Forged Alliance Forever Forged Alliance Forever Forums 2015-02-27T05:22:36+02:00 /feed.php?f=2&t=9422 2015-02-27T05:22:36+02:00 2015-02-27T05:22:36+02:00 /viewtopic.php?t=9422&p=94863#p94863 <![CDATA[Re: Denial and Error: Lag, console commands and stupidity]]> Statistics: Posted by SeraphimLeftNut — 27 Feb 2015, 05:22


]]>
2015-02-26T15:16:38+02:00 2015-02-26T15:16:38+02:00 /viewtopic.php?t=9422&p=94828#p94828 <![CDATA[Re: Denial and Error: Lag, console commands and stupidity]]>
SeraphimLeftNut wrote:
http://www.overclock.net/t/1433882/gaming-and-mouse-response-bios-optimization-guide-for-modern-pc-hardware
This helped me


Seems like you have found about everything there is to know about mouse input lag. :D

Statistics: Posted by Col_Walter_Kurtz — 26 Feb 2015, 15:16


]]>
2015-02-26T05:19:19+02:00 2015-02-26T05:19:19+02:00 /viewtopic.php?t=9422&p=94810#p94810 <![CDATA[Re: Denial and Error: Lag, console commands and stupidity]]> http://www.overclock.net/t/1433882/gami ... c-hardware
This helped me

Statistics: Posted by SeraphimLeftNut — 26 Feb 2015, 05:19


]]>
2015-02-17T04:39:11+02:00 2015-02-17T04:39:11+02:00 /viewtopic.php?t=9422&p=94088#p94088 <![CDATA[Re: Denial and Error: Lag, console commands and stupidity]]> I tried using a soundblaster sound card instead on onboard audio, but it made no difference.

One way to make these go away is to remove all sound files from the supcom FA folder.
I added some files back and calling to the microsoft/apl key correleates with the use of sound effects. If you zoom into a massive load of units all firing their weapon you can see a lot of this spam and associated drop in frame rate.

Statistics: Posted by SeraphimLeftNut — 17 Feb 2015, 04:39


]]>
2015-02-15T05:23:15+02:00 2015-02-15T05:23:15+02:00 /viewtopic.php?t=9422&p=93928#p93928 <![CDATA[Re: Denial and Error: Lag, console commands and stupidity]]>
Edit: I created the path the game was looking for and arrived at this:
HKLM\SOFTWARE\Wow6432Node\Microsoft\APL\Safe Mode
REG_DWORD 0x0

I don't actually know what kind of value it should be, I'm just guessing here. This did not stop the spam though, at least not with that value (0x0). Trying to google what this value is used for.

Edit again: It's a function in DSOUND that is spamming the APL key. Also if I'm reading this correctly there's another spammer in a module called xactengi (xactengine2_9.dll). I'm not good enough at assembly to tell you why any of this is happening though.

Final edit: It would appear that the value returned by RegQueryValueExA from the DSOUND module is never used. The next time the memory address is accessed is when it's being written to again. I think it doesn't matter what value Safe Mode has, the information retrieved from the registry is never used.

Statistics: Posted by TheDarkOne — 15 Feb 2015, 05:23


]]>
2015-02-14T02:29:32+02:00 2015-02-14T02:29:32+02:00 /viewtopic.php?t=9422&p=93869#p93869 <![CDATA[Re: Denial and Error: Lag, console commands and stupidity]]>
I can turn vsync on or off and gsync on or off and the performance will still be playable. What I was trying to present is how to get the best performance from what I have tried over the last few months. The point of this thread is not to solve my issues, but to get people to see if they can get better performance and then to report back progress. I would really love to see people's reactions when they turn cam_EntityBoxExpand to -1000.

Process monitor is a must have program for everyone. For me firefox is the biggest offender, followed by norton. Also Nvidia Inspector likes to create lots of noise when its on.
My process interrupts are at about 0.1% when nothing is happening. Turning on audio adds another 0.1%, then moving the mouse adds less than 0.1%. If I had a pci slot I would have a dedicated sound card.

Thanks to process monitor, I have just found out that I have this problem: http://forums.gaspowered.com/viewtopic.php?t=7428
specifically FA keeps looking for the HKLM\Software\Wow6432Node\Microsoft\APL registry key, which doesn't exist. If anyone has ideas I would be interested

Statistics: Posted by SeraphimLeftNut — 14 Feb 2015, 02:29


]]>
2015-02-14T00:22:33+02:00 2015-02-14T00:22:33+02:00 /viewtopic.php?t=9422&p=93865#p93865 <![CDATA[Re: Denial and Error: Lag, console commands and stupidity]]>
SeraphimLeftNut wrote:
I would also submit that when you are dealing with a complex system designed by 10 different parties, you will get surprising results. I have been reading forums on all these issues on for the past 6 months, and everywhere I have seen "experts" be very sure about their statements. This confidence is very nearsighted.


