Ergebnis 1 bis 17 von 17

Computer Freezes with loud buzzing sound from speakers

  1. #1 Zitieren
    Ranger Avatar von Marek33
    Registriert seit
    Nov 2016
    Beiträge
    173
    Greetings!

    I have a problem with my PC when I want to play games with a very high graphic settings like the Crysis 1, Warhead or the Gothic (1 & 2) with the DirectX11 mod.

    At first, I was thinking it was just an issue with the DirectX 11 mod since it's in the Alpha and the author himself told me that in the last version was a problem with the water tessellation. Unfortunately, I have this problem also with the previous versions + from what I read across the internet I'm the only one which has a problem with the hard crashes on the Gothic 1 & 2 with the Dx11. The strange thing is, that Gothic 2 crashes basically when I load a game or just a few minutes after I run the game (always in less than a four minutes) and PC has good temperatures, plus power supply is not hot so it's probably the PSU. When I lower the GPU memory with the Core clock, then it works fine (for almost two hours). Gothic 1 with Dx11 works for over 30 minutes, but I have to do a hard restart when I play it for more than 40 minutes.

    Then I tested the Crysis 1 + Warhead since they are the best games for the graphic test. Crysis 1 crash after 6 to 12 minutes with a black screen, but the Crysis Warhead (called Crysis.exe) caused a hard restart, which is strange. But, when I used, different exe, it crashes after a 10 minutes with a black screen without hard restart. However, a lot of people have problems with running Crysis 1 or Warhead on Windows 7 so I don't know what to believe. I also tried Crysis 2 which surprisingly, works perfectly for over 30 minutes (probably more, but I don't much like it so I don't play it for long). Of course that I'm using the official HD Texture pack with the DirectX 11 since the original looks very poorly and it looks very bad without it.

    I have the latest drivers, I did Memtest86 on RAM (3 passes, without any no problem). It's definitely not HDDs since I monitor them daily + I'm testing them once a month with a program like Hard Disk Sentinel and HD Tune. I'm suspecting that it could be either GPU or PSU even if I don't think it's GPU since I tried Unique benchmarks and it doesn't crash even once (I do everyone for over 20 minutes). Furthermore, I have in both benchmarks, well over 60fps (Sanctuary and Tropics). Even if I had only 20fps in the Heaven, but only when I use extreme tessellation and other things. Nonetheless, that was because those things were made for GPU cards like the GTX780, GTX980 and for others. The strange thing is that lowest fps I had on Dx9 and on Dx11 with Moderate tessellation (the lowest one). With the normal and Extreme tessellation I have always more than 20fps even if the highest fps. The average was lower than the one in the Dx9 I still have average fps around 45FPS. The temperature shouldn't be a problem since when I played Metro 2033 (in summer), I have over 90°C temperature and it doesn't crash. Also, the Red Faction Guerrilla works great for over 40 minutes (after 50 minutes or mostly one hour it crashes with a black screen). BF3 works fine for over an hour, but the BF4 crashes after 10 minutes because EA never fixed the issue with the drivers (AMD), but I don't care much about it since it's mostly the same game like BF3. I never had a crash when my GPU has more than 80°C Degrees, but when it has 50 or 60°C I had a hard crash which is strange. So, it's definitely not a temperature issue, but probably a problem with the power supply or something else.

    Here is my complete system build so if you know what should I change, then please tell me: ( http://pcpartpicker.com/list/9RHw7h ). Only the keyboard is different from the one which I have, since it's old + it's not in the list. I have I bought the PC in the middle of the 2015 because the previous one was very old. I haven't noticed the problem until now since I played mostly the S.T.A.L.K.E.R. series where I never experienced this problem until Call of Pripyat.
    Marek33 ist offline

  2. #2 Zitieren
    Legende Avatar von jabu
    Registriert seit
    Jul 2011
    Beiträge
    7.328
    Zitat Zitat von Marek33 Beitrag anzeigen
    I never had a crash when my GPU has more than 80°C Degrees, but when it has 50 or 60°C I had a hard crash which is strange. So, it's definitely not a temperature issue, but probably a problem with the power supply or something else.
    That's very interesting.

    Briefly outlined:
    Memory (also graphic memory) reacts very sensitively to temperature when it comes to its limits. At decreasing temperatures, higher voltages are required to work properly and vice versa. For this reason, tests with lower temperatures are required in the case of memory overclocking. The same is true for GPUs, but it is less common.
    Aged electrolytic capacitors often reach a usable capacity at high temperatures (although aging increases with temperature) so that the voltage supply becomes unstable rather at low temperatures. This applies to voltage regulators on the graphics card as well as on the board or in the power supply.

    It would be a possible approach to test the entire graphics memory at low GPU load. However, the driver influences the memory clock as a function of the load, and perhaps also the memory voltage, which is why the automatic adjustments had to be deactivated for testing and the values, possibly by a given profile, had to be determined.

    It could be a try to test the graphics memory at reduced clock without lowering the voltage, even in normal game scenarios (for more stable operation).
    jabu ist offline

  3. #3 Zitieren
    Ranger Avatar von Marek33
    Registriert seit
    Nov 2016
    Beiträge
    173
    Zitat Zitat von jabu Beitrag anzeigen
    That's very interesting.

