I'm betting that when it says it recovered that just means we got the picture back... which we do, just not the video from the game. So it crashes, we get the blue screen, then it recovers and we get the desktop. There are multiple identical entries of this, sometimes a few in quick succession. So maybe these are just stumbles and not actual crashes.
Under the "Application" branch of the tree I found one stretch of particularly interesting logs. Starting at 05:19:39 with a "LiveKernelEvent" message there were dozens of logs over the next several seconds, culminating in;
Log Name: Application
Source: Application Hang
Date: 9/27/2017 05:19:47
Event ID: 1002
Task Category: (101)
Level: Error
Keywords: Classic
User: N/A
Computer: DESKTOP-EVO7BM7
Description:
The program Subnautica.exe version 5.6.2.0 stopped interacting with Windows and was closed. To see if more information about the problem is available, check the problem history in the Security and Maintenance control panel.
Process ID: 58c
Start Time: 01d33779243b2308
Termination Time: 4294967295
Application Path: C:\Program Files (x86)\Steam\steamapps\common\Subnautica\Subnautica.exe
Report Id: 62740cf6-a36d-11e7-a7f8-001d60555739
Faulting package full name:
Faulting package-relative application ID:
Event Xml:
<Event xmlns="http://schemas.microsoft.com/win/2004/08/events/event">
<System>
<Provider Name="Application Hang" />
<EventID Qualifiers="0">1002</EventID>
<Level>2</Level>
<Task>101</Task>
<Keywords>0x80000000000000</Keywords>
<TimeCreated SystemTime="2017-09-27T10:19:47.751635400Z" />
<EventRecordID>61600</EventRecordID>
<Channel>Application</Channel>
<Computer>DESKTOP-EVO7BM7</Computer>
<Security />
</System>
<EventData>
<Data>Subnautica.exe</Data>
<Data>5.6.2.0</Data>
<Data>58c</Data>
<Data>01d33779243b2308</Data>
<Data>4294967295</Data>
<Data>C:\Program Files (x86)\Steam\steamapps\common\Subnautica\Subnautica.exe</Data>
<Data>62740cf6-a36d-11e7-a7f8-001d60555739</Data>
<Data>
</Data>
<Data>
</Data>
<Binary>54006F00700020006C006500760065006C002000770069006E0064006F0077002000690073002000690064006C00650000000000</Binary>
</EventData>
</Event>
There were multiple "AppHang81" errors, multiple "PnPGenericDriverFound" messages, multiple "StoreAgentDownloadFailure1" messages, multiple "PerfWatsonVS12Data" messages, a few "WindowsUpdateFailure3" messages, and an "APPCRASH" message for Radeon.exe all within a span of 7-8 seconds. It looks like it may be that Windows is trying to update behind the scenes and that causes too much draw on the system. So we get multiple crashes because Windows keeps trying to update and we keep trying to play.
You'll want something that stresses both the CPU and GPU, I'd think, because that's what a game would do. (Especially one that's not finished being optimized).
Oh. You could also try this (but if you already reset your PC, probably no need. But it can't hurt):
From an admin cmd prompt using Win+X+A (opens PowerShell, almost the same thing as cmd, although if you really need the old cmd prompt, put in cmd and enter in PowerShell and it'll open the old prompt. For these tasks, however, PowerShell will work fine.)
dism /online /cleanup-image /restorehealth
and then
sfc /scannow
EDIT: curious, what brand & model PSU did you get?
Running that reset code now. I feel all Linux-y playing in the command prompt.
The PSU is an Evga 500W unit. It's not the fanciest available. But It's a huge improvement over the Coolmax unit that was in the rig when I bought it.
EDIT:
The OCCT stress test at least appears to be very thorough. It does run both the GPU and CPU to their limits. It even has a PSU "test" where it ramps up everything to draw as much power as possible.
Also, components can have the same issues: GPU works fine under light load, but under heavy load, the voltage regulators have broken down to the point where they can't handle the normal required power to perform, and screw everything up.
@0x6A7232 This is exactly what I was thinking when he said it had started to crash with other games.
