PC keeps cutting out when livestreaming

[d3anio97]

So I have a slight issue that literally only affects my system when I'm livestreaming. This isn't the first time it's happened and it's also been literally IMPOSSIBLE to get a straight answer from anyone anywhere... Hoping here will be better because I know this forum's extremely competent with regards to gaming and streaming as well as troubleshooting some pretty damn complex issues. This hopefully being one of them. Here's a little bit of info on my config;

NOTE: This will be a bit long but I've also done a LOT of troubleshooting with this issue in the past, so please bear with me...

here's my configuration;

SOFTWARE:
OBS Studio 0.14
Windows 10 64-bit

(Info about the render: Using the AMD AMF H.264 encoder at a 6600KBPS bitrate at 1920 x 1080 60FPS)

HARDWARE: 
CPU: AMD FX8150 (NO OC)
MB: Gigabyte 970A-DS3P
GPU: AMD Radeon  HD7950 (Crossfire, 2 GPUs)
Disks: 1x WD Red 1TB (WD10EFRX), 1x WD Blue 2TB (WD20EZRZ), Toshiba Q300 SSD (Boot)
PSU: Corsair RM750

Network: TPLink TL-851ND
RAM: 16GB DDR3-1333
Other information:

I verified this system under AIDA64 and the temperatures are (With an OUTSIDE ambient of 25*c)

GPU0: 80*c
GPU1: 75*c
CPU: 56*c

This eliminates overheating from this issue.

The power supply is 750W and also from a reputable brand and has a filtered intake. It can't possibly be the issue or the system would have powered off while stressing the entire system with AIDA64.


Here's a list of what I've already done with this machine to try and make it stream for more than 5 minutes before abruptly powering off:

I've checked the system error logs and there were no kernel panics or severe errors that could have caused the shutdown

Also, I'm able to record until I fill a drive with OBS, but it's just streaming that has trouble. When the system powers down, the logs for OBS show no abnormalities whatsoever.

I have plans of looking into altering my networking settings and seeing if that resolves the issue... However, from past experience with this WiFi card, I have good reason to believe it's NOT the issue... It doesn't cause a problem in Windows, it's not shorted out on the board (Or I'd have no WiFi!) and it handles stress testing well (I DOS'd my laptop to make sure the card was working properly and it handled a sustained UDP stream of 300Mb/s for 6 hours)

Also, I've tried multiple different power supplies, updating graphics drivers with a clean install, removing and swapping out memory and GPUs... all to no avail.

On the note of memory: I ran it for 3 days under MemTest86+ and it passed with flying colors.

Also, I tried underclocking my GPUs to 810MHz and dropping the power limits... Again, this gave me no positive result... well, besides a 2*c drop in idle temps and a whopping 9*c in load temps!

The CPU is at stock frequencies and hasn't been overclocked much... I tried to get it past 4.3GHz once and even after playing with the HTLink and NB multipliers, I couldn't get the machine to remain stable, so I backed the OC off back to stock (3600 to 3900 on my specific chip), and again... after testing it for 48 hours straight, I can confirm with absolute confidence that it is not the culprit!

I've pretty much exhausted every piece of advice I can get and I'm completely at my wit's end with this machine. If anyone can give me a hand with this pos I'll be eternally in your debt!

[d3anio97]

MENTAL NOTE: I should recheck and verify some BIOS settings... if I get a resolution from that then I'll let y'all know.. but if not, I'll be back!

[d3anio97]

49 minutes ago, d3anio97 said:

MENTAL NOTE: I should recheck and verify some BIOS settings... if I get a resolution from that then I'll let y'all know.. but if not, I'll be back!

I can now 100% say for sure it's not a temperature issue. Disabling thermal limits did nothing at all to resolve the issue... and I'm not prepared to intentionally pin my CPU at 60+*c to see if it causes a shutdown... that's just a plain bad idea!

[Captain_WD]

7 hours ago, d3anio97 said:

~snip~

Hey there d3anio97

 

I'm sorry to hear about the issues you are going through. 

So, just to make sure we are on the same page, your system shuts down ONLY when you are streaming your gameplay? Does it simply restart or it shuts down as if you pulled the plug? Do you have any error messages? Does it start back up on its own or do you have to manually start it? 

 

Just to be on the safe side, I'd run diagnostic tests on the storage drives so you know your data is safe and you can safely exclude them as potential causes of the situation. I'd get tools from the manufacturers and see if the drives pass all the tests. For the WD drives that would be WD Data Lifeguard Diagnostic. 

 

Have you monitored the load on your parts while streaming? Maybe something is hitting 100% usage and causes the issue? 

 

If I'm not mistaken your computer should be consuming around 660W under load. Do you have a custom CPU cooler or any other parts connected to your computer? This wattage is rather close to the maximum supported by your PSU so during higher loads your system may be getting its power cut due to overloading the PSU. If you have one with more wattage I'd try that just to verify the results and completely exclude the PSU as the potential cause of the issue. 