The confidence is based on hundreds, in many cases thousands of hours dealing with people who continually take magic, smoke, and mirrors over real, science-based solutions, then continue to have problems months and years into the future. You appear to have been struggling with this stuff in SupCom for at least a year now, and I bet that if you had the appropriate diagnostic tools you'd have solved it instantly, and wouldn't be fooling yourself that this setting or that setting makes any difference. Only when they finally take the machine itself to a professional does that pro then fix whatever the issue was in 5 seconds flat.

Unfortunately I don't know where you can find such tools. Try online PC hardware stores, or electronics stores, or forums. I just use the stuff from dad's old work...

Yes, there can be odd interactions. But GSync doing anything whatsoever when you're using VGA output? Nope.

By the way, UI lag /=/ slower mouse motion. UI lag on the mouse would manifest as an input lag. You move your hand, THEN the cursor moves. If you start bringing mouse sensitivity and movement accuracy, then you open up a whole new can of worms!



To be honest, I dunno why you're having any trouble at all with your system :/ I have a very similar desktop back home, running an i5 4690 and an R9 280X, which *should* be slower, and it runs supcom as tight as a drum, never lagging and never skipping a beat. And I'm the kind of gamer who can't STAND anything <55 FPS in any game, and who can tell the difference between 5ms and 15ms of latency in TF2...

Statistics: Posted by IceDreamer — 14 Feb 2015, 00:22


]]>
2015-02-13T22:22:10+02:00 2015-02-13T22:22:10+02:00 /viewtopic.php?t=9422&p=93859#p93859 <![CDATA[Re: Denial and Error: Lag, console commands and stupidity]]>
I'm running:
955BE AM2 CPU, 3.21Ghz, 4 cores
nVidia 780TI, drv 337.88
8GiB (2x4) 1333Mhz DDR3

Things I'd look into when it comes to performance issues:

1) http://www.sysinternals.com. Windows techie's must-have tools. For this situation: Process Monitor and Process Explorer.

Process Explorer can be used to get a slightly more in-depth view of processes. Perhaps the most relevant features here would be process affinity and the System.Interrupts numbers.

Process Affinity
There's a question I've never found the answer to that goes something like this:
If you have a process that runs 2 threads that use up all processing power of the cpu thread they are assigned to, which of these scenarios will occur:
a) A hyperthreaded cpu will assign the 2 threads to the first 2 cpu threads using the power of only one core.
b) A hyperthreaded cpu will assign the 2 threads to the first and third cpu thread, using up the power of 2 cores.

In case of 'a' that would, perhaps overly simplified, mean that a non-hyperthreaded cpu with identical performance will in fact have double performance over a hyperthreaded cpu.

With that crazy idea out of the way, uncheck every other cpu thread for the FAF process. Force it to use the same number of threads as you have cores - one thread for one core.

System.Interrupts
The cpu usage of this tree item can sometimes be useful for spotting performance issues with drivers. In general you don't want this to have a high number (>1%). Examples (while idle) from my system:
0.37% usage while not doing anything
0.56% while moving the mouse around
I'd look into drivers if this number was 1-2% or over, especially when idle.

Process Monitor
This tool requires a lot more getting used to but it sounds like you've already used this one. Generally when running this tool you don't want to see spam of hundereds of thousands of lines per minute. It should be more like 50 lines every 2-5 seconds. Look into and disable (if possible) any application breaking this rule.

Generally the noisiest applications are:
1) SearchIndexer.exe (Windows Search service)
2) Media applications and software. Printers, Scanners, Camera, Phone, iTunes...
3) Recent versions of Steam.exe
4) svchost.exe but only because it's the host process(es) for windows services - use process monitor or "tasklist /svc" to find the services hosted by a noisy host process. Eliminate them one by one (if possible) until the noise stops.

This application causes a noticeable performance hit when running which is something to keep in mind.

2) Wireshark. Another must-have tool for debugging anything network related.

For this I don't have much advice. It requires a whole lot of understanding about network protocols to get anything out of this application, but for those who have a moment to sit down and google things, it will definitely allow you to see things going on with the network. The WinPcap driver that is used by Wireshark can add a noticeable performance hit on it's own which is something to keep in mind.

3) System cleaning.