@SouthernGorilla It's possible for new stresses to degrade an old part fairly quickly. My rig is pretty old, but the video card in it was doing fine for the light duty I'd been giving it for a while. I'd stopped gaming on it for a long time, living on my XBox instead, but when I was gaming, it ran fine. Things like Starcraft 2, for instance.
Theeeen my wife got me to install Overwatch so I could play with her (she's aiming for Grandmaster Junkrat by sometime next year). It played fine for about a week and then started crashing. And then other things started crashing. And then my computer wouldn't boot... until I replaced the video card. Seems Overwatch put enough stress on the video card to cause it to break when otherwise it would have lasted a lot longer.
I'm thinking Subnautica may have done the same thing for yours, but it's just a theory.
Also, by the time you're done with this issue, you can put this kind of troubleshooting on your resume'. This is some serious low-level troubleshooting you're going at it like a champ. When I was a Microsoft contractor for several years, I lived and breathed this kind of thing and it fills me with a kind of nostalgic terror to see it again. Good times!
Also, components can have the same issues: GPU works fine under light load, but under heavy load, the voltage regulators have broken down to the point where they can't handle the normal required power to perform, and screw everything up.
@0x6A7232 This is exactly what I was thinking when he said it had started to crash with other games.
@SouthernGorilla It's possible for new stresses to degrade an old part fairly quickly. My rig is pretty old, but the video card in it was doing fine for the light duty I'd been giving it for a while. I'd stopped gaming on it for a long time, living on my XBox instead, but when I was gaming, it ran fine. Things like Starcraft 2, for instance.
Theeeen my wife got me to install Overwatch so I could play with her (she's aiming for Grandmaster Junkrat by sometime next year). It played fine for about a week and then started crashing. And then other things started crashing. And then my computer wouldn't boot... until I replaced the video card. Seems Overwatch put enough stress on the video card to cause it to break when otherwise it would have lasted a lot longer.
I'm thinking Subnautica may have done the same thing for yours, but it's just a theory.
Also, by the time you're done with this issue, you can put this kind of troubleshooting on your resume'. This is some serious low-level troubleshooting you're going at it like a champ. When I was a Microsoft contractor for several years, I lived and breathed this kind of thing and it fills me with a kind of nostalgic terror to see it again. Good times!
I could see it hurting an old card, but I bought this card specifically so I could play games. The card is *maybe* six months old. Unless playing for hours at a time has sped up the degradation process. But if GPUs wear out that easily I'm going to quit games. I can't afford to buy a new card every few months. Based on what I saw from the event viewer I tend to think it's just an issue with the rig itself being too old to keep up. It's PCI-E 1.1, which can't be very fast compared to 3.0. Plus the SATA to the HD is slower. It's a 7-yr old motherboard. There may be all manner of bottlenecks causing issues. Not to mention the CPU itself.
I didn't realize I was digging that deep with all this. I spent a decade playing with Linux on several desktops and laptops. So troubleshooting is just "normal" to me since nothing works out of the box with Linux.
I tried windowed mode and ran for a good 45 minutes or so before I got annoyed with being in windowed mode. So I went back to fullscreen, but kept the lower resolution. Then I got a crash about twenty minutes in. So maybe windowed mode would run fine or maybe the crash would have still happened. I'm thinking within the next couple weeks I'll be upgrading to an SSD. Maybe having quicker access to data will help. After that it will be a matter of waiting until I have the money to upgrade the whole rig. I just don't think this rig can move bytes like it needs to despite being overclocked.
Well, SSD is one... you can fix the SATA bottleneck by buying a SATA PCI-E card (probably not worth it, just upgrade the rig, you've got a lot of other bottlenecks).
You could also try installing, say, an older version of Windows on a separate partition and seeing if that helps (mobo not quite compatible with Win 10? I have a laptop like that, it can be a pain sometimes, but it does run 10, it just needs a lot of encouragement lol)
I don't see any problems with Windows in general. It boots promptly and doesn't seem to have any hiccups. But it is obviously a bigger strain on the system than a lightweight Linux distro would be. Maybe Windows and gaming are just a deadly combo in this case. It won't hurt to get the SSD since, as we've already mentioned, it can swap over with the rebuild. But I'm convinced there's nothing more that can be done to save this rig. It seems to be running as well as it can. It's just not up to the task.