 

Checking if all cables are plugged all the way and reseating all your parts on the motherboard may be a good idea. 

 

Checking the system log and Event Viewer may give you some info on the reason behind the shutdowns. 

 

Captain_WD.

[d3anio97]

7 hours ago, Captain_WD said:

Hey there d3anio97

 

I'm sorry to hear about the issues you are going through. 

So, just to make sure we are on the same page, your system shuts down ONLY when you are streaming your gameplay? Does it simply restart or it shuts down as if you pulled the plug? Do you have any error messages? Does it start back up on its own or do you have to manually start it? 

 

Just to be on the safe side, I'd run diagnostic tests on the storage drives so you know your data is safe and you can safely exclude them as potential causes of the situation. I'd get tools from the manufacturers and see if the drives pass all the tests. For the WD drives that would be WD Data Lifeguard Diagnostic. 

 

Have you monitored the load on your parts while streaming? Maybe something is hitting 100% usage and causes the issue? 

 

If I'm not mistaken your computer should be consuming around 660W under load. Do you have a custom CPU cooler or any other parts connected to your computer? This wattage is rather close to the maximum supported by your PSU so during higher loads your system may be getting its power cut due to overloading the PSU. If you have one with more wattage I'd try that just to verify the results and completely exclude the PSU as the potential cause of the issue. 

 

Checking if all cables are plugged all the way and reseating all your parts on the motherboard may be a good idea. 

 

Checking the system log and Event Viewer may give you some info on the reason behind the shutdowns. 

 

Captain_WD.

Yeah, it shuts down as if the plug's been pulled whenever I start streaming... Even when I'm not in games!

> The drives all pass a SMART test with 100% health, and the SSD is almost brand new (Not even 2 months old) with the oldest disk being 1 year old.

> Average system loads vary a slight amount, but generally, OBS takes up 8% total CPU time and 10% of the GPU. Disk utilization is never above 50% while streaming. Global load sits around 50% on the CPU and 99% on GPU 0, while the load on GPU 1 varies depending on if the game's in Crossfire or not.

> I use a Hyper212 on the CPU, the GPUs are non-Reference (GigaByte Windforce and a Sapphire DualX) and a whacktonne of case fans to shift as much air as possible (Probably takes up about 8 more W with the controller).

> As for power, I've tried disabling and removing graphics cards before now and the same issue happens. I've also stress tested everything (including disks) in AIDA64, the total power consumption at the wall was at 680W and the system ran the entire test period without an error (this was a good 4 hours)
> Everything was checked and reseated multiple times each time this issue occurred... All the cables are solidly plugged in

> Event log shows absolutely nothing with regards to shutdown issues. I've looked for kernel panics or anything that would be attributed to the issue, and there's nothing that could be linked to the problem. The worst thing in there is the system clock adjusting... again, harmless at worst!

[d3anio97]

UPDATE: Apparently it's affecting Sony Vegas now!! I'm gonna go try doing a clean install of my GPU drivers and see if anything changes  what a pain in the arse!

[d3anio97]

On 08/07/2016 at 9:06 PM, d3anio97 said:

UPDATE: Apparently it's affecting Sony Vegas now!! I'm gonna go try doing a clean install of my GPU drivers and see if anything changes  what a pain in the arse!

The issue with Vegas is now fixed after a fresh driver install... MAYBE OBS is working now? (Not counting my hens yet!!)

[Keesler]

Did it work? 

[Keesler]

Saw these under SLI problems figure I post them

[d3anio97]

On 08/07/2016 at 9:06 PM, d3anio97 said:

12 hours ago, Keesler said:

Did it work? 

The troubleshooting steps mostly don't apply here because I have AMD GPUs, and the featuresets are slightly different. the MA scaling is a little better with OBS and the AMF capture compared to NVENC with SLI. HOWEVER... after resetting and tinkering with the render settings, I got one version of OBS to work for 20 minutes and no shutdown... I'm gonna test more extensively with both OBS and OBS studio before I mark this as solved... but it's looking promising so far!

[Keesler]

5 hours ago, d3anio97 said:

 

Great to hear Good Luck!

[Captain_WD]

On 8.07.2016 г. at 6:23 PM, d3anio97 said:

~snip~

The age of a storage drive doesn't necessarily mean it is functioning properly. If the drives pass their tests you should be able to safely exclude them as the potential cause of the problem.

 

The 99% load on the GPU may have something in common. Try working without the multiple GPU configuration and see if the crashes persist. 

 

I do hope the problem is fixed with the proper drivers. Multi-GPU setups do lead to such problems at times.

 

Post back if you need more help! 

 

Captain_WD.

[d3anio97]

On 7/11/2016 at 0:10 PM, Captain_WD said:

The age of a storage drive doesn't necessarily mean it is functioning properly. If the drives pass their tests you should be able to safely exclude them as the potential cause of the problem.

 

The 99% load on the GPU may have something in common. Try working without the multiple GPU configuration and see if the crashes persist. 