Generally by the time you get here, Process Monitor has already located all troublemakers. But one effective step is to scan through the Programs and Features list (appwiz.cpl) and getting rid of.... well I guess that's something I don't know how to explain. I personally see bad applications stick out like a sore thumb but trying to explain what to look for is a different matter. In general:

1) Device drivers and software no longer in use.
2) Toolbars for browers.
3) Bundled applications that came with your computer (only really applies to OEM computers)
4) "30-day" trial software.
5) Registry and system cleaners.
6) Security suites, if there's multiple.

Specifically for FAF:

I can't say I've ever seen any other kind of lag other than unit command lag with large groups. This I work around by selecting only a part of the units in the group I want to command, and issuing several identical commands. This way by the time I'm done issuing the last command, the first batch of units are already on their way. Basically just gets the units moving faster.

When it comes to compatibility with nVidia, there's definitely a problem with dual monitors. While yes, this dual monitor problem probably has nothing to do with performance - it's just a crashing issue - but the idea is that perhaps FAF just generally works better with the older drivers. I recently spoke (although it was a one-way conversation) with BRNKofINSANITY about his dual monitor crashing issue. I'm no longer able to run FAF dual monitors either, but I remember these issues started after an nVidia driver update. What I don't know is whether or not BRNKofINSANITY saw my message and tried out the driver version I suggested. He did mention in one of his latest videos that a driver regression did fix the issue, but I do not know how far back in history he had to go; and whether or not this version of the driver actually supports our 7xx series cards at all. The version I suggested to him was the oldest one in my local driver repository: 301.42

Obviously there's a lot more that can be done but I'd be here all week typing if I tried include everything.
HTH

Statistics: Posted by TheDarkOne — 13 Feb 2015, 22:22


]]>
2015-02-13T22:14:27+02:00 2015-02-13T22:14:27+02:00 /viewtopic.php?t=9422&p=93858#p93858 <![CDATA[Re: Denial and Error: Lag, console commands and stupidity]]>
IceDreamer wrote:
I'd love to see you try to prove that by getting input lag measurements...

There's no way that VSync on can reduce input lag. It simply can't, the way it's programmed and operates simply doesn't allow it. QED: Placebo. If you got a USB polling device and sophisticated hardware and input monitoring software, you'd be able to prove it to yourself.

GSync also... eeeeeeehhh. It COULD be a bug I suppose... But the thing is, you then said you're using VGA. The GSync hardware on the chip isn't even hooked up to the VGA output, so its utterly impossible the two things can be connected. Toggling GSync on/off should make no impact over VGA. If you can prove it does... Well, Nvidia 0, TA 1 I guess, but I find it highly unlikely. It's possible that the GSync is causing some issue somehow, but that it was the move from Displayport (GSync is hooked up to this) to VGA which made the difference, rather than the act of toggling the setting on/off.


Could you direct me to a place where I can get this hardware/software, I would love to play with it. The reason I mentioned the VGA cable was precisely because it makes no sense. I have tried gsync on and off with VGA. I will admit that the difference is very minor, it is probably one of the least important things I have written in my original post.

Also perhaps Vsync is just a good frame limiter, which makes it seem like input lag is lowered. I can turn vsync off and set sc_FrameTimeClamp to 0 and get 400 fps, however the cursor/ui lag are increased. Maybe calling it input lag is the wrong way to call it, but the cursor moves slower.

I would also submit that when you are dealing with a complex system designed by 10 different parties, you will get surprising results. I have been reading forums on all these issues on for the past 6 months, and everywhere I have seen "experts" be very sure about their statements. This confidence is very nearsighted.

Statistics: Posted by SeraphimLeftNut — 13 Feb 2015, 22:14


]]>
2015-02-13T21:38:14+02:00 2015-02-13T21:38:14+02:00 /viewtopic.php?t=9422&p=93857#p93857 <![CDATA[Re: Denial and Error: Lag, console commands and stupidity]]>
There's no way that VSync on can reduce input lag. It simply can't, the way it's programmed and operates simply doesn't allow it. QED: Placebo. If you got a USB polling device and sophisticated hardware and input monitoring software, you'd be able to prove it to yourself.

GSync also... eeeeeeehhh. It COULD be a bug I suppose... But the thing is, you then said you're using VGA. The GSync hardware on the chip isn't even hooked up to the VGA output, so its utterly impossible the two things can be connected. Toggling GSync on/off should make no impact over VGA. If you can prove it does... Well, Nvidia 0, TA 1 I guess, but I find it highly unlikely. It's possible that the GSync is causing some issue somehow, but that it was the move from Displayport (GSync is hooked up to this) to VGA which made the difference, rather than the act of toggling the setting on/off.

