Ryzen bsod

[johndms]

The only other thing I can think of is a reinstallation of Windows. Maybe reinstall video drivers.

[johndms]

2 minutes ago, kk027 said:

Ok. I just wanted to know why they paired a high end CPU and an RX 560.

I may have been a little harsh, but this thread's already full of people going off topic and asking their own questions and seeking help for their own problems. My apologies.

[Entermaker]

30 minutes ago, kk027 said:

Why do you have a Ryzen 7 and a 560?

Because I'm doing programming and running VMs. I don't need much video processing power.

[Entermaker]

I have already reinstalled Windows 7 like 5 times. I always update AMD Chipset Drivers and Radeon Drivers as well.

 

I don't have any other powerful PSU, just a 300W old Codegen. Even if it was powerful enough I still don't trust it.

 

I'm downloading Unigine Valley.

 

I shut down the system a little bit. When I started it, after the boot logo the screen goes black and the debug display shows 00.

I pressed reset and then it started up correctly without changing anything in the bios.

[johndms]

7 minutes ago, Entermaker said:

I shut down the system a little bit. When I started it, after the boot logo the screen goes black and the debug display shows 00.

I pressed reset and then it started up correctly without changing anything in the bios.

Yeah, it seems many Ryzen boards have a "cold boot" bug. I'm not entirely sure what causes it, but if you shut the PC down completely, there's a chance it'll go through it's normal power cycles and may wipe the overclocks. I've noticed this happens to me when I'm using a memory overclock higher than my rated ram speed of 3000. 2933 will boot fine, 3200 may power cycle and wipe overclock.

I *think* it has to do with the bios/board forgetting how much DRAM Voltage is necessary to post with the ram overclocked. I think instead of trying what you've manually entered (1.35, 1.40), it tries 1.20 and fails. I've read some boards have a DRAM Boot Voltage, but mine doesn't. This is mostly just guess work, though. I have gotten into the habit of just putting my PC to sleep instead of powering off.

 

Have you considered trying Windows 10 just to see if it causes any blue screens? You can download the full version from Microsoft and not need a key. Just don't activate it and it's fully functional. Well, I think you can't change colors, or desktop wallpaper or something, but many choose that route. The whole "Windows 7 not supported" comes to mind.

[Entermaker]

You mentioned posting problems because of the cold boot bug, but my system posts perfectly. The black screen happens after Windows starts loading (after the Win7 boot animation).

So I'm not sure if its a cold bug.

 

I ran Unigine Valley for an hour and a half, and no problem.

 

I will try Win10 to eliminate possible software/driver failures (if there is any).

[johndms]

3 minutes ago, Entermaker said:

You mentioned posting problems because of the cold boot bug, but my system posts perfectly. The black screen happens after Windows starts loading (after the Win7 boot animation).

So I'm not sure if its a cold bug.

yeah, I mis-read what you said. I kinda skimmed over not paying good enough attention. I saw "boot logo" and assumed you meant motherboard logo before post.

[Santo22]

do you have another PSU you can try out? It doesn't really seem like a power delivery issue, but anything is possible and process of elimination is the only way to figure this out. 

[Entermaker]

38 minutes ago, Santo22 said:

do you have another PSU you can try out? It doesn't really seem like a power delivery issue, but anything is possible and process of elimination is the only way to figure this out. 

Unfortunately I don't have another PSU.

[enfinders]

Why don't you use a monitoring program and log everything to file? Start it, make sure it's logging every single possible parameter (temps, voltages, usage) and start playing or start a stress test. this way you will have a more complete view of what's happening. Make sure it's logging to file at least twice a second. this way if it crashes at least you have a log saved to your hdd/ssd.

[Entermaker]

Good idea.

This is exactly what I'm trying to do.

[Entermaker]

I spent a lot of time trying but I can't make the system produce bsod while running Aida64 monitoring.

I don't really understand.

Anyway, I captured a lot of system info, so hopefully it can be found out if the PSU is the problem.

 

I did one more Aida64 stability test, but this time I took more screenshots about the failure.

 

 

I also attach some pics about the voltage fluctuation while using the system.

 

 

Is there anything else I should upload?

 

Can you help me find out if it's acceptable or not?

 

Thanks in advance.

[Santo22]

Your CPU Voltage is very low. I would assume this is probably issue right here. Remind me, are you overclocking or is bios at default? My stock voltage on Ryzen 7 1700x is about 1.3-1.4. 

[Entermaker]

Everything is default except the memory speed.

[Santo22]

do you have ryzen master or HWMonitor installed? I would like to know what that is showing for voltage

[Salacious B Crum]

11 minutes ago, Entermaker said:

Everything is default except the memory speed.

Have you cleared the cmos? it may be retaining a corruption from a previous issue also it looks like based on your screenshots that you're stressing several parameters at the same time maybe try one at a time?

[Entermaker]

Yes I have HWMonitor installed, but I'm in college so I can't do it now.

 

Yes I tryed both clearing the CMOS and restoring bios defaults.

 

I have already tested the components individually for six hours and no problem. The fail occurs when I test them together.

[johndms]

2 hours ago, Entermaker said:

Yes I have HWMonitor installed, but I'm in college so I can't do it now.