I could see it hurting an old card, but I bought this card specifically so I could play games. The card is *maybe* six months old.
Ahah, okay. Yeah that would sink my initial theory, though it's also possible you got a defective video card, and it only causes a crash under very specific conditions. But I think that's unlikely, and as you say, there are a lot of other factors/bottlenecks we can't really test, and a new rig will probably solve it all anyway. Here's to hoping you can get it sooner rather than later.
So troubleshooting is just "normal" to me since nothing works out of the box with Linux.
Hahahah, you got that right, bro. At one of my MS jobs, I spent /seven weeks/ on a customer's Linux problem. I got so deep into the troubleshooting that I inadvertently became the team's "Linux expert", lol.
Ahah, okay. Yeah that would sink my initial theory, though it's also possible you got a defective video card, and it only causes a crash under very specific conditions. But I think that's unlikely, and as you say, there are a lot of other factors/bottlenecks we can't really test, and a new rig will probably solve it all anyway. Here's to hoping you can get it sooner rather than later.
I'm really thinking about the fullscreen question now. Every crash we've ever had has been with a fullscreen program. It's entirely possible that an RX 460 just can't drive a 42" TV at 1080p. Or maybe it can just barely handle it and the slightest hiccup crashes it. We're going to experiment this weekend and see if we can run windowed mode for an extended period. That wouldn't explain why we didn't have any problems for the first few weeks though.
Hahahah, you got that right, bro. At one of my MS jobs, I spent /seven weeks/ on a customer's Linux problem. I got so deep into the troubleshooting that I inadvertently became the team's "Linux expert", lol.
I thoroughly enjoyed Linux for the first few years. But it's become such a cluster that it's simpler to deal with Windows. The problem is too many arrogant devs think every Linux box should be built to suit their ideals rather than designing their programs to suit the inherent flexibility of the OS. So you need to have at least two different graphics toolkits and who knows how many different versions of other libraries and programs just to run the basics of a desktop. It's absurd. [/rant]
Just did something I hadn't tried before. I ran GPU-Z while playing in windowed mode so I could see what was going on in real-time. Even during the load screen and while I had the game paused the GPU usage was almost constantly pegged at 100%. The memory controller usage fluctuated pretty wildly with multiple spikes over 90%. Power draw was also pretty erratic, with spikes over 70W. To me, it does look like the card is operating at its limits and the slightest extra stress puts it over. This card doesn't have an auxiliary power input, so the power draw is perilously close to the PCI 75W limit. I don't know if these numbers are really out of the ordinary or if they're even accurate. So I may be wrong in thinking they indicate a potential problem. At least temperatures were stable and reasonable.
My R7 360 I just got has a 6-pin power connector; you're positive your RX 460 doesn't have one (I almost missed the one on my 360)? According to this, your 460 beats out my 360 by about 16%. Might be the 460 is more efficient though? I know at least some 460s require a 6-pin.
Also, Linux is cool. Try Manjaro with the KDE desktop (unless you have a DE that you prefer; Manjaro has several, but KDE is pretty flexible and popular). See if that improves your GNU experience any. My preference: Manjaro > Linux Mint Debian Edition > Ubuntu (so far, anyways) -- I'd also love to get Ubuntu's Unity DE on other distros, but the standard response is "well then just use Ubuntu" but, if I wanted Mac OS X's interface over a Windows backbone, I wouldn't appreciate being told to "just use a Mac then". You're right, a lot of Linux guys are pretty stuck up about stuff. (Side note: I started tinkering with Linux back when Mandriva came out. Fun times!) One more thing: you can run the Linux subsystem in Windows 10 now (no, it's not an emulator, yes you can run WINE on it to support old Windows programs better than Windows can, or at least that's the eventual hope)
My R7 360 I just got has a 6-pin power connector; you're positive your RX 460 doesn't have one (I almost missed the one on my 360)? According to this, your 460 beats out my 360 by about 16%. Might be the 460 is more efficient though? I know at least some 460s require a 6-pin.