    Briefly outlined:
    Memory (also graphic memory) reacts very sensitively to temperature when it comes to its limits. At decreasing temperatures, higher voltages are required to work properly and vice versa. For this reason, tests with lower temperatures are required in the case of memory overclocking. The same is true for GPUs, but it is less common.
    Aged electrolytic capacitors often reach a usable capacity at high temperatures (although aging increases with temperature) so that the voltage supply becomes unstable rather at low temperatures. This applies to voltage regulators on the graphics card as well as on the board or in the power supply.
    So, should I do a test with a lower temperature or do you mean something else? When I play games like Gothic 1 (with a Reloaded Mod), Mafia 1, Max Payne 1 (or 2), Assassins Creed 1 (or 2), Battlefield 2 and Far Cry 1 in which the GPU has temperature a bit more than 50°C (depends on a game) I don't have a single issue. Well, I have this issue before having those 80 degrees temperature, but I haven't known about it since I didn't know that the Dx11 for Gothic doesn't have this issue (and newer version doesn't work much better because of the AMD drivers).

    Zitat Zitat von jabu Beitrag anzeigen
    It would be a possible approach to test the entire graphics memory at low GPU load. However, the driver influences the memory clock as a function of the load, and perhaps also the memory voltage, which is why the automatic adjustments had to be deactivated for testing and the values, possibly by a given profile, had to be determined.
    What exactly do you mean by this?

    Zitat Zitat von jabu Beitrag anzeigen
    It could be a try to test the graphics memory at reduced clock without lowering the voltage, even in normal game scenarios (for more stable operation).
    Do you mean that I should try some games with a reduced clock? Because I don't know why this would be helpful.

    Btw: Do know about a Discord? If you want we can chat there since its in real time and then I can (or you) write the solution here.
    Marek33 ist offline

  4. #4 Zitieren
    Legende Avatar von jabu
    Registriert seit
    Jul 2011
    Beiträge
    7.328
    Zitat Zitat von Marek33 Beitrag anzeigen
    So, should I do a test with a lower temperature
    According to the information so far, but because of your new hints, this should not be necessary at first. But it could be helpful anyway!

    When I play games like Gothic 1 (with a Reloaded Mod), Mafia 1, Max Payne 1 (or 2), Assassins Creed 1 (or 2), Battlefield 2 and Far Cry 1 in which the GPU has temperature a bit more than 50°C (depends on a game) I don't have a single issue.
    So is it reproducible that you never have symptoms at low temperatures?

    Well, I have this issue before having those 80 degrees temperature,
    I do not understand this, unfortunately. So you have symptoms at low temperatures?

    I guess: the problems emerge at a transition from a low to a high temperature on always at approximately 80 °C, right? This would be an information of essential importance.

    but I haven't known about it since I didn't know that the Dx11 for Gothic doesn't have this issue (and newer version doesn't work much better because of the AMD drivers).
    We should try to isolate the problem cleanly and do not mix symptoms of different planes. Maybe I do not quite understand you correctly and you have the symptoms exclusively with the DirectX 11 mod? We would have to separate quite clean.

    What exactly do you mean by this?
    Excuse my strictly speaking, but it would be even more difficult if we take the details apart as long as the observations are not reproducible. Maybe they are? We should work on this in order to isolate them. Error as a function of temperature, for example.

    Tests with an open housing can prevent the air heated by the graphics card from flowing through the power supply. Also this may be necessary or at least be helpful to isolate the problem.

    Do you mean that I should try some games with a reduced clock? Because I don't know why this would be helpful.
    It should be obvious that reduced clocks have a stabilizing effect. It is often possible to set the storage cycle independently of the GPU clock. From the connection with load and temperature some conclusions could be drawn.
    jabu ist offline Geändert von jabu (05.06.2017 um 15:04 Uhr)

  5. #5 Zitieren
    Ranger Avatar von Marek33
    Registriert seit
    Nov 2016
    Beiträge
    173
    Zitat Zitat von jabu Beitrag anzeigen
    According to the information so far, but because of your new hints, this should not be necessary at first. But it could be helpful anyway!
    How can I limit the temperature? Even if Its definitely not because of the temperature since no matter the temperature, it depends on a game. Since no matter if I have: 40, 50, 60, 70 or 80°C it depends purely on a game. For example, Crysis 1, Crysis 1 Warhead, Gothic with DirectX 11 (, but only when I have that shoulder mod), Battlefield 4 even if the Battlefield 3 works fine for over an hour.

    Zitat Zitat von jabu Beitrag anzeigen
    So is it reproducible that you never have symptoms at low temperatures?
    I have only at the Gothic 1 & 2 because the DirectX 11 mod is very taxing for the system. And unfortunately it doesn't work with the SystemPack so I could have from 50°C to 60°C, maybe 70°C, but definitely not more since its crash then. Sometimes it even crashes a few seconds after I load the latest save, but it doesn't work for more than four minutes.

    Zitat Zitat von jabu Beitrag anzeigen
    I do not understand this, unfortunately. So you have symptoms at low temperatures? I guess: the problems emerge at a transition from a low to a high temperature on always at approximately 80 °C, right? This would be an information of essential importance.
    No, I have these symptoms only when I play games which are very intensive like the ones which I mentioned in the previous answer so it's not because of the transition from a low to a high temperature.

    Zitat Zitat von jabu Beitrag anzeigen
    We should try to isolate the problem cleanly and do not mix symptoms of different planes. Maybe I do not quite understand you correctly and you have the symptoms exclusively with the DirectX 11 mod? We would have to separate quite clean.
    Yes, but only when I play a very high intensity game and now it doesn't happen then, since when my GPU has more than 70°C I never had that crash.

    Zitat Zitat von jabu Beitrag anzeigen
    Excuse my strictly speaking, but it would be even more difficult if we take the details apart as long as the observations are not reproducible. Maybe they are? We should work on this in order to isolate them. Error as a function of temperature, for example.
    Yes, exactly. And since it's a very demanding mod, it requires a completely different system than the vanilla game.