 

Yes I tryed both clearing the CMOS and restoring bios defaults.

 

I have already tested the components individually for six hours and no problem. The fail occurs when I test them together.

So this still hasn't been figured out? If I remember correctly, you're about to stress CPU/FPU/Cache for several hours and it works fine. You're able to stress the GPU for several hours and it works fine. But the moment you add the GPU to the CPU/FPU/Cache test, it fails? This sounds like the PSU simply can't keep up with the power demands. You did mention it was making a noise. The low Core Voltage makes me curious.

 

https://seasonic.com/product/g-650/ - If this is your PSU, it does have a 5 year warranty. Maybe there's a local computer shop that can test your system. Use one of their PSUs to run a stress test with GPU CPU/FPU/Cache. It it doesn't fail, then RMA that PSU if it's still under warranty. But depending on what they charge, it may be cheaper just to purchase a Corsair CX550M or Seasonic FOCUS Plus 550 Gold (depending on budget).

 

You mentioned you have HWMonitor installed. Can you post a small screenshot, similar to the one I'll show below? I'm interested in seeing your the CPU VDD line after running Aida64 for at least 15 to 30 minutes stressing CPU/FPU/Cache. No need to crash it, just interested in the cpu core voltage during load.

 

Leave the stress test running while you save that screenshot. While it's running, check GPU adding it to the test and watch the CPU VDD line to see if the 'Value' column drops significantly. I assume this will cause Aida64 to fail, but I'd like to know what happens to the cpu core voltage when the GPU is added to the test. If the voltage drops for any reason after adding the GPU, then the PSU has to be faulty, or something is seriously screwed with the board and it's incapable of handling that much current. This is just my opinion, however. I'm no expert.

 

It does appear is though your Aida64 can monitor CPU VDD, so perhaps you can unselect all of those other sensors, only showing the one.

[Entermaker]

I'm in college, but as soon as I got home I will do the test.

For now I have these graphs, that I took yesterday.

I was running Aida64 then I was playing a bit during the monitoring.

[johndms]

7 minutes ago, Entermaker said:

I'm in college, but as soon as I got home I will do the test.

For now I have these graphs, that I took yesterday.

I was running Aida64 then I was playing a bit during the monitoring.

Well, CPU VDD is the reading for CPU Core Voltage that I pay attention to. It seems to be most accurate. From that first graph, it appears your core voltage bottoms out when playing games. I assume that's when the GPU starts drawing power and the CPU doesn't have enough voltage to continue. Umm. I wonder if it's something with the Windows power saver options in whatever power plan you're using. When you get home from college, try choosing the High Performance Power Plan and see if it still happens.

 

It's as if when you start gaming, your CPU goes into power save mode.

[Entermaker]

I am using High Performance Power Plan all the time, so its not the problem.

[johndms]

Just now, Entermaker said:

I am using High Performance Power Plan all the time, so its not the problem.

Still brainstorming ideas. As a temporary test, go into your bios and disable Cool'n'Quiet, C6 State, and Global C-State Control (remember what they were so you can revert afterwards). That should disable any power saving options, causing the cpu core voltage to run at whatever you've manually entered. You should be able to leave the CPU frequency at stock while manually entering 1.35v as CPU Core Voltage. If not, just enter 3600 for the frequency. Nice and low so it can't cause problems.

 

What I'm looking for here is for CPU VDD to stay at, or very near 1.35v under load. But it shouldn't be going to 0.6v.

 

I've personally run with all those particular bios options both enabled and disabled, overclocked and not overclocked. Right now, at my current overclock, I have only CnQ disabled, with the others set to Auto. Hopefully my ideas don't sound too crazy.

[Entermaker]

Alright, I will do that.

[Entermaker]

I installed Windows 10 and for the first time it ran perfectly. But it started to produce somewhat similar problems.

It starts fine and it doesn't seem to fail during Aida64 stability test.

However gaming is another story. Rise of the Tomb Raider sometimes fails to start, crashes to desktop or freezes after a few minutes randomly.

I don't think it is a problem with the game, because my 4 year old laptop can run it for even 8 hours without a problem.

 

Anyway, I stressed the CPU/FPU/Cache for half an hour and then I added the GPU to the test.

I didn't see any problem.

 

I set the VCORE to fix 1.35V and it was stable during stress test, but the game still crashed.

 

Once I monitored when a game froze the entire pc. It seems like that the GPU used dynamic memory goes up and then the crash happens.

I ran MemtestCL to test the GPU memory and it passed everything.

 

Any ideas?

 

 

 

 

 

 

 

 

 

[johndms]

47 minutes ago, Entermaker said:

Any ideas?

First off, this is the only screenshot that make sense. All that other stuff is useless to me. I have no idea what I'm looking at.

 

If you set 1.35 in the bios and the vcore (CPU VDD) is dropping to 1.044 DURING a game or stress test of cpu+gpu, then something is seriously screwed up. Especially if it carries over between two different operating systems and fresh installs. Either take it to a repair shop asking them to verify your PSU is faulty by testing your system with one of their own, or just buy a new PSU. 550w or higher should be plenty.