Yup. I actually bought the card in part because it didn't need extra power. The original PSU didn't have an extra connection so I couldn't have hooked it up. Power may not be the issue. I'm just tossing out every idea I have at this point. We'll see if it runs in windowed mode. Then since it has a DVI connection I'll hook it to a 17" monitor I have laying around to see if it runs that. If it runs on the monitor I'll definitely know the card is just overwhelmed by the TV.
My R7 360 I just got has a 6-pin power connector; you're positive your RX 460 doesn't have one (I almost missed the one on my 360)? According to this, your 460 beats out my 360 by about 16%. Might be the 460 is more efficient though? I know at least some 460s require a 6-pin.
Yup. I actually bought the card in part because it didn't need extra power. The original PSU didn't have an extra connection so I couldn't have hooked it up. Power may not be the issue. I'm just tossing out every idea I have at this point. We'll see if it runs in windowed mode. Then since it has a DVI connection I'll hook it to a 17" monitor I have laying around to see if it runs that. If it runs on the monitor I'll definitely know the card is just overwhelmed by the TV.
If it was the TV, it would be the resolution, so you could fix it by lowering the resolution (the inches make no difference to the card; it's just putting out whatever the resolution is set to, so, for 1080p, that's 1920 pixels across x 1080 pixels high).
I guess what I'm trying to say is if your 17" is set to 1920x1080 you won't see a difference.
Upgrading to an SSD didn't fix it. We expect a decent refund on our taxes this year. A portion of that money is going to a full rebuild of the rig. That rebuild may include a new GPU. I might just build a whole new rig and then upgrade this one a bit to send to my folks.
Once upon a time, I used to build my own computer every couple of years. These days, I shop for deals and just buy pre-built computers from Dell. They have a section on their website called, I think, Dell Outlet or something where they sell refurbished computers. Yeah, yeah, I know. Who wants to buy a refurbished computer? But included in that section of the website, they actually sell three different categories of "not new":
1. Refurbished - broken computers that were then fixed
2. Scratch and Dent - minor cosmetic issues that don't affect internal function
3. Previously purchased new - computers sold and shipped to a customer and then returned without powering on
They sell the "Scratch and Dent" and "Previously purchased New" computers for the same discounted prices as the refurbished ones. So, if you can shop around and wait before buying a new PC, you can often get the equivalent of a brand new Dell PC for almost half off the regular price. That's where I bought my last 3-4 computers. I'd get either a Scratch and Dent or previously purchased new, and I've gotten 3-4 years of use out of each computer.
At the moment, I'm actually playing Subnautica on a Dell server I picked up for only $300 and then added 16 GB of memory for around $100 and an nVidia 750TI for around $130 and a Samsung 512GB SSD for $170. So, let's call it $700 for my gaming PC. It is certainly not the cheapest PC I've every bought. But it plays Subnautic just fine. I'm very happy with its performance. And, since it was a server class computer, it is definitely rock solid reliable. (If I had it to do over, 256 GB SSD would have been more than enough as my 512 is still only half full.)
I've used CrystalDiskInfo and HDD Health before, and Acronis is good. Anyways, get a SMART tool, and see if it says your drive has issues.
According to CrystalDiskInfo the drive is fine. But I don't understand most of the info it prints out.
I was going to post a screenshot but it isn't giving me the option to upload a photo. Don't feel like putting it up somewhere else just so I can link to it.
This board, for whatever reason, doesn't allow you to upload photos. The 'General' one does, but this one doesn't. You'd think the bugfix board is the one where they would *want* to allow uploading!
I've used CrystalDiskInfo and HDD Health before, and Acronis is good. Anyways, get a SMART tool, and see if it says your drive has issues.
According to CrystalDiskInfo the drive is fine. But I don't understand most of the info it prints out.
I was going to post a screenshot but it isn't giving me the option to upload a photo. Don't feel like putting it up somewhere else just so I can link to it.
