Showing Posts For crpNOP.4165:
Well, there’s obviously something wrong, but what makes it noticeable is Windows resetting the driver. Now, there could be a variety of reasons windows may think the driver needs to be reset, like memory leaks or hardware failure, as pointed out before; but also it could be the GPU being busy and unable to respond to the system, like I’ve seen programming for CUDA. What I did, which doesn’t fix the problem if there’s actually one, but it just makes it seamless: set the timer that controls how much time has the graphic card to answer to the system checks to a higher value. It’s called Timeout Detection and Recovery, and it’s an usefull system failsafe that prevents a malfunctioning program to propagate the error to the point of crashing the OS. At least in my case, never happened again since the change, and my pc looks stable after hours of playing, so It could be windows being an a**.
So, without further dilation, here are the steps I followed:
-Access to the registry through “regedit”
-Go to HKEY_LOCAL_MACHINE/SYSTEM/ControlSet001/Control/GraphicsDrivers
-Go to Edit>New>DWord (or QWord, depending if it’s 64 bit system or 32)
-Name the new entry TdrDelay, and set its value to 4 or more, but I wouldn’t recommend giving it a too high value, just between 4-40 top.
-Reset the PC to activate the changes.
Hope this helps more people.
P.S.: You should make a restoration point beforehand, for restore the system if anything goes wrong with the registry. It’s a simple procedure, but messing with the registry is delicate thing at best, do it at your own risk.
Updating to windows 8.1 preview has reduced substantially the occurrence of this issue to the point it rarely happens.
To the people still experiencing this problem, have you a dual monitor settup?
It may very well be an electronic engineer thinking It’s hardware related, and a software engineer thinking It’s software.
At least in my case, time will tell. If It’s hardware related, the issue just can get worse.
Salutes.
Monitors don’t work that way electronically – whether they are analog (CRT) or digital (LCD). I presume by “information” you mean data (sync issues are different but I’ll address that next). The screen will display ANYTHING (data) the graphics card sends to it. Let me try and articulate that a little better:
You have digital data on the DVI output of your graphics card. Each bit can only be high or low. It doesn’t matter what the data bits are – the monitor WILL display it.
Sync issues, on the other hand, are different. CRT monitors (well most anyway) will blank the video if the sync frequency is out of range. This is to protect the CRT, which can be damaged if the sweep circuitry fails. I don’t know of any way to repair burnt phosphor in a CRT.
LCD monitors do the same, although sync being the wrong frequency won’t damage an LCD monitor.
Of course, it could be a driver issue – however, if the hardware is working properly; there is absolutely no reason for it to suddenly “malfunction”. In addition, DirectX is controlling the graphics driver when running GW2 (or any D3D game).
If it is indeed a driver issue, then I would expect that everyone with the same card by the same manufacturer would have the exact same problem. That is obviously not the case.
This problem points to failing or misconfigured (overclocked, etc.) hardware, specifically the GPU.
Edited to add: It’s nearly impossible for me to try and cover every possible aspect regarding this problem without writing a novel. Yes, other component failures such as motherboard and power supply could cause this problem.
I’m sure I forgot to mention something. Either way, I certainly hope the OP and others are able to resolve the problems they are having.
I use “information” as a collection of “data”, though data is just a value and says nothing about the order or the cohesion/coherence of the group. Information, in the other hand, as It’s taught in computer science, Is a sequence of data that can be interpreted as a message, which happens after the myriad of systems the data has to go through that treated It before reaching the monitor, and using information actually gives more information than using solely data. This way I cover RGB data, and clock data as a whole.
Though I admit english semantics may very well be out of my reach, for It’s not my first language, technical and scientific terms should not vary too much.
Anyhow, it’s irrelevant.
In regard of how does a LCD screen work electronically or how it does not, I beg to disagree. I’m afraid LCDs are a bit more complex than just Color Mapping. Granted DVI use a collection of digital data, which are just low and high voltages, but not all the information given to the monitor is plain displayable data, nor always the data is in the range of what a panel can actually display. Some LCD monitors will show the message “input signal out of range”, others just aren’t so friendly. The clock, for example, runs through a dedicated DVI channel, It’s given by the graphic card and the rate is set by the system. A clock failure can mimic the behaviour explained, or give a “mosaic” image. But I agree with you in that is hard to cover all the possibilities.
