6 replies to this topic

[dD0ntStopTh3Beat]

What A **** is going on? Before latest patch game crashed once a few hours during startup sequence on "Sensors - online" phrase. This crash was and still is connected to nVidia graphical drivers, but that is as much as I know. Now game crashes once 5-10 matches in quick play/ faction play and funniest thing is that even when I get pop-up window with suggestion to send crash report to developers - error reporting program stops responding. Earlier it crashed during start-up mostly, now it crashes even during game. So if you had cruise control on - and game crashes - you will walk right in the enemy group.
AND it is impossible to restart a game until reporting program is closed which leads to delay before man can login into game and rejoin match. And, you neither mention these crashes on the patch list nor saying that you are working on this issue and trying to solve it.
Suggestions like restarting the game once so often is out of question - we all know that it is not the answer.

[G4LV4TR0N]

Windows 10 x64?

Edited by G4LV4TR0N, 23 October 2017 - 08:03 AM.

[NARC BAIT]

sounds like problems, at your end ... which tend to be hard for the developers to provide a fix for, seeing as they don't know what goes on at your end ....

to have windows error reporting crash on you also, implies to me that you might want to do a thorough check for viruses, not many things will bring down windows error reporting, but malware wont want you sending info to anyone about how they made your app crash ... furthermore, if you cant send the report about a crash, and don't contact technical support, then they don't really have much way of even knowing, that you crashed ... simply disconnecting from the game, doesn't tell them anything, but they do tend to like to throw out the punishments ... because ... well 'farmers' beat special forces or something from history ...

beyond the realm of software you didn't intend to be running, you might want to check your RAM stability ... one bit incorrectly flipped at any point in the process will make MWO die, probably thanks to hashing ... not really something you can bypass ...

if your sure its related to your video card, whats it like (model) ? whats it doing, whats the error message ? is it overheating ? is it factory overclocked ? thres more questions than answers here ...

[dD0ntStopTh3Beat]

G4LV4TR0N, on 23 October 2017 - 08:02 AM, said:

Windows 10 x64?

Win 7 x64

NARC BAIT, on 23 October 2017 - 04:14 PM, said:

sounds like problems, at your end ... which tend to be hard for the developers to provide a fix for, seeing as they don't know what goes on at your end ....

to have windows error reporting crash on you also, implies to me that you might want to do a thorough check for viruses, not many things will bring down windows error reporting, but malware wont want you sending info to anyone about how they made your app crash ... furthermore, if you cant send the report about a crash, and don't contact technical support, then they don't really have much way of even knowing, that you crashed ... simply disconnecting from the game, doesn't tell them anything, but they do tend to like to throw out the punishments ... because ... well 'farmers' beat special forces or something from history ...

beyond the realm of software you didn't intend to be running, you might want to check your RAM stability ... one bit incorrectly flipped at any point in the process will make MWO die, probably thanks to hashing ... not really something you can bypass ...

if your sure its related to your video card, whats it like (model) ? whats it doing, whats the error message ? is it overheating ? is it factory overclocked ? thres more questions than answers here ...

I am pretty sure that it is connected with graphic card since I've seen system messages about it, mostly "driver stopped responding and was recovered".
It is GeForce GTX 1070, and so far I've had such crashes in MWO only. And no, it is not overheating. I've looked into BIOS, average temp is around 35-37 Celsius. And it comes with possibility of overclocking, but I am not using it though.
Besides, it is almost every QP match I see a person or two, some times it is up to six, who lose connection during startup sequence. And you want to tell me that we all have problems at our end, not the client stability? Hard to believe.
And again, error messages about video driver stopping to respond show up mostly during startup sequence. And it will not happen in FP if your 1st drop was successful ( game didn't crash).

[NARC BAIT]

dD0ntStopTh3Beat, on 25 October 2017 - 12:39 AM, said:

I am pretty sure that it is connected with graphic card since I've seen system messages about it, mostly "driver stopped responding and was recovered".

this is one of those terrible error message that tend to be hard to resolve, and most people when (if) they manage to get it to stop happening, never make it back to wherever they asked for help, and the internet gets left with a lot of questions, but not a lot of good information about how to resolve it .... essentially, the card hasn't responded to the driver, by the time the driver expects that it should have .... and you could be looking at a number of scenarios as to why it has occurred, I'm honestly not sure how long the 'time response window' is, but its pretty thin, and I would expect you would be measuring it in a matter of microseconds, because that's how often the card is expected to be updating fresh frames .... other programs that tend to be really good as exposing this issue are just about any title in the need for speed series after about 2003, but that's kinda beside the point ( I may be drunk at the moment, sorry for all rambling ) .... the last time that I had these issues occur frequently was with a NVidia 6600GT, right as we were swapping over from the AGP port, to PCIE ports .... and that might be the first thing to consider, is there anything about your PC that might cause delays or corruptions on the PCIE bus, in case that's junk jargon, were talking about maybe PCIE wifi cards, sound cards, network cards, maybe even an M2 NVME hard disk .... I know my current ryzen system has internal stuff to counteract 'poisoning' on that bus, and I recently found that my 'older' sound blaster plug in card ( that I love ) wasn't always playing so well at all frequencies ... on my current board, it drops into a protection mode, where it makes the CPU safe, by stopping it from doing anything, yeah, thanks for that one AMD/ASUS ....