    Zitat Zitat von jabu Beitrag anzeigen
    Tests with an open housing can prevent the air heated by the graphics card from flowing through the power supply. Also this may be necessary or at least be helpful to isolate the problem.
    When I had the old case I do multiple open housing tests, but unsuccessfully (back in the January).

    Zitat Zitat von jabu Beitrag anzeigen
    It should be obvious that reduced clocks have a stabilizing effect. It is often possible to set the storage cycle independently of the GPU clock. From the connection with load and temperature some conclusions could be drawn.
    I have done a test in the November 2016 with the Gothic 2 and DirectX 11 (version 17.5) and it worked completely fine. I lowered Core Clock 1100 to 800 and Memory clock 1500 to 1100. It worked for over two hours and it would probably work even for three hours, but I get a game crash. It was a basic game crash which is not caused by my problem. It happened because I put a bandit armor near a bandit cave and the bandit (which saw me in the normal clothes) wanted to tell me that I was attacking him (I thought you were trying to pick a fight with me) and the game crash because it doesn't find that text.
    Marek33 ist offline Geändert von Marek33 (05.06.2017 um 20:01 Uhr)

  6. #6 Zitieren
    Legende Avatar von jabu
    Registriert seit
    Jul 2011
    Beiträge
    7.328
    Zitat Zitat von Marek33 Beitrag anzeigen
    How can I limit the temperature?
    Open the case, use additional fan. Try using a cold computer (in the morning at 15 °C, for example). I had forgotten to mention the mainboard, which is a common instability. You can have blown it with a fan. Almost all non-permanent hardware errors are temperature-dependent in any way! There should be sufficient possibilities for delimitation. And you can also try the other way around (but not too hot!). Accurate observation and description is important. There is unfortunately no tool, which solves all problems.

    Even if Its definitely not because of the temperature since no matter the temperature, it depends on a game. Since no matter if I have: 40, 50, 60, 70 or 80°C it depends purely on a game. For example, Crysis 1, Crysis 1 Warhead, Gothic with DirectX 11 (, but only when I have that shoulder mod), Battlefield 4 even if the Battlefield 3 works fine for over an hour.
    So you have the error only with Gothic with DirectX 11 and shoulder mod (and never with another game)? And temperature is completely irrelevant? Do you have the same problem with other games (I suppose not)? Sorry that I ask so stupid, because that is really important!

    Although the symptom is so brutal as to speak for a hardware fault, it does not necessarily have to be the case. During my programming (where it was about older versions of DirectX) I occasionally had to realize that it is possible to force a system into the knee by mistake in the code. This has something to do with the fact that the focus is more on performance than on stability. Although windows is usually still functional, but the error image through the bug can be so brutal that it looks like a hardware error. It looks even more so if the error causes the game to take too many resources. It is usually possible, but not guaranteed, to terminate the application with the Task Manager. This has something to do with the hardware proximity of DirectX, especially with Direct Input (which is why deprecated in the meantime). A layman has hardly a chance to distinguish from a hardware fault. It becomes even more difficult if the graphics driver is not able to intercept a programming error that causes an undefined state, but instead passes it on to the hardware level. The symptom can be indistinguishable from a hardware fault. Another driver might react differently!

    A more precise description of the symptoms could perhaps give a more accurate indication, the more precise, the better.

    I have only at the Gothic 1 & 2 because the DirectX 11 mod is very taxing for the system. And unfortunately it doesn't work with the SystemPack so I could have from 50°C to 60°C, maybe 70°C, but definitely not more since its crash then. Sometimes it even crashes a few seconds after I load the latest save, but it doesn't work for more than four minutes.
    It could be because of the heavily fluctuating load, just because the game is basically not particularly demanding (from today's point of view), but due to the DirectX 11 mod within each frame the graphics card for milliseconds or less heavily loaded. This can overstate the voltage controls, which remains unclear which (power supply or graphics card as main error sources, possibly also the mainboard (Short load pulses could put the mainboard into a faulty condition!)). Unfortunately, I can not make measurements by remote diagnostics. With a little luck, however, there is a slight but fully reproducible temperature dependency (if your test series are sufficiently precise and reproducible, which usually failed with remote diagnostics). In that case it would be clarified that it is not a software error.

    No, I have these symptoms only when I play games which are very intensive like the ones which I mentioned in the previous answer so it's not because of the transition from a low to a high temperature.
    Thanks for claryfying this, but now we have new confusion:
    ... like the ones you have mentioned or only the ones...

    ^Big difference, very important! What do you mean exactly with game? Do you mean intensive settings instead of intensive games? Or do you mean intensive (other, but which?) games?

    Yes, but only when I play a very high intensity game and now it doesn't happen then, since when my GPU has more than 70°C I never had that crash.
    So the error is strongly temperature-dependent, because it never occurs at more than 70 °C (it should be easy to preheat)?

    That has been the reason for my annoying questions since days. If it is so, you should have quite a hardware fault, if it can be reproduced, even very safe.

    With such symptoms I would recommend to test the video memory.
    jabu ist offline Geändert von jabu (07.06.2017 um 19:19 Uhr) Grund: important additions, corrections

  7. #7 Zitieren
    Ranger Avatar von Marek33
    Registriert seit
    Nov 2016
    Beiträge
    173
    Zitat Zitat von jabu Beitrag anzeigen
    Open the case, use additional fan. Try using a cold computer (in the morning at 15 °C, for example). I had forgotten to mention the mainboard, which is a common instability. You can have blown it with a fan. Almost all non-permanent hardware errors are temperature-dependent in any way! There should be sufficient possibilities for delimitation. And you can also try the other way around (but not too hot!). Accurate observation and description is important. There is unfortunately no tool, which solves all problems.
    I will try it, but I think that I have a good enough cooling with my case.