This board, for whatever reason, doesn't allow you to upload photos. The 'General' one does, but this one doesn't. You'd think the bugfix board is the one where they would *want* to allow uploading!
What's really strange is you can upload the photo to a post (in progress but not yet posted) in general, copy the link, paste it to a post here and then... just not complete the post on general and the photo uploaded to the post you *didn't* complete on general will link here just fine.
I might look at the Dell option. But a huge part of why I build my own is just the enjoyment/satisfaction of the process and not the "I can do it better" factor.
I might look at the Dell option. But a huge part of why I build my own is just the enjoyment/satisfaction of the process and not the "I can do it better" factor.
For my last PC, I *wanted* to buy a Dell Home PC that could take 32 GB of RAM but the Dell Sales Rep LIED to me and said only their super high end Business models could take that much memory. So, I decided to put one together myself. I bought all the parts and then re-used a case from one of my older PC's. But I made a critical mistake and didn't remove all the motherboard standouts on the back plane of the case. These are the little things that make metal to metal contact with the insulated parts of the motherboard and that you put the screws in. I was in a hurry and put new standouts to align with the screw holes for the new motherboard but left the old ones on the back plane. I assembled the entire computer and powered it on. Oops!
After that, I decided I didn't need 32 GB of memory and bought a Dell... which as it turned out could accommodate 32 Gig just fine. *sigh*
I may try building my own PC again someday. Maybe.
I know the feeling. My first (only, so far) build I used a dual-socket server board and maxed out a credit card getting all the shiniest bits Newegg and Tiger Direct had to offer. Then I bought the cheapest PSU I could find... because PSUs don't matter, right? So I wound up with a $3,500 paperweight and a major shot to my credit when the PSU had a seizure. Needless to say, my next full build will have a very expensive PSU.
Also, for PSUs, you don't necessarily want the most expensive (although, expect to pay a fair price). You want a reliable brand. The review sites above are helpful in that regard. Also, get a surge protector with a power filter. Look for something with a lot of joules of protection. (2-3,000?) Watch the surge protection light. Once it goes out, (usually in a few years unless a big surge hits) your surge protector has finally been chewed through by micro surges and needs replacement (as it's no longer filtering the power coming in). They aren't necessarily all that expensive, either. Also, if you want, you can get a crappy 500 or 1,000 joule protector, plug that into the wall, then plug the more expensive protector into that. So hopefully the crappy one protects the expensive one and gets replaced every so often, but if a big spike hits, the heavy duty protector won't let it pass through to your equipment.
This entire thread is in a language I'm too stupid to understand
It's not stupid to be unfamiliar with jargon.
By way of updating this situation, the crashes are getting worse. I get full crashes now sometimes, lose keyboard and everything and have to do a hard reboot. It's definitely not the game or the GPU. But I'm still at a loss to figure it out. We're holding out for our tax refund to rebuild the system. Especially now that it looks like my daily driver has blown a head gasket and needs immediate repair.
This entire thread is in a language I'm too stupid to understand
It's not stupid to be unfamiliar with jargon.
By way of updating this situation, the crashes are getting worse. I get full crashes now sometimes, lose keyboard and everything and have to do a hard reboot. It's definitely not the game or the GPU. But I'm still at a loss to figure it out. We're holding out for our tax refund to rebuild the system. Especially now that it looks like my daily driver has blown a head gasket and needs immediate repair.
I forget (hey, it's been 3 pages) - have you run a Memtest on it? Make a Memtest86+ boot disk (just grab a FalconFour boot CD 4.6 + 4.61 patch, apply the patch to the 4.6 ISO) and put that on a USB key using Rufus.) and let that run overnight. (Only really need a few passes, though.)
See if that runs through a few times with no problems.
There's a built-in memory test that I have run. But I have not tried the fancy one. Mostly because it's not very convenient to use and I'm getting lazy in my old age.
Well, I need to experiment some more still... but it looks like the issue may be solved. I just played for about an hour with no issues whatsoever.