 

I do hope the problem is fixed with the proper drivers. Multi-GPU setups do lead to such problems at times.

 

Post back if you need more help! 

 

Captain_WD.

ok, so I have the system "idling" (i.e no games running, just streaming a single screen to Twitch), Attempt 1 caused the system to power off literally 5 seconds into streaming, and the seconds attempt (after having to run a repair because the shutdown lead to a crash at boot) seems to be doing okay... I'm 36 minutes in as of posting this and I've not (fingers crossed) had the system power off yet!! I'll report back if everything turns to $h*t

[d3anio97]

ok, UPDATE:

40 minutes in and it just died again... I might try reinstalling Windows or refreshing it and see if anything changes

 

 

[samiscool51]

from what i can tell is your power supply is not capable of handling your system when doing capable loads...

it's hard to explain why this happens but with every computer being different it's hard to pin point what is the problem

if you have another PSU use that and if you are getting the same results then it might be something else

[d3anio97]

1 hour ago, samiscool51 said:

from what i can tell is your power supply is not capable of handling your system when doing capable loads...

it's hard to explain why this happens but with every computer being different it's hard to pin point what is the problem

if you have another PSU use that and if you are getting the same results then it might be something else

I highly doubt it's power supply related, and the reason I can verify that has just happened...

1: The system has a 750W power supply. the system draws 85-680W from the wall (that's idle to load) and the power supply can handle a power input of about 780W with an output of 750 continuous.

2: the system has just cut out again, but this time, I was browsing Facebook when it happened. This has lead me to the suspicion that it's related to the latching circuit that sends +5VSB to ground, which should trigger PS_ON and send a PS_OK signal to the motherboard... which it does, but there's a high chance there's an issue with that section of the board since the entire thing loses all power for a split second before the USB ports get their +5V back (The USB ports are powered via +5VSB when in standby mode.

Now, that doesn't mean I'm not gonna look at this more in-depth. I have a spare power supply that's got EPS and 20+4 power, so I'm gonna use that to power the board, then all my drives and the GPUs will be powered off the main power supply. If I lose power to the disks and GPUs then I know there is a fault with my power supply. I'll probably look into having it RMA'd if that turns out to be the issue. If not, and the known working PSU has the same issue, then I know it's an issue with the PS_on latching circuit and I need a new board... let's see how things go!!... #990FXHYPE XDD

[d3anio97]

OKAY... I CAN NOW 100% WITHOUT A SHADOW OF A DOUBT CONFIRM THAT THE PROBLEM IS MOTHERBOARD RELATED. This means 2 things:

1: I'm totally screwed for 6 whole Weeks!!
2: I get a motherboard upgrade

But the fact I'm getting a board upgrade on the same CPU is depressing T^T I at least wanted to wait for AMD's "Zen" chips to arrive before I jumped the gun and get a new board!

As for confirmation of the issue?:

I took 2 power supplies. The one in my rig and a spare I had lying around that I knew works 100%. I had my main supply powering the GPUs, hard disks and the secondary specifically powering the motherboard... Whichever power supply cut out first would either show a different issue or confirm my belief... The result was that after 30 seconds of AIDA64... the power supply controlling the motherboard lost power!! This reinforced my thought that the motherboard had issues... However, the issue appears to manifest itself as a PCI-e VRM issue and not a PS_ON latch issue like I initially thought!! This means I'm now gonna have to get a new board so I can use the system again. Wish me luck on not gong stir-crazy without my fix of 1440p Goat Simulator!!

[samiscool51]

I'm sorry that it's your mother board and that you have to get a new one, it could be worse the PSU could fail and kill the other components 

[d3anio97]

17 minutes ago, samiscool51 said:

I'm sorry that it's your mother board and that you have to get a new one, it could be worse the PSU could fail and kill the other components 

Yeah... the brightside is that my hard disks, SSD, fan controller and both my GPUs live to livestream again XDD It's not too hard to find a cheap 990FX board nowadays though, so it shouldn't be more than a few weeks or so before I have BLU3_5N0W (That's my system's codename btw) back in fighting form

[samiscool51]

[Captain_WD]

13 hours ago, d3anio97 said:

~snip~

Hey, I do hope the issue was finally identified. Troubleshooting and identifying a motherboard as the cause of the issue is a hard and irritating thing so I do hope replacing it fixes everything. Post back if you have any development!

 

Captain_WD.

[d3anio97]

7 hours ago, Captain_WD said:

Hey, I do hope the issue was finally identified. Troubleshooting and identifying a motherboard as the cause of the issue is a hard and irritating thing so I do hope replacing it fixes everything. Post back if you have any development!

 

Captain_WD.

Yeah, it was defo board related. I have the GPUs and power supply in a different machine and they're humming away without a problem... Only downside is I'm stuck with a Core2 Duo E6850 and 4GB DDR2 for 6 weeks!! At least I can still play my less CPU intensive titles in 1440p though!! That would bug me to hell and back if I was stuck without Goat Simulator XDD