    Zitat Zitat von jabu Beitrag anzeigen
    So you have the error only with Gothic with DirectX 11 and shoulder mod (and never with another game)? And temperature is completely irrelevant? Do you have the same problem with other games (I suppose not)? Sorry that I ask so stupid, because that is really important!
    No, I have it in the Gothic when I use DirectX 11 mod or in the Crysis 1 and Crysis Warhead. And also the Hitman Absolution.Yes, the temperature is completely irrelevant and I have it mostly in these (Gothic when I use DirectX 11 mod or in the Crysis 1 and Crysis Warhead).

    Zitat Zitat von jabu Beitrag anzeigen
    Although the symptom is so brutal as to speak for a hardware fault, it does not necessarily have to be the case. During my programming (where it was about older versions of DirectX) I occasionally had to realize that it is possible to force a system into the knee by mistake in the code. This has something to do with the fact that the focus is more on performance than on stability. Although windows is usually still functional, but the error image through the bug can be so brutal that it looks like a hardware error. It looks even more so if the error causes the game to take too many resources. It is usually possible, but not guaranteed, to terminate the application with the Task Manager. This has something to do with the hardware proximity of DirectX, especially with Direct Input (which is why deprecated in the meantime). A layman has hardly a chance to distinguish from a hardware fault. It becomes even more difficult if the graphics driver is not able to intercept a programming error that causes an undefined state, but instead passes it on to the hardware level. The symptom can be indistinguishable from a hardware fault. Another driver might react differently!

    A more precise description of the symptoms could perhaps give a more accurate indication, the more precise, the better.

    It could be because of the heavily fluctuating load, just because the game is basically not particularly demanding (from today's point of view), but due to the DirectX 11 mod within each frame the graphics card for milliseconds or less heavily loaded. This can overstate the voltage controls, which remains unclear which (power supply or graphics card as main error sources, possibly also the mainboard (Short load pulses could put the mainboard into a faulty condition!)). Unfortunately, I can not make measurements by remote diagnostics. With a little luck, however, there is a slight but fully reproducible temperature dependency (if your test series are sufficiently precise and reproducible, which usually failed with remote diagnostics). In that case it would be clarified that it is not a software error.
    If it would be a hardware error, there would be issues with almost every game and not a few. Plus, I think that there would be hard restarts on a game like Metro 2033, Battlefield 3, Battlefield 4, Far Cry 3 and so on.

    Yes, it definitely isn't (as a Vanilla game).

    Zitat Zitat von jabu Beitrag anzeigen
    Thanks for claryfying this, but now we have new confusion:
    ... like the ones you have mentioned or only the ones...

    ^Big difference, very important! What do you mean exactly with game? Do you mean intensive settings instead of intensive games? Or do you mean intensive (other, but which?) games?
    I mean games which are very demanding for the system. So, Crysis 1, Crysis Warhead Gothic with the DirectX 11 and once the Hitman Absolution. Also, I get a black screen in the Battlefield 4 after 10 minutes and I get sometimes from 10 to 20 minutes a black screen in the Metro 2033 Redux.

    Zitat Zitat von jabu Beitrag anzeigen
    So the error is strongly temperature-dependent, because it never occurs at more than 70 °C (it should be easy to preheat)?
    It's not a temperature dependent since it never occurs when I have more than 70 °C. I think that it's because there is a load of things to load in the game so it either crash at the start or under five minutes in (when the GPU is relatively cold and then before 60 °C it crashes).

    Zitat Zitat von jabu Beitrag anzeigen
    That has been the reason for my annoying questions since days. If it is so, you should have quite a hardware fault, if it can be reproduced, even very safe.

    With such symptoms I would recommend to test the video memory.
    So, the GPU is faulty? And could it be the PSU?
    Marek33 ist offline

  8. #8 Zitieren
    Legende Avatar von jabu
    Registriert seit
    Jul 2011
    Beiträge
    7.328
    Zitat Zitat von Marek33 Beitrag anzeigen
    I will try it, but I think that I have a good enough cooling with my case.
    It is not a matter of whether your cooling would have been sufficient under normal circumstances, but whether a temperature dependence exists, which can be shown by the fact that an error occurs at other temperatures sooner or later. That would be a strong indicator for a hardware error, almost a proof. The reverse conclusion that there is none due to unchanged or unobtrusive behavior would be inadmissible.

    No, I have it in the Gothic when I use DirectX 11 mod or in the Crysis 1 and Crysis Warhead. And also the Hitman Absolution.Yes, the temperature is completely irrelevant and I have it mostly in these (Gothic when I use DirectX 11 mod or in the Crysis 1 and Crysis Warhead).
    [...]
    I mean games which are very demanding for the system. So, Crysis 1, Crysis Warhead Gothic with the DirectX 11 and once the Hitman Absolution. Also, I get a black screen in the Battlefield 4 after 10 minutes and I get sometimes from 10 to 20 minutes a black screen in the Metro 2033 Redux.
    So the problem also occurs in some other games. This normally indicates a hardware error, a misconfiguration (e.g., overclocking), or a driver error (You have several other drivers tested, even with conservative settings?). But it is difficult to judge from afar.

    Are these the most demanding ones?

    What exactly do you mean with black screen? I absolutely know what it means. Nevertheless, I ask dumb again:
    Is it possible in any way to return to Windows?
    What happens at [Win]?
    What happens at [Alt] + [Tab]?
    What happens at [Ctrl] + [Alt] + [Del]?
    Is the screen just black?
    Does the pc restart by itself?
    Does the pc shut off?