Found a video that showed where to find the crash dump files (C:/Windows/Minidump/) and how to analyze them via an online dmp file analyzer. That led me to checking for a bad "atikmpag.sys" file in the System32/drivers folder. Turns out, the file wasn't bad... it was completely missing. Copied it over from the latest version in the appropriate AMD folder and voila! I'm guessing the file didn't get installed with one of the many updates and that's why we just suddenly started having trouble. I'll mark this thread SOLVED as soon as I'm sure this did the trick. As unpredictable as the crashes are it may take a few days to be fully confident in the solution. But I'm actually optimistic for once. I'll also post links to the info I used to find this out since the actual solution may be somewhat different for others.
Comments
Log Name: System Source: Display Date: 7/25/2017 16:08:52 Event ID: 4101 Task Category: None Level: Warning Keywords: Classic User: N/A Computer: DESKTOP-EVO7BM7 Description: Display driver amdkmdap stopped responding and has successfully recovered. Event Xml: <Event xmlns="http://schemas.microsoft.com/win/2004/08/events/event"> <System> <Provider Name="Display" /> <EventID Qualifiers="0">4101</EventID> <Level>3</Level> <Task>0</Task> <Keywords>0x80000000000000</Keywords> <TimeCreated SystemTime="2017-07-25T21:08:52.682910200Z" /> <EventRecordID>4964</EventRecordID> <Channel>System</Channel> <Computer>DESKTOP-EVO7BM7</Computer> <Security /> </System> <EventData> <Data>amdkmdap</Data> <Data> </Data> </EventData> </Event>Under the "Application" branch of the tree I found one stretch of particularly interesting logs. Starting at 05:19:39 with a "LiveKernelEvent" message there were dozens of logs over the next several seconds, culminating in;
Log Name: Application Source: Application Hang Date: 9/27/2017 05:19:47 Event ID: 1002 Task Category: (101) Level: Error Keywords: Classic User: N/A Computer: DESKTOP-EVO7BM7 Description: The program Subnautica.exe version 5.6.2.0 stopped interacting with Windows and was closed. To see if more information about the problem is available, check the problem history in the Security and Maintenance control panel. Process ID: 58c Start Time: 01d33779243b2308 Termination Time: 4294967295 Application Path: C:\Program Files (x86)\Steam\steamapps\common\Subnautica\Subnautica.exe Report Id: 62740cf6-a36d-11e7-a7f8-001d60555739 Faulting package full name: Faulting package-relative application ID: Event Xml: <Event xmlns="http://schemas.microsoft.com/win/2004/08/events/event"> <System> <Provider Name="Application Hang" /> <EventID Qualifiers="0">1002</EventID> <Level>2</Level> <Task>101</Task> <Keywords>0x80000000000000</Keywords> <TimeCreated SystemTime="2017-09-27T10:19:47.751635400Z" /> <EventRecordID>61600</EventRecordID> <Channel>Application</Channel> <Computer>DESKTOP-EVO7BM7</Computer> <Security /> </System> <EventData> <Data>Subnautica.exe</Data> <Data>5.6.2.0</Data> <Data>58c</Data> <Data>01d33779243b2308</Data> <Data>4294967295</Data> <Data>C:\Program Files (x86)\Steam\steamapps\common\Subnautica\Subnautica.exe</Data> <Data>62740cf6-a36d-11e7-a7f8-001d60555739</Data> <Data> </Data> <Data> </Data> <Binary>54006F00700020006C006500760065006C002000770069006E0064006F0077002000690073002000690064006C00650000000000</Binary> </EventData> </Event>The PSU is an Evga 500W unit. It's not the fanciest available. But It's a huge improvement over the Coolmax unit that was in the rig when I bought it.
EDIT:
The OCCT stress test at least appears to be very thorough. It does run both the GPU and CPU to their limits. It even has a PSU "test" where it ramps up everything to draw as much power as possible.
@0x6A7232 This is exactly what I was thinking when he said it had started to crash with other games.
@SouthernGorilla It's possible for new stresses to degrade an old part fairly quickly. My rig is pretty old, but the video card in it was doing fine for the light duty I'd been giving it for a while. I'd stopped gaming on it for a long time, living on my XBox instead, but when I was gaming, it ran fine. Things like Starcraft 2, for instance.
