Have I killed my NVMe?

[AliT]

Unfortunate :\
You shouldn't spend money until you've narrowed down the cause of the crashes. Eliminating the hard drive as a suspect was the goal but I guess you can't do that just yet.



There are likely different revisions of the SF600 - it's been around on the market for a long time so there's no guarantee that the SF600 you saw that worked will be the same as the one you own. PSUs also behave very differently when on 115V or 230V and if it's a multi-rail design, which cables and how many different cables you use can also have an effect.

When your GPU is idle or under low load, that's when it crashes, and the GPU is by far the most power-hungry part of your PC. You should install Furmark and see if that causes an instant crash. On the assumption it will, you can see if it's too much GPU power draw causing the problem by installing MSI afterburner and reducing your power limit as low as the slider will go. If that helps, then you know you have a power draw issue and the next step is to try and isolate where the problem is, as your 3070 is drawing power from both the PSU directly, and also the motherboard slot.

Hi mate,
I have just downloaded Furmark and I am going to take the steps you have listed, could you let me know if these settings are okay for the GPU stress test or if I should change anything?

thank you.

Unfortunate :\
You shouldn't spend money until you've narrowed down the cause of the crashes. Eliminating the hard drive as a suspect was the goal but I guess you can't do that just yet.



There are likely different revisions of the SF600 - it's been around on the market for a long time so there's no guarantee that the SF600 you saw that worked will be the same as the one you own. PSUs also behave very differently when on 115V or 230V and if it's a multi-rail design, which cables and how many different cables you use can also have an effect.

When your GPU is idle or under low load, that's when it crashes, and the GPU is by far the most power-hungry part of your PC. You should install Furmark and see if that causes an instant crash. On the assumption it will, you can see if it's too much GPU power draw causing the problem by installing MSI afterburner and reducing your power limit as low as the slider will go. If that helps, then you know you have a power draw issue and the next step is to try and isolate where the problem is, as your 3070 is drawing power from both the PSU directly, and also the motherboard slot.

And how long should I run it for if it doesn’t crash?

Unfortunate :\
You shouldn't spend money until you've narrowed down the cause of the crashes. Eliminating the hard drive as a suspect was the goal but I guess you can't do that just yet.



There are likely different revisions of the SF600 - it's been around on the market for a long time so there's no guarantee that the SF600 you saw that worked will be the same as the one you own. PSUs also behave very differently when on 115V or 230V and if it's a multi-rail design, which cables and how many different cables you use can also have an effect.

When your GPU is idle or under low load, that's when it crashes, and the GPU is by far the most power-hungry part of your PC. You should install Furmark and see if that causes an instant crash. On the assumption it will, you can see if it's too much GPU power draw causing the problem by installing MSI afterburner and reducing your power limit as low as the slider will go. If that helps, then you know you have a power draw issue and the next step is to try and isolate where the problem is, as your 3070 is drawing power from both the PSU directly, and also the motherboard slot.

Currently been running the Furmark test with the settings I showed on the screenshot and it’s been 18 minutes and no crash or signs of crash whatsoever. My GPU min temp says 45C and max temp is 76C.

Unfortunate :\
You shouldn't spend money until you've narrowed down the cause of the crashes. Eliminating the hard drive as a suspect was the goal but I guess you can't do that just yet.



There are likely different revisions of the SF600 - it's been around on the market for a long time so there's no guarantee that the SF600 you saw that worked will be the same as the one you own. PSUs also behave very differently when on 115V or 230V and if it's a multi-rail design, which cables and how many different cables you use can also have an effect.

When your GPU is idle or under low load, that's when it crashes, and the GPU is by far the most power-hungry part of your PC. You should install Furmark and see if that causes an instant crash. On the assumption it will, you can see if it's too much GPU power draw causing the problem by installing MSI afterburner and reducing your power limit as low as the slider will go. If that helps, then you know you have a power draw issue and the next step is to try and isolate where the problem is, as your 3070 is drawing power from both the PSU directly, and also the motherboard slot.

 

[Chrispy_]

So your PSU can handle the full GPU load. Now fire up the CPU burner with all cores at the same time. If it handles that, you can say with confidence that your SF600 PSU is good enough for your rig.

The next step will be to use Afterburner to reduce your GPU clocks. Early 3080 cards crashed to desktop and bluescreened because their firware boosted too aggressively. If you set a -300MHz offset on your 3070 to stop it boosting so aggressively and Fortnite still causes bluescreens, we know it's not your GPU boosting that's causing the issue, so you can put the core clock back to how it was. That, coupled with the Furmark stress test basically eliminates your GPU as the culprit.

As others have suggested, the next thing to do after that would be to run memtestx86 off a bootable USB stick and make sure your RAM isn't the problem. Since you can't rule out your NVMe drive as faulty, you have to instead verify all the other components are good, one by one, until the SSD is the only thing left that you haven't ruled out.