next you might want to look at power supply stability, these current ( and future ) NVidia cards operate their chips at a pretty low voltage, my 1060 has a normal range between 0.6v and 1.05v ( and an absolute maximum of 1.2v, but mine gets less unstable at 1.093v ) but to get that, it starts of with two voltage options, one at 12v, and the other at 5v, don't quote me on which one is trimmed down to the actual supply voltage, but they can be particularly sensitive to changes, being why GPU's ended up with their own power connections to most PSU's, to eliminate other pieces from introducing noise or voltage drooping, but that doesn't completely mean that it wont happen ... on my old thermaltake PSU, I have an 8 pin PCIE connector and two 6 pin connectors ... the 8 pin cable, doesn't provide a stable enough supply to keep the GPU happy, and it will crash ( my shits hard OC most of the time ) .... other things to watch out for might be using a MOLEX to PCIE adaptor for the power supply, something like an older hard drive could cause a voltage droop on the rail when it wakes up, or goes to sleep ... could be something as stupid as a RGB setup that just happens to push it out of tolerance, for one nanosecond longer than it was expected to respond ... in terms of finding hidden issues, grab a program like GPU-Z or HWINFO and check if your hitting the 'reliability operating voltage limiter' while your in game, and have it under load ... afterburner tends to not show rVOP .... but there is a good option in it that you can set it to force a constant voltage, which will encourage the voltage to 'jump less' which may help in some situations ...

most of the people whove had the issue, end up underclocking their cards to bypass the issue ... and that may be an option for you, because unless your doing something crazy like MSSA @ 4K, your probably not going to be getting the card upto 100% load ....

there was another fix, that we used to use ( for need for speed titles ), where you would edit the registry and add a key to increase the timeout value .... its been years since ive done that one, and my google-fu is failing me for it right at the moment ...

dD0ntStopTh3Beat, on 25 October 2017 - 12:39 AM, said:

Besides, it is almost every QP match I see a person or two, some times it is up to six, who lose connection during startup sequence. And you want to tell me that we all have problems at our end, not the client stability? Hard to believe.

at no point am I vouching for the client being 100% stable .... and your not the only person to have ever complained about it crashing, only at the start of a round .... but, on the other side of the coin, if people don't send in those windows error reports, how are the devs meant to know that the client crashed, as opposed to you just hit alt-f4 and weren't there anymore .... the servers cant tell wether the client crashed, the network route died, or power went off in your region ... if you see 6 people disconnect 'at the same time', that's probably network related because even though the clients are spread out in different parts of the world, theres only so many pathways in and out, close to the server ... and the servers, are well, off to the side of major networks, where its cheaper ... for example, someone in Europe, who gets connected to the oceanic server ( which is Singapore, about as far as you could get from the middle of oceania ) the clients only have so many potential pathways to the server, if one of the service providers along the way decide to change the active route, all connections will probably be dropped and unrecoverable, oh, and you still need to maintain a connection to the auth server, which is in a total back end of the internet, if that link goes down, nearly everyone leaves at the same time, but, not everyone ...

further compounding the issue, is that the devs only test a very narrow selection of hardware, and make no effort whatsoever to go outside of current intel / NVidia builds ... AMD FX's had issues for ages, and instead of acquiring a system for them to debug the issue, they eventually just changed the code for everybody, in an attempt to semi fix the issue, and, surprise, it didn't really work, it got slightly better, but certainly wasn't resolved ... leading to a problem around here where basically anything that was wrong was blamed on the CPU, and shielding the devs from actually diagnosing and improving the base code, which by the way, is stuck in a time warp back to 2012/2013, and seeing as weve gone this far without them having to improve much, you'll probably die of old age before it becomes a thing .... if they don't see the problem, than clearly the problem is yours, but if you look at how often patched code needs to be hot fixed, you start to get the idea, that they don't see much ....

a chunk of stuff happens during the start up sequence, and its pretty much all hidden from the user .... one great example is that your chosen settings, might become irrelevant, depending on which map was selected, some maps force values onto the client, with no regard to what that might do for the end user experience ( especially for people with settings set to 'low' ) ...

[G4LV4TR0N]

Next time it crashes you could also upload omicron error log, not that it will help but there are many people with crashes and it's quite interesting in finding what could be causing it.

[NARC BAIT]

in the default configuration the log file contains very little actual information ...