Tradicionally, TFT LCD panels could be damaged by sync mismanagement: they are sensitive to power sequences during power up and power down periods, and their operational lifetime could be dramatically decreased if this is disregarded. There are hardware built in failsafes that prevent this from happening in the Display Controller, very much à la CRT.
Answering to your edition, while not all the chips have failed, there are a fair amount that have. Moreover, not all the chips are exactly the same in a given architecture: there are hardware revisions and updates that are not notified to the public, remaining the specifications roughly the same. Nonetheless, we know other chips present this sympthoms, all the ones I’ve seen, AMD cards using the AMD driver, and sharing the fact that the issue is limited to this game. I think this is not just random hardware failure, as It has statistical significance.
I think It may be related with how shaders and post-processing effects are implemented.
Thanks for your thoughts,
Crp
Not necessarily. Different games use different resources in different ways.
Your graphics issue will occur again – it’s only a matter of WHEN it will occur.
What games use similar resources so that we can test this? Concrete examples will be helpful. Otherwise we can’t rule out a fault in GW2.
It’s irrelevant.
Evidently, I have to teach some people HOW to troubleshoot. Fine, here goes:
Symptom:
The video blanks out (no signal) after a certain period of time has passed. If this was a
driver issue, the driver would crash OR cause strange discolorations on the screen – the video signal would not turn off, leaving no signal to the monitor. That alone indicates one of three things:1) The video card is bad.
2) The cable from the video card to the monitor is bad.
3) The monitor is bad.
Now, the OP states that lowering the GPU clock speeds allows him to play longer before the problem occurs. That eliminates the cable AND monitor as being related to the problem. Lowering the GPU clock speed has no effect on the cable or the monitor.
When the GPU speed is lowered, this results in the video card operating at a lower temperature. This also means that it takes longer for the video card to reach the temperature where it fails.
You see, the video card is actually failing at a certain temperature – just not a HIGH temperature. This is quite common with BGA (look it up) mounted components (failing video cards, the Xbox360 “red ring of death”, some HP laptops, etc.).
I have seen this exact issue many times. It’s obviously a bad video card.
Good day.
The premise is wrong.
The driver can malfunction and the system still think it’s working properly, so it wouldn’t kill it, nor give any error message. The result would be the driver telling the graphic card to give information to the screen that the screen can’t interpret, hence the “no signal” message. I’ve seen this a lot back in the day of non-user-friendly unix enviroments while trying to figure out the synch of a laptop’s screen on a default driver.
Now, I’m not ruling out It could be a faulty GPU, what I’m saying is It could still be a driver issue, and your null hypothesis is flawed.
Cheers,
Crp
AMD display driver stopped responding and has successfully recovered
in Account & Technical Support
Posted by: crpNOP.4165
Setting post-processing and shaders to its minimum value seems to alleviate the problems.
AMD display driver stopped responding and has successfully recovered
in Account & Technical Support
Posted by: crpNOP.4165
Exact same problem here, AMD Radeon HD 4870. Overheating is not the problem, I’m quite positive. It’s a software issue, maybe the AMD driver but, since this is the only game that doesn’t get along with my driver, I can’t help but think It’s something gameclient wise, for I’ve been dragging this error for several driver updates now.
Any of you using a dual monitor settup?
Salutes,
Crp
You either counted just one heatseeker, or biased the entire calculation making the backstab hits at full damage and heartseeker at minimum. The fact is, I’ve seen, as a thief, 5k+ damage on mug and 5k+ on heartseeker (in fact, usually more) on the very same target that it could hit for 10k with a backstab. So, if you actually use a more reasonable aproach to the damage a heartseeker can do, to be fair with the rest displayed, the damage nerf begins to be blurred. It’s not a nerf in overall damage, as you said at the begin of the post, but a nerf in burst as you pointed out at the end.
This video, as an example of the numbers a thief can make. I picked this one because I saw it in the forums recently, but youtube is full of them.
But I won’t say just your numbers are wrong and mine are right. What I will say is that, when the difference is so small, like 1.1k, It’s quite easy to bias it just “being optimistic” in some skills, and less in others, as you did.
I’m not against of doing that as much as I am against the attitude that your post distilles.
Salutes,
crp
PD: English is not my first language, I apologize for any mistake.
(edited by crpNOP.4165)