[0.17.32] 5th Steam Engine Placement causes hard crash, computer reboots.

[xLocke]

Log File: https://pastebin.com/Fwf4qtpK

This has happened consistently since 0.17.
Whenever I place 5 or more Steam engines, usually on the 5th one, my computer will shut down suddenly, like the power supply was pulled or something. I did not have this issue during .16, so I suspect this has to do with fluid handling. The bug happens 100% of the time when water is connected to the boiler which is connected to the steam engine. It does not need to necessarily have power itself, or coal.

Right before I added the next boiler and steam engine: https://imgur.com/3V1zNsa

In the Event Viewer on Windows, I get the following:

Log Name: System
Source: Microsoft-Windows-Kernel-Power
Date: 4/19/2019 8:56:01 AM
Event ID: 41
Task Category: (63)
Level: Critical
Keywords: (70368744177664),(2)
User: SYSTEM
Computer: [user]
Description:
The system has rebooted without cleanly shutting down first. This error could be caused if the system stopped responding, crashed, or lost power unexpectedly.
Event Xml:
<Event xmlns="http://schemas.microsoft.com/win/2004/08/events/event">
<System>
<Provider Name="Microsoft-Windows-Kernel-Power" Guid="{331C3B3A-2005-44C2-AC5E-77220C37D6B4}" />
<EventID>41</EventID>
<Version>6</Version>
<Level>1</Level>
<Task>63</Task>
<Opcode>0</Opcode>
<Keywords>0x8000400000000002</Keywords>
<TimeCreated SystemTime="2019-04-19T12:56:01.752052100Z" />
<EventRecordID>15276</EventRecordID>
<Correlation />
<Execution ProcessID="4" ThreadID="8" />
<Channel>System</Channel>
<Computer>[user]</Computer>
<Security UserID="S-1-5-18" />
</System>
<EventData>
<Data Name="BugcheckCode">0</Data>
<Data Name="BugcheckParameter1">0x0</Data>
<Data Name="BugcheckParameter2">0x0</Data>
<Data Name="BugcheckParameter3">0x0</Data>
<Data Name="BugcheckParameter4">0x0</Data>
<Data Name="SleepInProgress">0</Data>
<Data Name="PowerButtonTimestamp">0</Data>
<Data Name="BootAppStatus">0</Data>
<Data Name="Checkpoint">0</Data>
<Data Name="ConnectedStandbyInProgress">false</Data>
<Data Name="SystemSleepTransitionsToOn">0</Data>
<Data Name="CsEntryScenarioInstanceId">0</Data>
<Data Name="BugcheckInfoFromEFI">false</Data>
<Data Name="CheckpointStatus">0</Data>
</EventData>
</Event>

I have done the following to rule out a hardware issue:
12 hour RAM stress testing.
6 hour GPU stress testing.
12 hour CPU stress testing.

No issues were logged.

Here are my PC specs:


Processor Intel(R) Core(TM) i5-4690K CPU @ 3.50GHz
Manufacturer Intel
Speed 3.5 GHz
Number of Cores 4
CPU ID BFEBFBFF000306C3
Family 06
Model 3C
Stepping 3
Revision

Video Card NVIDIA GeForce GTX 980 Ti 6GB

Video Card #2 Intel(R) HD Graphics 4600

RAM 16 GB

Operating System Windows 10

Motherboard: Z97X-Gaming 5

[Dominik]

Hi,
Factorio can't turn off your computer. You either have faulty hardware or a girlfriend that pulls your power cable whenever you play too much.

[xLocke]

I have also attached the save file. There are no mods.

I assure you, no Girlfriend.

And, unfortunately, I have consistently reproduced this bug over and over, and no other application causes it.

If it is a hardware error, I don't know where it is coming from. My temps are all groovy.

[Ultros]

Your save contains 3 boilers and 6 steam engines though, which means that the save succeeded before your computer crashed?

In either case, loading your save had no issues and I could build another 10 or so steam engines without any crashes. It's almost certainly a hardware problem.

[xLocke]

I realized I had RSO still checked for some reason. It crashed a moment after I managed the save, but had not yet hooked up the coal. I kept placing them one by one and then when I hooked up the coal and they started producing steam, it crashed. So it seems then that it might be a particular interaction between Factorio and something in my system?

[TruePikachu]

Firstly, if you're overclocking either the CPU or GPU, STOP!, at least until the cause of the issue can be determined.

Do you get the bugcheck screen (BSoD) or does the system just reboot? I ask because the event you listed above doesn't contain any bugcheck.

If you get a bugcheck screen, please upload a picture of it. Sometimes, Windows fails to record bugcheck information to the event log, but it should almost always be displayed on-screen. You might need to change a setting to prevent the system from automatically rebooting in the case of bugcheck, however.

If the system just reboots, it's probably either a hardware or driver problem which Factorio is not the cause of (or the system is configured to reboot on bugcheck -- see above). Potential hardware culprits could include the PSU (ironically) -- consider running a multithreaded CPU stress test at the same time as a GPU stress test for at least 10 minutes, to see if their combined power draw exceeds what the PSU can put out.

I'm 99% certain this will be 1/0, though.

[steinio]

Power Unit fails or is overloaded...?

[Dominik]

The most special thing about the steam engines are not the fluids but the smoke. So I would look in the GPU direction.

[Bilka]

Moving this to technical help to make the thread more visible so that you can hopefully figure out what is wrong with your PC.

[xLocke]

There is no bugcheck screen. It is as if I were to just flip the power switch on my PC, the whole thing shuts down and then the whole thing reboots. No blue screen or graphical overload (the weird color bar thing). It just turns off.

I will now attempt to disable smoke effects and see if that does anything.