    Is it completely the same as with Gothic?

    If it would be a hardware error, there would be issues with almost every game and not a few.
    Sometimes it is so. But modern hardware is far too complex to draw such conclusions. So no, the conclusion is not correct (although your assumption could be right).

    It's not a temperature dependent since it never occurs when I have more than 70 °C.
    You could be right again, but it is an inadmissible conclusion. It does not prove that there is no temperature dependency, it says that your GPU did not exceed 70 °C. These are two different statements. For example, it says nothing about whether you get the error with a different temperature at another place. Think of temperature as a useful indicator, like a tool. And it says little about the temperature of the other components, such as the video memory or the capacitors or anything else. Video memory can also be very warm when the GPU has little to calculate. It can also preferably produce errors at low temperatures.

    I think that it's because there is a load of things to load in the game so it either crash at the start or under five minutes in (when the GPU is relatively cold and then before 60 °C it crashes).
    Maybe you could have an error with the drive or the file system. Have you already checked your file system (CHKDSK) and have it repaired? Did you use the system file checker (SFC) afterwards?

    So, the GPU is faulty? And could it be the PSU?
    It is unfortunately impossible to answer this on the basis of the given information with sufficient certainty. The black screen could speak for a problem with the PSU, but to say something, I would have the black screen closer specified. And a question would have to be answered: are with gothic the symptoms different (not this black screen)? Perhaps these are two completely different scenarios and symptomatics? In the case at least the general black screen should be temperature-dependent, be it with the graphics card, with the mainboard or with the PSU. The question would be where the temperature dependence exists. If not, that would also be interesting.

    What about benchmarks that target specific load scenarios?

    A suggestion: your CPU has an integrated graphics solution and your mainboard has probably corresponding connections. How would it be to remove your graphics card and test with the integrated graphics?

    ;tldr

    But we have left the original track, so to link there:
    Zitat Zitat von Marek33 Beitrag anzeigen
    The strange thing is, that Gothic 2 crashes basically when I load a game or just a few minutes after I run the game (always in less than a four minutes) and PC has good temperatures,
    This could possibly be related to initializations or the allocation of the video memory. What happens with more conservative settings regarding the use of the video memory (in the original Gothic in the ini-file, with the mod you have to look yourself)?

    When I lower the GPU memory with the Core clock, then it works fine (for almost two hours). Gothic 1 with Dx11 works for over 30 minutes, but I have to do a hard restart when I play it for more than 40 minutes.
    If the correlation between clock and stability is reproducible, then a hardware error is very likely (in particular a defective graphics card). However, it should nevertheless somehow be localizable about its temperature dependence. But sometimes the thermal coupling is too great to make clear statements. It is helpful to design test conditions in such a way that the components do not heat each other.

    In order to isolate the power supply, it can help to produce full load, both with the CPU and with the graphics card (at normal clock; but it is difficult to get a graphic card to nearly full load), as well as at the same time. If you are not able to move the PSU to shutdown, then this speaks against a problem with the PSU. Otherwise, so in case of a crash it says almost nothing. But in combination with a finding that the same symptoms occur reproducibly also with low PSU-load and with regard to each line, then it leaves quite the presumption, that the PSU is not causative. On its own, the latter would not say much. So you have to be careful not to fall into a logic trap.

    VSync should be included in many considerations, because without VSync, even the simplest loading screens can lead to high load and high power consumption.
    jabu ist offline Geändert von jabu (08.06.2017 um 11:34 Uhr) Grund: corrected wrong shortcuts

  9. #9 Zitieren
    Ranger Avatar von Marek33
    Registriert seit
    Nov 2016
    Beiträge
    173
    Zitat Zitat von jabu Beitrag anzeigen
    It is not a matter of whether your cooling would have been sufficient under normal circumstances, but whether a temperature dependence exists, which can be shown by the fact that an error occurs at other temperatures sooner or later. That would be a strong indicator for a hardware error, almost a proof. The reverse conclusion that there is none due to unchanged or unobtrusive behavior would be inadmissible.
    I see. So It's very important for me to try it.

    Zitat Zitat von jabu Beitrag anzeigen
    So the problem also occurs in some other games. This normally indicates a hardware error, a misconfiguration (e.g., overclocking), or a driver error (You have several other drivers tested, even with conservative settings?). But it is difficult to judge from afar.

    Are these the most demanding ones?
    Yes, these are one of the most of the demanding ones.

    Zitat Zitat von jabu Beitrag anzeigen
    What exactly do you mean with black screen? I absolutely know what it means. Nevertheless, I ask dumb again:
    Is it possible in any way to return to Windows?
    What happens at [Win]?
    What happens at [Tab]?
    What happens at [Str] + [Alt] + [Del]?
    Is the screen just black?
    Does the pc restart by itself?
    Does the pc shut off?
    It depends on a game. With games like (Crysis 1, Crysis Warhead Gothic with the DirectX 1) my PC just froze and I can't do anything except a hard restart. At the Battlefield 4 and the Metro 2033 Redux, I have a black screen, I can Ctrl + ALT +Delete. If the PC doesn't freeze completely I can access the Windows, mostly do at + Tab and the PC does never restart by itself or it never shuts off even in a game where I need to do a hard restart.

    Zitat Zitat von jabu Beitrag anzeigen
    Is it completely the same as with Gothic?
    Unfortunately no. It just froze with Gothic.

    Zitat Zitat von jabu Beitrag anzeigen
    Sometimes it is so. But modern hardware is far too complex to draw such conclusions. So no, the conclusion is not correct (although your assumption could be right).

