle timer call. They simply refer to the problem of profiling correctly, if the timing call itself takes significant time. So if a timer call takes 3000 cycles and u want to time a 100 cycle call, u obviously need to eliminate the 3000 cycles of the timer call itself.

As example a expansive d3d “DrawIndexedPrimitive” and “SetStreamSource” call also takes around 1000-3000 cycles or even up to 10000 cycles. The difference is those calls are actually called 200000-1000000 times per frame on complex scenes.

So reducing a function call that makes up less then 1% of the per frame cycles, even by a factor of 1000 times wont yield any measurable results. As noted its used to correctly profile code or get correct timeouts for network or other timer critical tasks like video playback.

tekwiz.8635: no im saying what i did works for me. if you are too lazy to try it then dont. i never said it was a myth. I already spent 15 mins writing this down. i guess its perls for swines (mostly). I dont really consider it worthwhile my time trying to convince you. In fact I consider it the height of arrogance even requesting it.

@AndyPandy.3471, thanks for raising some significant issues.

Yes, you are correct about the function call time, it is mostly irrelevant (in the order of ~0.72uS for QPC with HPET on and ~0.0362uS with TSC). Function precision is however another story. QPC has a precision of ~0.0698uS with HPET on and ~0.427uS with TSC.

In the end any should make a reliable counter, if these were the only aspects to it. However, HPET responses are more immune to several aspects where TSC fails miserably. If you do a google search for useplatformclock usepmtimer, there are several topics on this issue.

@VirtualBS.3165 Sure thats all true, the HPET is the superior method to time things in applications period. However we are discussing if a more precise timer can have any FPS impact at all in a game engine and here the answer is no, since it don’t effect any performance relevant game engine subsystem.

The OP uses the common excuse, “Well it worked for me, so it might not work for everyone and u are all arrogant requesting proof.”

The problem i have with any of those “magic” FPS fixes is simply the magnitude, normally those differences are due to bad test setups in largely unstable environments or simple total overstated.

Its like saying: “Well if u use those engine additives u will get 300 more hourse-power!”
The OP did not even respond to the simple request if he could reliable reproduce this “phenomenon” of his, by enabling/disabling the tweak.

The problem is once u “spoke out” at whatever standpoint or claim, u will hardly ever review or take it back, if there is a chance u might be wrong. Thats basic human psychology, so we will probably not see any actual additional tests from the OP on this.

I have given you an opportunity to try it out for yourself, and instead you are here whining and b*******.

AndyPandy.3471:

@VirtualBS.3165 Sure thats all true, the HPET is the superior method to time things in applications period. However we are discussing if a more precise timer can have any FPS impact at all in a game engine and here the answer is no, since it don’t effect any performance relevant game engine subsystem.

Counter precision can affect a lot of game engine subsystems. Try to do physics on wrong counters!

AndyPandy.3471:

“And I turned hpet on and Voila I sit again at 58 fps stabily.”

Could u reproduce this effect, by disabling Hpet and gaining lower FPS again and than turning it back on again for the gain?

Can u maybe fraps a video to show the steps and the results?

I’m a bit confused, since i’m a programmer and i know what the HPET timer is used for and what it does and i see no reason at all, why it would effect gaming FPS. So either there must be a “crazy” software bug involved or your claim is simply wrong and the FPS gain is related to some other factors.

Hpet affects FPS and HID’s. With it turned on a lot of people have a bigger delay on their mouse control (some are less sensitive to this) it usually makes for a little more stable framerate but lower overall.

Hpet has been giving these issues for quite a few years now. For instance when playing ‘WAR’ the big tip always was to turn it off.

VirtualBS.3165:

AndyPandy.3471:

@VirtualBS.3165 Sure thats all true, the HPET is the superior method to time things in applications period. However we are discussing if a more precise timer can have any FPS impact at all in a game engine and here the answer is no, since it don’t effect any performance relevant game engine subsystem.

Counter precision can affect a lot of game engine subsystems. Try to do physics on wrong counters!

This is exactly where the problem lies , with that and collision detections.

Conditioned.2467:

I have given you an opportunity to try it out for yourself, and instead you are here whining and b*******.

Try to ignore it. I noticed he has a tendency to ask others to test and vid/screencap things FOR him and W&B why it hasn’t been done already and complain if people answer him without video/graphic proof.

Like here: https://forum-en.gw2archive.eu/forum/support/tech/Lucid-Virtu-MVP-and-GW2-HD3000-vs-HD4000/first#post312691

I think you made a great effort in your post. Most have been posted already but it is always good to see it in one post and that people are reminded of it or have a fresh chance of learning of it.

Smackjack.5071:

Hpet affects FPS and HID’s. With it turned on a lot of people have a bigger delay on their mouse control (some are less sensitive to this) it usually makes for a little more stable framerate but lower overall.