Statistics: Posted by IceDreamer — 13 Feb 2015, 21:38


]]>
2015-02-13T20:16:42+02:00 2015-02-13T20:16:42+02:00 /viewtopic.php?t=9422&p=93848#p93848 <![CDATA[Re: Denial and Error: Lag, console commands and stupidity]]>
IceDreamer wrote:
Sure thing :) Remember, the ideal settings will be based on whether your system is GPU or CPU walled. For SupCom, I'd want to free up as much CPU as possible since a GTX 780 should run the game at 60FPS + with no trouble whatsoever.

Turn off shader Cache - This should be on. Shader cache is simple a memory setting, it takes supcom's Graphics shaders and saves them to disk, freeing up CPU time.

Maximum pre rendered frames 1 - This one is tricky. If you want to free up CPU, this should be at 1, or even 0, so that the CPU doesn't work on a new frame until the GPU is ready. If you're GPU bound (As I am) you want to put this up, as it uses spare CPU cycles to minimise the time the GPU spends waiting for the next frame.

Triple buffering on - Good

Anisotropic Filtering on (really helps to make icons more clear) - Good

Threaded optimization off - Supcom's engine can use multiple CPU cores, so this should be set to Auto or On such that the drivers will be able to take advantage of multiple cores as and when the engine tells them they can.

Frame rate limited 60 fps - Sure

Highest available refresh rate - Yup

Allow vsync to vary refresh rate - VSync should be off anyway, see below

vsync smooth afr on - VSync should be off. Always. It will always cause additional lag. All it will ever do is fix tearing, so if that really bothers you so much then leave it on, but for responsiveness, especially mouse response, this should be permanently disabled.

Gsync on (I don't get this at all, but it makes things smoother) - This will make no difference on your machine. None. Zilch. Nil. If you claim to see any difference, I'm telling you it's a placebo. Since you don't have a GSync monitor, there is no chance whatsoever that this setting will make any difference at all.

AA off - You can turn this on if the GPU is kicking out solid framerate, if you want to.



The most crucial thing to remember for you though is that as long as you're at 60+ FPS, changing any of the settings here other than pre-rendered frames (Can free up CPU) and VSync (On WILL introduce input lag) will make very little, if any difference to your SupCom experience. You should see far more difference from all the ingame console commands you're tweaking, because your system is more than fast enough to run the game :)


With regard to shader cache. Look at your hdd activity as you move in and out with the camera with 100's of units. FA tries to get new sound and texture info off the disk, therefore freeing up hdd activity for a very small cpu price made a difference for me.(I have gone through many different driver versions for my sata controller, and I have settled on one that doesn't make isr/dpc during work)

I played without vsync for 2 years because everyone kept saying exactly what you have said, I completely disagree based on my experience. I have never seen any tearing with it on or off, what I have seen is less input lag from the mouse with it on.(I know this is completely against all the geniuses out there who write on pc forums)
The gsync options intrigued me and I played for hours with gsync on and off an I can say conclusively that on my system it makes a big difference in the smoothness of the cursor. (I have tested both a display port connection and a standard vga cable, and VGA cable won for me, with gsync on) Reading about gsync online, this makes no sense, so I just assume enabling gsync removes some bug. I am sure there are bugs because I have gone though about 5 different versions of GPU drivers from nvidia, which where all latest at the time, and the system functioned quite differently with different ones. I have absolutely no faith in people who make these drivers.

Thank you for the threaded optimization suggestion, I will play with it on for a while.

Decreasing the number of prerendered frames decreases input lag.

You are correct, console commands can have a significantly bigger effect than all these options.

Placebo effect really exists, especially when you get trained to play under certain settings, but it is not the case above.

Statistics: Posted by SeraphimLeftNut — 13 Feb 2015, 20:16


]]>
2015-02-13T19:18:54+02:00 2015-02-13T19:18:54+02:00 /viewtopic.php?t=9422&p=93838#p93838 <![CDATA[Re: Denial and Error: Lag, console commands and stupidity]]> Remember, the ideal settings will be based on whether your system is GPU or CPU walled. For SupCom, I'd want to free up as much CPU as possible since a GTX 780 should run the game at 60FPS + with no trouble whatsoever.

Turn off shader Cache - This should be on. Shader cache is simple a memory setting, it takes supcom's Graphics shaders and saves them to disk, freeing up CPU time.

Maximum pre rendered frames 1 - This one is tricky. If you want to free up CPU, this should be at 1, or even 0, so that the CPU doesn't work on a new frame until the GPU is ready. If you're GPU bound (As I am) you want to put this up, as it uses spare CPU cycles to minimise the time the GPU spends waiting for the next frame.