 

[AliT]

So your PSU can handle the full GPU load. Now fire up the CPU burner with all cores at the same time. If it handles that, you can say with confidence that your SF600 PSU is good enough for your rig.

The next step will be to use Afterburner to reduce your GPU clocks. Early 3080 cards crashed to desktop and bluescreened because their firware boosted too aggressively. If you set a -300MHz offset on your 3070 to stop it boosting so aggressively and Fortnite still causes bluescreens, we know it's not your GPU boosting that's causing the issue, so you can put the core clock back to how it was. That, coupled with the Furmark stress test basically eliminates your GPU as the culprit.

As others have suggested, the next thing to do after that would be to run memtestx86 off a bootable USB stick and make sure your RAM isn't the problem. Since you can't rule out your NVMe drive as faulty, you have to instead verify all the other components are good, one by one, until the SSD is the only thing left that you haven't ruled out.

Hi,
I have no words to explain how appreciate I am of your time and efforts. It’s really humbling.

As you said the next step is to stress test the CPU, my CPU is the Ryzen 9 3900x which programme do you recommend I use to stress test it?

 

[Chrispy_]

Hi,
I have no words to explain how appreciate I am of your time and efforts. It’s really humbling.

As you said the next step is to stress test the CPU, my CPU is the Ryzen 9 3900x which programme do you recommend I use to stress test it?

The CPU burner built into Furmark is fine. There are other tools like AIDA64 or Prime95 but you already have Furmark installed. To rule out the PSU you should definitely run a combined CPU burner and Furmark GPU stress test at the same time. If it's good once your GPU temps stabilise then it's good enough to call done.

I'm heading out in a bit but I'll check in tomorrow.

You need to rule out:

CPU - Stress test with Furmark's CPU burner, Prime95, or AIDA64
GPU - Stress test with Furmark or OCCT but those synthetic power-virus tests won't push GPU boost clocks so verify boost clock stability isn't the issue by running 3DMark or Superposition
Power supply - Stress test by running a combined GPU and CPU stress test simultaneously
RAM - Memtest x86 from bootable USB (you'll need to follow their instructions to make one). If you get errors, first try disabling XMP, running at the default 2133MHz and testing with Memtestx86 again.

 

[AliT]

The CPU burner built into Furmark is fine. There are other tools like AIDA64 or Prime95 but you already have Furmark installed. To rule out the PSU you should definitely run a combined CPU burner and Furmark GPU stress test at the same time. If it's good once your GPU temps stabilise then it's good enough to call done.

I'm heading out in a bit but I'll check in tomorrow.

You need to rule out:

CPU - Stress test with Furmark's CPU burner, Prime95, or AIDA64
GPU - Stress test with Furmark or OCCT but those synthetic power-virus tests won't push GPU boost clocks so verify boost clock stability isn't the issue by running 3DMark or Superposition
Power supply - Stress test by running a combined GPU and CPU stress test simultaneously
RAM - Memtest x86 from bootable USB (you'll need to follow their instructions to make one). If you get errors, first try disabling XMP, running at the default 2133MHz and testing with Memtestx86 again.

Hi,

Thank you, and hope you have a good rest of your day. And will catch up tomorrow!

I have uploaded below my results (Also displayed my CPU temps with Ryzen Master):

CPU - Stress test with Furmark's CPU burner

GPU - Stress test with Furmark (I RAN FOR A TOTAL OF 20 MINS)

Power supply - Stress test by running a combined GPU and CPU stress test simultaneously
(I RAN FOR A TOTAL OF 27 MINS)

I will be doing the memtest tomorrow and I will let you know the results! I have also provided images of my temps, please let me know if there are okay also!

 

[Chrispy_]

Looks solid so far. If the memory is stable then it probably is your SSD that needs replacing. It's just a shame you don't have any other drives you can swap into the machine to test with.

It wouldn't hurt to make sure you're on the latest motherboard BIOS and AMD chipset driver but I doubt older versions would result in bluescreens.

 

[AliT]

Looks solid so far. If the memory is stable then it probably is your SSD that needs replacing. It's just a shame you don't have any other drives you can swap into the machine to test with.

It wouldn't hurt to make sure you're on the latest motherboard BIOS and AMD chipset driver but I doubt older versions would result in bluescreens.

To be honest, I’ve been running Fortnite again and nothings happened, it’s almost like the blue screen disappeared, which is weird. I will run the memtest tomorrow and update results!

Looks solid so far. If the memory is stable then it probably is your SSD that needs replacing. It's just a shame you don't have any other drives you can swap into the machine to test with.

It wouldn't hurt to make sure you're on the latest motherboard BIOS and AMD chipset driver but I doubt older versions would result in bluescreens.

And yes, I have the latest AMD chipset drivers.