Theeeen my wife got me to install Overwatch so I could play with her (she's aiming for Grandmaster Junkrat by sometime next year). It played fine for about a week and then started crashing. And then other things started crashing. And then my computer wouldn't boot... until I replaced the video card. Seems Overwatch put enough stress on the video card to cause it to break when otherwise it would have lasted a lot longer.
I'm thinking Subnautica may have done the same thing for yours, but it's just a theory.
Also, by the time you're done with this issue, you can put this kind of troubleshooting on your resume'. This is some serious low-level troubleshooting you're going at it like a champ. When I was a Microsoft contractor for several years, I lived and breathed this kind of thing and it fills me with a kind of nostalgic terror to see it again. Good times!
@SouthernGorilla
I could see it hurting an old card, but I bought this card specifically so I could play games. The card is *maybe* six months old. Unless playing for hours at a time has sped up the degradation process. But if GPUs wear out that easily I'm going to quit games. I can't afford to buy a new card every few months. Based on what I saw from the event viewer I tend to think it's just an issue with the rig itself being too old to keep up. It's PCI-E 1.1, which can't be very fast compared to 3.0. Plus the SATA to the HD is slower. It's a 7-yr old motherboard. There may be all manner of bottlenecks causing issues. Not to mention the CPU itself.
I didn't realize I was digging that deep with all this. I spent a decade playing with Linux on several desktops and laptops. So troubleshooting is just "normal" to me since nothing works out of the box with Linux.
You could also try installing, say, an older version of Windows on a separate partition and seeing if that helps (mobo not quite compatible with Win 10? I have a laptop like that, it can be a pain sometimes, but it does run 10, it just needs a lot of encouragement lol)
Ahah, okay. Yeah that would sink my initial theory, though it's also possible you got a defective video card, and it only causes a crash under very specific conditions. But I think that's unlikely, and as you say, there are a lot of other factors/bottlenecks we can't really test, and a new rig will probably solve it all anyway. Here's to hoping you can get it sooner rather than later.
Hahahah, you got that right, bro. At one of my MS jobs, I spent /seven weeks/ on a customer's Linux problem. I got so deep into the troubleshooting that I inadvertently became the team's "Linux expert", lol.
I thoroughly enjoyed Linux for the first few years. But it's become such a cluster that it's simpler to deal with Windows. The problem is too many arrogant devs think every Linux box should be built to suit their ideals rather than designing their programs to suit the inherent flexibility of the OS. So you need to have at least two different graphics toolkits and who knows how many different versions of other libraries and programs just to run the basics of a desktop. It's absurd. [/rant]
Also, Linux is cool. Try Manjaro with the KDE desktop (unless you have a DE that you prefer; Manjaro has several, but KDE is pretty flexible and popular). See if that improves your GNU experience any. My preference: Manjaro > Linux Mint Debian Edition > Ubuntu (so far, anyways) -- I'd also love to get Ubuntu's Unity DE on other distros, but the standard response is "well then just use Ubuntu" but, if I wanted Mac OS X's interface over a Windows backbone, I wouldn't appreciate being told to "just use a Mac then". You're right, a lot of Linux guys are pretty stuck up about stuff. (Side note: I started tinkering with Linux back when Mandriva came out. Fun times!) One more thing: you can run the Linux subsystem in Windows 10 now (no, it's not an emulator, yes you can run WINE on it to support old Windows programs better than Windows can, or at least that's the eventual hope)
If it was the TV, it would be the resolution, so you could fix it by lowering the resolution (the inches make no difference to the card; it's just putting out whatever the resolution is set to, so, for 1080p, that's 1920 pixels across x 1080 pixels high).
I guess what I'm trying to say is if your 17" is set to 1920x1080 you won't see a difference.
Could try halving the res, and using some sort of smoothing (antialiasing).