    You could be right again, but it is an inadmissible conclusion. It does not prove that there is no temperature dependency, it says that your GPU did not exceed 70 °C. These are two different statements. For example, it says nothing about whether you get the error with a different temperature at another place. Think of temperature as a useful indicator, like a tool. And it says little about the temperature of the other components, such as the video memory or the capacitors or anything else. Video memory can also be very warm when the GPU has little to calculate. It can also preferably produce errors at low temperatures.

    Maybe you could have an error with the drive or the file system. Have you already checked your file system (CHKDSK) and have it repaired? Did you use the system file checker (SFC) afterwards?
    Yes, I have checked the file system (CHKDSK).

    Zitat Zitat von jabu Beitrag anzeigen
    It is unfortunately impossible to answer this on the basis of the given information with sufficient certainty. The black screen could speak for a problem with the PSU, but to say something, I would have the black screen closer specified. And a question would have to be answered: are with gothic the symptoms different (not this black screen)? Perhaps these are two completely different scenarios and symptomatics? In the case at least the general black screen should be temperature-dependent, be it with the graphics card, with the mainboard or with the PSU. The question would be where the temperature dependence exists. If not, that would also be interesting.

    What about benchmarks that target specific load scenarios?
    Yes, they are completely different scenarios because some games froze PC completely (Gothic with Dx11) and some have just black screen. Do you mean benchmarks like the ones PC Wizard has, Graphic benchmarks or some other?

    Zitat Zitat von jabu Beitrag anzeigen
    A suggestion: your CPU has an integrated graphics solution and your mainboard has probably corresponding connections. How would it be to remove your graphics card and test with the integrated graphics?

    ;tldr
    I will try it.

    Zitat Zitat von jabu Beitrag anzeigen
    But we have left the original track, so to link there:

    This could possibly be related to initializations or the allocation of the video memory. What happens with more conservative settings regarding the use of the video memory (in the original Gothic in the ini-file, with the mod you have to look yourself)?
    It's around 20% most of the time.

    Zitat Zitat von jabu Beitrag anzeigen
    If the correlation between clock and stability is reproducible, then a hardware error is very likely (in particular a defective graphics card). However, it should nevertheless somehow be localizable about its temperature dependence. But sometimes the thermal coupling is too great to make clear statements. It is helpful to design test conditions in such a way that the components do not heat each other.

    In order to isolate the power supply, it can help to produce full load, both with the CPU and with the graphics card (at normal clock; but it is difficult to get a graphic card to nearly full load), as well as at the same time. If you are not able to move the PSU to shutdown, then this speaks against a problem with the PSU. Otherwise, so in case of a crash it says almost nothing. But in combination with a finding that the same symptoms occur reproducibly also with low PSU-load and with regard to each line, then it leaves quite the presumption, that the PSU is not causative. On its own, the latter would not say much. So you have to be careful not to fall into a logic trap.
    When I have a lower Core clock and Memory clock I have much lower temperatures and they are could be at 50 °C maximum (, but only at lowest possible values and I have around 36 FPS). When I tried medium values, then I had a bit more than 60 °C, little better FPS, but I don't remember how much exactly. Around 50 FPS probably. And how can I properly isolate the power supply?

    Zitat Zitat von jabu Beitrag anzeigen
    VSync should be included in many considerations, because without VSync, even the simplest loading screens can lead to high load and high power consumption.
    I have used VSync in some games, but when I don't, then I use RTSS to limit framerate.
    Marek33 ist offline Geändert von Marek33 (08.06.2017 um 11:33 Uhr)

  10. #10 Zitieren
    Legende Avatar von jabu
    Registriert seit
    Jul 2011
    Beiträge
    7.328
    For time reasons I can unfortunately not yet go into everything, so here the first part:

    Zitat Zitat von Marek33 Beitrag anzeigen
    I see. So It's very important for me to try it.
    Yes. Make sure that the graphics card does not heat the mainboard and not the PSU. Make sure that the PSU does not draw any hot air.

    Yes, these are one of the most of the demanding ones.
    So we now know that your PSU does not trigger a protective circuit even with the largest GPU load (so far, but not absolute!). Otherwise the pc would completely switch off.

    Meanwhile, I've looked at your PSU more closely, is a VS650 (not Vengeance)? It is a cost-effective model that provides the entire 12 V power supply via a single rail. How the other voltages are kept stable, I could not find, because Corsair is silent about it. Sometimes inexpensive PSUs can not keep the voltage constant enough at the heavily fluctuating loads of the GPUs. I do not claim that this is the case here, but I would be careful with the tests with a suitable software for a constant base load, at about 50 ... 75 % CPU load and test the games under this condition. This does not have to bring anything, but if this leads to an reproducible improvement (or the opposite), that would be quite meaningful. To improve the stability, I would disable the boost mode and set the CPU clock to a fixed value, about 80% of the nominal value (can be varied by test).
    jabu ist offline Geändert von jabu (08.06.2017 um 15:17 Uhr)

  11. #11 Zitieren
    Ranger Avatar von Marek33
    Registriert seit
    Nov 2016
    Beiträge
    173
    Zitat Zitat von jabu Beitrag anzeigen
    For time reasons I can unfortunately not yet go into everything, so here the first part:

    Yes. Make sure that the graphics card does not heat the mainboard and not the PSU. Make sure that the PSU does not draw any hot air.
    Alright, thanks and I will be waiting. I will try it tomorrow.

    Zitat Zitat von jabu Beitrag anzeigen
    So we now know that your PSU does not trigger a protective circuit even with the largest GPU load (so far, but not absolute!). Otherwise the pc would completely switch off.
    So, it doesn't seem to be caused by the PSU then?