Triple buffering on - Good

Anisotropic Filtering on (really helps to make icons more clear) - Good

Threaded optimization off - Supcom's engine can use multiple CPU cores, so this should be set to Auto or On such that the drivers will be able to take advantage of multiple cores as and when the engine tells them they can.

Frame rate limited 60 fps - Sure

Highest available refresh rate - Yup

Allow vsync to vary refresh rate - VSync should be off anyway, see below

vsync smooth afr on - VSync should be off. Always. It will always cause additional lag. All it will ever do is fix tearing, so if that really bothers you so much then leave it on, but for responsiveness, especially mouse response, this should be permanently disabled.

Gsync on (I don't get this at all, but it makes things smoother) - This will make no difference on your machine. None. Zilch. Nil. If you claim to see any difference, I'm telling you it's a placebo. Since you don't have a GSync monitor, there is no chance whatsoever that this setting will make any difference at all.

AA off - You can turn this on if the GPU is kicking out solid framerate, if you want to.



The most crucial thing to remember for you though is that as long as you're at 60+ FPS, changing any of the settings here other than pre-rendered frames (Can free up CPU) and VSync (On WILL introduce input lag) will make very little, if any difference to your SupCom experience. You should see far more difference from all the ingame console commands you're tweaking, because your system is more than fast enough to run the game :)

Statistics: Posted by IceDreamer — 13 Feb 2015, 19:18


]]>
2015-02-13T18:58:03+02:00 2015-02-13T18:58:03+02:00 /viewtopic.php?t=9422&p=93834#p93834 <![CDATA[Re: Denial and Error: Lag, console commands and stupidity]]>
IceDreamer wrote:
Some of this is wrong.

Net lag won't necessarily cause projectiles etc to slow down at all. You can have horrendous packet loss or dropout issues with all players at +10 sim, and everything will move fine but you'll get big outright pauses.

CPU lag manifesting as simspeed lag won't necessarily cause any UI or mouse lag either, as the OS can shift those things onto another available core.

GPU lag won't cause camera stutter when you zoom in or out, that's a memory bottleneck at work. GPU lag will cause your FPS to drop, which some users will notice once it hits 50, some 30, some 15, depending on your general tolerance.

Oh, and your GPU Settings section is total crap. Looks like that's been based on smoke, mirrors and magic rather than what those settings actually do. Some of those settings will outright make things worse if you're GPU limited.

Course, the ultimate solution is just a new PC ;)


Your statements about the different kinds of lag are actually exactly what I said, if you read what I said again. Except maybe the GPU lag part.
This pc is a 4790K 780gtx 3200mhz ddr3 corsairs samsung ssd on an asus z97 ws motherboard win 8.1(latest drivers, latest bios, frimware, etc.). I agree that the gpu settings are probably system dependent, which is exactly what I stated. I did everything by trial and error.

I appreciate people trying to correct things, but the internet is full of suggestions like "restart pc", "get a new pc", "update drivers", etc.

I want some real knowledge about what different options do. Please enlighten about what your optimal GPU settings would be.

Statistics: Posted by SeraphimLeftNut — 13 Feb 2015, 18:58


]]>
2015-02-13T14:22:06+02:00 2015-02-13T14:22:06+02:00 /viewtopic.php?t=9422&p=93824#p93824 <![CDATA[Re: Denial and Error: Lag, console commands and stupidity]]>
Net lag won't necessarily cause projectiles etc to slow down at all. You can have horrendous packet loss or dropout issues with all players at +10 sim, and everything will move fine but you'll get big outright pauses.

CPU lag manifesting as simspeed lag won't necessarily cause any UI or mouse lag either, as the OS can shift those things onto another available core.

GPU lag won't cause camera stutter when you zoom in or out, that's a memory bottleneck at work. GPU lag will cause your FPS to drop, which some users will notice once it hits 50, some 30, some 15, depending on your general tolerance.

Oh, and your GPU Settings section is total crap. Looks like that's been based on smoke, mirrors and magic rather than what those settings actually do. Some of those settings will outright make things worse if you're GPU limited.

Course, the ultimate solution is just a new PC ;)

Statistics: Posted by IceDreamer — 13 Feb 2015, 14:22


]]>
2015-02-13T14:03:30+02:00 2015-02-13T14:03:30+02:00 /viewtopic.php?t=9422&p=93821#p93821 <![CDATA[Re: Denial and Error: Lag, console commands and stupidity]]>
I would strongly suspect the motherboard is the culprit.

Statistics: Posted by ZOB — 13 Feb 2015, 14:03


]]>