Hpet has been giving these issues for quite a few years now. For instance when playing ‘WAR’ the big tip always was to turn it off.

As far as I know, there are 4 possible combinations for timer usage:

• HPET on in BIOS and useplatformclock=yes in windows BCD — Windows 7 will use mainly HPET for timers
• HPET on in BIOS and no useplatformclock in windows BCD — Windows 7 will use HPET+TSCs for timers
• HPET off in BIOS and useplatformclock=yes in windows BCD — Windows 7 will use LAPICs for timers
• HPET off in BIOS and no useplatformclock in windows BCD — Windows 7 will use LAPICs+TSCs for timers

Which of the 2 first ones with HPET on were you referring to, for the slowdowns?

Good sir my hat is off, this seems to have fixed it i always seem to have the stuttering issue after 2 waypoint jumps. Issue is non existent , i even had this random occurence in PS2 bete i have yet to check it again.

Warruz.8096:

Good sir my hat is off, this seems to have fixed it i always seem to have the stuttering issue after 2 waypoint jumps. Issue is non existent , i even had this random occurence in PS2 bete i have yet to check it again.

You are very welcome

Smackjack.5071:

Conditioned.2467:

I have given you an opportunity to try it out for yourself, and instead you are here whining and b*******.

Try to ignore it. I noticed he has a tendency to ask others to test and vid/screencap things FOR him and W&B why it hasn’t been done already and complain if people answer him without video/graphic proof.

Like here: https://forum-en.gw2archive.eu/forum/support/tech/Lucid-Virtu-MVP-and-GW2-HD3000-vs-HD4000/first#post312691

I think you made a great effort in your post. Most have been posted already but it is always good to see it in one post and that people are reminded of it or have a fresh chance of learning of it.

Thanks smack. I noticed (after having posted) the disable core parking has been mentioned. I havent seen a post about hpet though. I just changed 2 things, got a result and posted it for others to use.

Conditioned.2467:

Warruz.8096:

Good sir my hat is off, this seems to have fixed it i always seem to have the stuttering issue after 2 waypoint jumps. Issue is non existent , i even had this random occurence in PS2 bete i have yet to check it again.

You are very welcome

Smackjack.5071:

Conditioned.2467:

I have given you an opportunity to try it out for yourself, and instead you are here whining and b*******.

Try to ignore it. I noticed he has a tendency to ask others to test and vid/screencap things FOR him and W&B why it hasn’t been done already and complain if people answer him without video/graphic proof.

Like here: https://forum-en.gw2archive.eu/forum/support/tech/Lucid-Virtu-MVP-and-GW2-HD3000-vs-HD4000/first#post312691

I think you made a great effort in your post. Most have been posted already but it is always good to see it in one post and that people are reminded of it or have a fresh chance of learning of it.

Thanks smack. I noticed (after having posted) the disable core parking has been mentioned. I havent seen a post about hpet though. I just changed 2 things, got a result and posted it for others to use.

Your welcome

Its ok it doesn’t matter your post was useful to players so that is all that counts.

just to extend this finding, i researched it a little before i went through with it and it seems to mainly have an effect on AMD systems. Not to say intel systems cant also benefit from this it seems they have run into more issues when attempting the fix.

Im running a 1100T @ 4.0

amd athlon 7550 dual-core processor 2.50ghz any chance it could work? Thanks

Warruz: Im running a sb (intel).

Lani, yes there is a chance it could work.

Just wanted to leave a note: Thanks for the tip.
My machine is kinda oldish and these adjustments seem to have given me a smallish boost to fps.

Disabled core packing without success. My motherboard doesn’t support Hpet, at least i can’t find it.

If anyone know a guide how to unlock amd athlon tm 7750 dual-core 2.50 GHz i can’t find one. I lost hope already and probably gonne buy a new computer because it sickens me.

Halvaard.1034:

Just wanted to leave a note: Thanks for the tip.
My machine is kinda oldish and these adjustments seem to have given me a smallish boost to fps.

Your welcome. How much fps gain did you get?

Lani.1429:

Disabled core packing without success. My motherboard doesn’t support Hpet, at least i can’t find it.

If anyone know a guide how to unlock amd athlon tm 7750 dual-core 2.50 GHz i can’t find one. I lost hope already and probably gonne buy a new computer because it sickens me.

Yea hpet seems to only be in newer bioses, say ~18-24 months and newer.

Any PC newer than 2005 will almost surely have HPET, even if the BIOS doesn’t show the option to turn it on/off. It’s part of the ACPI spec.

Thanks for this – I’ll have to check it out when I get home. I’ve been having some issues since upgrading from XP 32bit to 7 64bit. Two questions:
Does this have any effect on cpu temperature? Any known issues if the user has an overclocked CPU?