    Zitat Zitat von jabu Beitrag anzeigen
    Meanwhile, I've looked at your PSU more closely, is a VS650 (not Vengeance)? It is a cost-effective model that provides the entire 12 V power supply via a single rail. How the other voltages are kept stable, I could not find, because Corsair is silent about it. Sometimes inexpensive PSUs can not keep the voltage constant enough at the heavily fluctuating loads of the GPUs. I do not claim that this is the case here, but I would be careful with the tests with a suitable software for a constant base load, at about 50 ... 75 % CPU load and test the games under this condition. This does not have to bring anything, but if this leads to an reproducible improvement (or the opposite), that would be quite meaningful. To improve the stability, I would disable the boost mode and set the CPU clock to a fixed value, about 80% of the nominal value (can be varied by test).
    I do have a high CPU load (over 60% or even over 70% only when I format a video). How can I set the CPU clock to a fixed value?
    Marek33 ist offline

  12. #12 Zitieren
    Ranger Avatar von Marek33
    Registriert seit
    Nov 2016
    Beiträge
    173
    Zitat Zitat von jabu Beitrag anzeigen
    Yes. Make sure that the graphics card does not heat the mainboard and not the PSU. Make sure that the PSU does not draw any hot air.
    I have tried what you suggest me and it unfortunately didn't help. The strange thing is that first and second time (in Gothic 2) I had an error message: Device Removed — DXGI_ERROR_DEVICE_REMOVED. Then I tried Gothic 1 and I have played it for a few minutes. I wait for about 10 minutes for lower temperature of my PC and I was going to play Gothic 2. I have played it for about four minutes and then I need to do a hard restart. After that I tried Classic Mod, but I have to do a restart again. And the graphic card was under 70°C + it didn't heat other components.
    Marek33 ist offline

  13. #13 Zitieren
    Legende Avatar von jabu
    Registriert seit
    Jul 2011
    Beiträge
    7.328
    Zitat Zitat von Marek33 Beitrag anzeigen
    I have tried what you suggest me and it unfortunately didn't help.
    The sense of the open housing does not consist in a success. It is not primarily about a lower temperature, but about controllable test conditions.

    Now it could be possible to influence the ambient temperature (but not too cold and not too hot and not too abruptly to avoid damaging the hardware) to pinpoint the error. The temperature would be targeted only to the component to be tested.

    With a closed housing that would be impossible.

    So have you run tests in the meantime with different ambient temperatures, 10 °C, 20 °C, 30 °C (not more than 40 °C)? I would assume that makes a difference. If there is no difference in gothic, then at least in the more demanding games differences should arise.

    If crashes occur reproducibly depending on the temperature of only a particular component sooner or later, then this is usually very helpful. As long as you do not carry out appropriate tests, you do not know if it helps.

    The strange thing is that first and second time (in Gothic 2) I had an error message: Device Removed — DXGI_ERROR_DEVICE_REMOVED.
    That says what it means. The driver no longer responds to certain DirectX requests. But there can be many different reasons for this:
    Problems with the driver (try some others), with the GPU or with the VRAM (do tests and benchmarks), with the stability of the power supply (bad contact in the PCI-E socket (remove and re-insert), bad contact of the auxiliary plug (remove and re-insert)) or an unstable mainboard or power supply caused by the enormously abrupt load peaks from your GPU etc.

    Any advantages with lowered GPU and VRAM clock? Perhaps this video could give some advice.

    The four minutes could be interesting because they seem to repeat. Is it always exaclly the same time span? That would be very interesting. I remember roughly that there is in an ini file of Gothic a value for the retention time of textures in the cache. Maybe this is exactly this time? But I do not know what influence the mod has on it.
    jabu ist offline Geändert von jabu (10.06.2017 um 15:36 Uhr)

  14. #14 Zitieren
    Ranger Avatar von Marek33
    Registriert seit
    Nov 2016
    Beiträge
    173
    Zitat Zitat von jabu Beitrag anzeigen
    The sense of the open housing does not consist in a success. It is not primarily about a lower temperature, but about controllable test conditions.

    Now it could be possible to influence the ambient temperature (but not too cold and not too hot and not too abruptly to avoid damaging the hardware) to pinpoint the error. The temperature would be targeted only to the component to be tested.

    With a closed housing that would be impossible.

    So have you run tests in the meantime with different ambient temperatures, 10 °C, 20 °C, 30 °C (not more than 40 °C)? I would assume that makes a difference. If there is no difference in gothic, then at least in the more demanding games differences should arise.
    Unfortunately, I don't have a climatisation to lower the ambient temperatures under 20 °C.

    Zitat Zitat von jabu Beitrag anzeigen
    If crashes occur reproducibly depending on the temperature of only a particular component sooner or later, then this is usually very helpful. As long as you do not carry out appropriate tests, you do not know if it helps.

    That says what it means. The driver no longer responds to certain DirectX requests. But there can be many different reasons for this:
    Problems with the driver (try some others), with the GPU or with the VRAM (do tests and benchmarks), with the stability of the power supply (bad contact in the PCI-E socket (remove and re-insert), bad contact of the auxiliary plug (remove and re-insert)) or an unstable mainboard or power supply caused by the enormously abrupt load peaks from your GPU etc.