1. Refurbished - broken computers that were then fixed
2. Scratch and Dent - minor cosmetic issues that don't affect internal function
3. Previously purchased new - computers sold and shipped to a customer and then returned without powering on
They sell the "Scratch and Dent" and "Previously purchased New" computers for the same discounted prices as the refurbished ones. So, if you can shop around and wait before buying a new PC, you can often get the equivalent of a brand new Dell PC for almost half off the regular price. That's where I bought my last 3-4 computers. I'd get either a Scratch and Dent or previously purchased new, and I've gotten 3-4 years of use out of each computer.
At the moment, I'm actually playing Subnautica on a Dell server I picked up for only $300 and then added 16 GB of memory for around $100 and an nVidia 750TI for around $130 and a Samsung 512GB SSD for $170. So, let's call it $700 for my gaming PC. It is certainly not the cheapest PC I've every bought. But it plays Subnautic just fine. I'm very happy with its performance. And, since it was a server class computer, it is definitely rock solid reliable. (If I had it to do over, 256 GB SSD would have been more than enough as my 512 is still only half full.)
This board, for whatever reason, doesn't allow you to upload photos. The 'General' one does, but this one doesn't. You'd think the bugfix board is the one where they would *want* to allow uploading!
What's really strange is you can upload the photo to a post (in progress but not yet posted) in general, copy the link, paste it to a post here and then... just not complete the post on general and the photo uploaded to the post you *didn't* complete on general will link here just fine.
For my last PC, I *wanted* to buy a Dell Home PC that could take 32 GB of RAM but the Dell Sales Rep LIED to me and said only their super high end Business models could take that much memory. So, I decided to put one together myself. I bought all the parts and then re-used a case from one of my older PC's. But I made a critical mistake and didn't remove all the motherboard standouts on the back plane of the case. These are the little things that make metal to metal contact with the insulated parts of the motherboard and that you put the screws in. I was in a hurry and put new standouts to align with the screw holes for the new motherboard but left the old ones on the back plane. I assembled the entire computer and powered it on. Oops!
After that, I decided I didn't need 32 GB of memory and bought a Dell... which as it turned out could accommodate 32 Gig just fine. *sigh*
I may try building my own PC again someday. Maybe.
Also, for PSUs, you don't necessarily want the most expensive (although, expect to pay a fair price). You want a reliable brand. The review sites above are helpful in that regard. Also, get a surge protector with a power filter. Look for something with a lot of joules of protection. (2-3,000?) Watch the surge protection light. Once it goes out, (usually in a few years unless a big surge hits) your surge protector has finally been chewed through by micro surges and needs replacement (as it's no longer filtering the power coming in). They aren't necessarily all that expensive, either. Also, if you want, you can get a crappy 500 or 1,000 joule protector, plug that into the wall, then plug the more expensive protector into that. So hopefully the crappy one protects the expensive one and gets replaced every so often, but if a big spike hits, the heavy duty protector won't let it pass through to your equipment.
By way of updating this situation, the crashes are getting worse. I get full crashes now sometimes, lose keyboard and everything and have to do a hard reboot. It's definitely not the game or the GPU. But I'm still at a loss to figure it out. We're holding out for our tax refund to rebuild the system. Especially now that it looks like my daily driver has blown a head gasket and needs immediate repair.
I forget (hey, it's been 3 pages) - have you run a Memtest on it? Make a Memtest86+ boot disk (just grab a FalconFour boot CD 4.6 + 4.61 patch, apply the patch to the 4.6 ISO) and put that on a USB key using Rufus.) and let that run overnight. (Only really need a few passes, though.)
See if that runs through a few times with no problems.
Found a video that showed where to find the crash dump files (C:/Windows/Minidump/) and how to analyze them via an online dmp file analyzer. That led me to checking for a bad "atikmpag.sys" file in the System32/drivers folder. Turns out, the file wasn't bad... it was completely missing. Copied it over from the latest version in the appropriate AMD folder and voila! I'm guessing the file didn't get installed with one of the many updates and that's why we just suddenly started having trouble. I'll mark this thread SOLVED as soon as I'm sure this did the trick. As unpredictable as the crashes are it may take a few days to be fully confident in the solution. But I'm actually optimistic for once. I'll also post links to the info I used to find this out since the actual solution may be somewhat different for others.