The PSU might be getting overloaded, but its a 750 watt PSU in a system that only requires about 600.

I definitely do not overclock anything.

I will do a GPU and CPU stress test at the same time.

It just feels crazy that this would happen on Factorio of all games when I can run Total War Warhammer at max settings and unit sizes with no issues.

[xLocke]

OKay I ran the simultaneous stress tests and I had no issues with power or shut downs for about 20 minutes. I used Furmark and Prime95.

So I disabled smoke. And for a few minutes, I thought it had done it. I managed to place down several more burners and powerplants.

Then it crashed again.

And then this happened. https://imgur.com/KXuevjb

I have linked the log file and attached the save. https://pastebin.com/ETzE5k3S

[Ultros]

OKay I ran the simultaneous stress tests and I had no issues with power or shut downs for about 20 minutes. I used Furmark and Prime95.

So I disabled smoke. And for a few minutes, I thought it had done it. I managed to place down several more burners and powerplants.

Then it crashed again.

And then this happened. https://imgur.com/KXuevjb

I have linked the log file and attached the save. https://pastebin.com/ETzE5k3S

Your save file is corrupted and contains only zeros.It probably got messed up when the game crashed. You need to restore a backup.

[TruePikachu]

Another thing to try: run the game in OpenGL mode instead of DX11. This can be done by changing the line

Code: Select all

; force-opengl=false

(in %APPDATA%/Factorio/config/config.ini) to

Code: Select all

force-opengl=true

(note also the removal of the semicolon).

This can be used to isolate the DX9->DX11 migration from being related to the crash.

[xLocke]

Unfortunately, no luck with the OpenGL. Still shut down a few minutes later.

[PunPun]

How instant is the shutdown? Is there a momentary blackscreen before the computer turns off or does the screen go black at the same time as the computer? My educated guess is that if the screen goes black first then it is very likely that there is something wrong with the gpu or the power supplied to the gpu. Either way it sounds like it happens when the gpu goes from one powerstate to another while under load. There is a brief inrush of power at that moment that exceeds the limits of a fuse and trips either in the gpu or in the psu. If it is in the gpu then the rest of the computer will stay on for a moment where it tries to restart the gpu and when that fails it will fallback to restarting the whole system. If it's in the psu then all of the computer will shutdown at the same time.

If you go to nvidia control panel->manage 3d settings and change power management mode to different settings does it change the behaviour of the crash in anyway?

[xLocke]

It shuts down instantly. No cut to black screen, just suddenly off (I have my tower next to me, so I can see the GPU's lights flick off for a second as the screen does, then it boots back up. This definitely feels like a PSU issue. I fiddled with the GPU power settings but it made no difference. Still shut down.

[PunPun]

The chances of it being the psu does indeed look high. The problem could still be somewhere else. Just to be certain. Try and unplug all the front panel IO aka usb ports, headphone jacks, leds, reset switch. Basically anything that has a cable to your motherboard except the cpu fan and harddrives. I've seen a computer that would not boot if something was plugged in the headphone jack in the front panel. I also have seen one that would exhibit the same symptoms you describe that was caused by a faulty resetswitch. One of the pins in the switch side was contancting the case and was grounded and when the gpu started to pull more amps it would raise the voltage potential of the case just enough to cause the motherboard to think the reset switch was pressed. Just be carefull when you bridge the powerbutton contacts to boot the pc. Shouldn't be too hard if you have proper lighting and a short screwdriver.

If it still crashes then at this point I would try hooking another powersupply to test. No need to put it into the case just plug the cables in and do a short testrun. If you don't have any extra powersupplies you could go to an store that sells them and ask if you can buy one and return it if it doesn't fix the problem. The other choise would be if you have friends that have gaming computers and live close enough that you could take your tower there and try out the easily swappable parts in each others computers to see if you can identify the problem.

Also note that the test powersupply doesn't actually need to be powerfull enough to handle the cpu&gpu at full load. Just powerfull enough that you can test if the crashing in factorio is fixed and it will help narrow it down.

[Jon8RFC]

TLDR;
Anyway, does your card have a dual BIOS? A quick google shows that the GTX 980 has at least a Gigabyte and EVGA card with dual BIOS, but there may be others. If so, power off, flip that switch, power on, and see what happens.


I want to chime in on a magnificently and beautifully stupid implementation that an AMD card manufacturer used, and I spent a couple of weeks trying to figure it out with a friend, remotely.

Their card had a "neat feature" (pronounced with the most derogatory, sarcastic tone) of dual BIOSes for cryptocurrency crunching or gaming which wasn't blatantly described by the manufacturer...and which I didn't know, and was wanting to blame bad VRAM. Their card did NOT come with any blatant, brightly-colored paper when they opened the box which should've said "Please note that we've set this gaming card's BIOS to the cryptocurrency BIOS, which will crash 100% of the time when it goes over a certain vram usage threshold--enjoy!" Things like that are typically done for unusual or deal-breaking situations, but I guess a very highly-paid employee opted not to do that. Flipped the switch, and all was well.

Theirs would run games, for a short while, and crash at predictable points upon loading a map, or moving the mouse after loading. Seems a lot like what you're experiencing with the predictability. We did individual stick ram testing, ran prime95, 3dmark, confirmed very cool temps, ran sfc, glossed over event viewer (absolutely nothing helpful), did stupid graphics voltage tweaks at the suggestion of forum posts, and did a few other "why would you even test that?!" things at my request. I was ready to start blaming the power supply since we had already replaced the "faulty" graphics card (because I confidently blamed bad VRAM) and the old card was fine under all circumstances. Theirs would be ok, go black, power off, and power back on...just like yours. Nope, not the trusty Corsair PSU, just a dual BIOS switch on the GPU.