    Any advantages with lowered GPU and VRAM clock? Perhaps this video could give some advice.
    I have tried to lower the Core Clock from the 1100 to the 1000, but it crashed after 14 minutes with the message device removed which it should eliminate as it says in the video. Second time I play it for 24 minutes and I closed the game because I have the game was widowed and not Fullscreen. Temperature was around 46 when I closed the game. In the game, it could be 56 or 58 at max. Then I open the game and in less than a minute I had a buzzing sound again. Hopefully I have some time to try Crysis 1 today or tomorrow with the lower Core Clock and then I will let you know. Plus, I have done some benchmarks from the UNIGINE and it worked completely fine for thirty minutes (max temperature 70°C) and min FPS was 20 in the Heaven and 23 in the Valley. Plus, I run Furmark for 30 minutes (yesterday) with max temperature 77°C and default Clock Value, but without any problem

    Zitat Zitat von jabu Beitrag anzeigen
    The four minutes could be interesting because they seem to repeat. Is it always exaclly the same time span? That would be very interesting. I remember roughly that there is in an ini file of Gothic a value for the retention time of textures in the cache. Maybe this is exactly this time? But I do not know what influence the mod has on it.
    Unfortunately no, and That's the strangest part of it. Four minutes at maximum and sometimes it's less than 30 seconds into the game (which I forgot to mention before).
    Marek33 ist offline

  15. #15 Zitieren
    Ranger Avatar von Marek33
    Registriert seit
    Nov 2016
    Beiträge
    173
    I think that I solved the issue. The problem was in the GPU drivers. I have to completely uninstall them and install only the latest one. Also, I have decreased the Core Clock to the Base value 1000 and then I have played games like Metro 2033, Metro 2033 Redux, Far Cry 2 all for over two hours without any issue and also Crysis Warhead for over an hour without a single problem. Unfortunately, I still have that buzzing sound in the Gothic 2 (probably because the GPU drivers which are not optimized for it) and its crash in less than two minutes in the Gothic 2. Battlefield 4 also works for over an hour without issue. It works definitely much more, but I can't play it with that low FOV.
    Marek33 ist offline

  16. #16 Zitieren
    Legende Avatar von jabu
    Registriert seit
    Jul 2011
    Beiträge
    7.328
    Hi,
    I am glad that you have handled the matter consequently and that you have registered back.

    Zitat Zitat von Marek33 Beitrag anzeigen
    I think that I solved the issue. The problem was in the GPU drivers. I have to completely uninstall them and install only the latest one.
    That would be the normal case. I do not understand why you did not always have it that way.

    Also, I have decreased the Core Clock to the Base value 1000 and then I have played games like Metro 2033, Metro 2033 Redux, Far Cry 2 all for over two hours without any issue and also Crysis Warhead for over an hour without a single problem.
    The slight factory overclocking is not worthwhile if one counters the thermal and stability disadvantages. It's more about marketing.

    Unfortunately, I still have that buzzing sound in the Gothic 2 (probably because the GPU drivers which are not optimized for it) and its crash in less than two minutes in the Gothic 2.
    It is a widespread misunderstanding that GPU drivers need to be optimized for an application. Somewhere will still be an error: Bug in the application, bug in the driver, bug in windows, an additional program (which steals the input focus etc. (or interferes illegally in DirectX or in the driver (recording, benchmark or tuning software etc. (interfering in buffer locking etc.))) or a hardware problem (for example instability caused by the fast load changes of the GPU etc.) ...

    Battlefield 4 also works for over an hour without issue. It works definitely much more, but I can't play it with that low FOV.
    Great, but what do you mean by "I can't play it with that low FOV"? A performance issue? Why?
    jabu ist offline

  17. #17 Zitieren
    Ranger Avatar von Marek33
    Registriert seit
    Nov 2016
    Beiträge
    173
    Zitat Zitat von jabu Beitrag anzeigen
    Hi,
    I am glad that you have handled the matter consequently and that you have registered back.

    That would be the normal case. I do not understand why you did not always have it that way.
    I have only installed the new drivers, but never uninstall them since I haven't thought that it would cause any issues. Plus, since that program is very popular (Iobit Driver Booster) it proved me once again that it's garbage and to be honest, most of their products became total garbage over the years.

    Zitat Zitat von jabu Beitrag anzeigen
    The slight factory overclocking is not worthwhile if one counters the thermal and stability disadvantages. It's more about marketing.
    Exactly! I'm not exactly sure if that also causes issues, but it's a very small boost so it's not important.

    Zitat Zitat von jabu Beitrag anzeigen
    It is a widespread misunderstanding that GPU drivers need to be optimized for an application. Somewhere will still be an error: Bug in the application, bug in the driver, bug in windows, an additional program (which steals the input focus etc. (or interferes illegally in DirectX or in the driver (recording, benchmark or tuning software etc. (interfering in buffer locking etc.))) or a hardware problem (for example instability caused by the fast load changes of the GPU etc.) ...
    It's still in the Alpha and there are listed that there are problems with the current AMD drivers, but I have no idea what causes that problem since 99% of other much more demanding games works perfectly (also no black screens anymore). Maybe there is some bug in the application, or maybe even in the driver, but I think that there shouldn't be any issues with the windows. Even if some additional applications like Fraps, (I should close it before the test and I will when I test it sometime in the future) HitmanProAlert or maybe something else.

    Zitat Zitat von jabu Beitrag anzeigen
    Great, but what do you mean by "I can't play it with that low FOV"? A performance issue? Why?
    I forgot to mention that I play the only single player and also that option is not available in the single player game unfortunately. They should just make a bots for the next battlefield and no campaign since they always fail (BBC2 had a good campaign even if the bots are way better).

    And thank you very much for all the help!
    Marek33 ist offline

Berechtigungen

  • Neue Themen erstellen: Nein
  • Themen beantworten: Nein
  • Anhänge hochladen: Nein
  • Beiträge bearbeiten: Nein
  •