I have NO IDEA what the issue is!

[Morrob95]

Guys please help! I've been having issues for months and months with random crashes. Sometimes my PC doesnt boot at all. Sometimes it says code 00, sometimes it says hdd error. Sometimes it boots up just fine and runs for hours. I can run an aida64 stress on gpu, cache, mem, fpu and cpu for hours with no issues. Although, I notice my voltages seem off in aida 64 with my 3.3 reporting 3.05 and my 5v reporting down to 4.6. Not sure how accurate that is. When I say crash, I mean the computer literally just powers down and then restarts. I've built countless computers and never had an issue like this so I'm genuinely lost.

 

My temps under full load 

60-65C CPU

70ish GPU and up to 90c memory (while mining only, not gaming. Liquid metal and pads have been done, now it doesnt throttle).

Ram used to get in the 65c range so I added a corsair ram cooler and now it never goes above 40c.

 

I've just ran memcheck86 and it passed with no errors. My event logs just show that the PC crashed unexpectedly but doesnt say specifically what caused it.  I thought that my boot ssd was failing (970 pro) so I went and replaced with a 2tb 860 qvo and I'm having the same problem.

 

My specs are as follows

Motherboard is ASUS Formula VIII (on latest bios)

Ram is 64 Gigs trident Z neo 3600

CPU is 3950X

GPU is Founders 3080 (Notable that I had the same issues when I had a 3090 so I dont think its the gpu)

PSU is an EVGA 850 GQ

360 enermax AIO

lian li case

 

I've also performed a fresh install of both windows 10 and now a fresh install of windows 11 to no avail, still having the same issue. I can be doing absolutely nothing other than loading a page on chrome and it'll crash or I could be playing EFT it'll crash. Notable: I can mine for days and days with absolutely no issues.

 

 

Please help, I'm at my wits end here. Anything would be appreciated.

 

Also, I'm not overclocked when these issues occur. Or I can be overclocked and they'll occur. It doesnt seem to matter. 

 

Edit: I should also mention I went to bestbuy and swapped my psu for the same one and had the same issue so I returned it, so I think we can mark off the psu.

 

[Jurrunio]

13 minutes ago, Morrob95 said:

PSU is an EVGA 800

800w? Which model?

[Morrob95]

Just now, Jurrunio said:

800w? Which model?

Sorry, 850. its the 850 GQ

[Indian pc builder]

u might want to try a 1000 watt, not evga psu, u might not have enough power

[Indian pc builder]

other thing to check would be reseating the ram, trying new ram, or maybe a new motherboard, there might be an issue with the mobo. also check event viewer for errors, if there are power errors or something, try changing cpu and gpu power settings

[Jurrunio]

44 minutes ago, Morrob95 said:

Notable: I can mine for days and days with absolutely no issues.

So it's fine if you keep the workload steady, otherwise it throws a tantrum. I suppose the CPU's not stable enough at max clocks the turbo boost algorithm pulls it to during sudden workloads (say, moving the cursor around).

 

If this is true, you have a problem because Ryzen CPUs hit max safe voltages periodically at stock already so you can't just raise it and keep everything constant. You'll have to drop the boost clocks and I'm not sure if such an option exist.

 

The way I know is to get rid of the turbo boost system altogether and manually set a core clock multiplier. It does not drastically increase power draw because Ryzen reduces idle power by not powering unused parts of the CPU instead of only low clocks and voltages like on Intel, but it will still go up somewhat.

 

Alternatively is make the motherboard work harder, you can turn up loadline calibration (to the middle level) and VRM switching frequency to improve voltage regulation and hopefully can stabilize the CPU. You can prevent the average voltage from going too high by applying negative voltage offset to the CPU core voltage, but you cannot stop the power loss from the VRM when you raise the switching frequency from going up which means it will run hotter. Not that your board can't handle that, it's overkill enough, but this also doesn't help much from what I know about high end boards.

[Morrob95]

1 hour ago, Jurrunio said:

So it's fine if you keep the workload steady, otherwise it throws a tantrum. I suppose the CPU's not stable enough at max clocks the turbo boost algorithm pulls it to during sudden workloads (say, moving the cursor around).

 

If this is true, you have a problem because Ryzen CPUs hit max safe voltages periodically at stock already so you can't just raise it and keep everything constant. You'll have to drop the boost clocks and I'm not sure if such an option exist.

 

The way I know is to get rid of the turbo boost system altogether and manually set a core clock multiplier. It does not drastically increase power draw because Ryzen reduces idle power by not powering unused parts of the CPU instead of only low clocks and voltages like on Intel, but it will still go up somewhat.

 

Alternatively is make the motherboard work harder, you can turn up loadline calibration (to the middle level) and VRM switching frequency to improve voltage regulation and hopefully can stabilize the CPU. You can prevent the average voltage from going too high by applying negative voltage offset to the CPU core voltage, but you cannot stop the power loss from the VRM when you raise the switching frequency from going up which means it will run hotter. Not that your board can't handle that, it's overkill enough, but this also doesn't help much from what I know about high end boards.

That would be really odd though wouldn't it? Would it be the MOBO at that point or the cpu? Not gonna lie, this formula viii had to be replaced under warranty. Which imo is a bit ridiculous for a 600$ board. One thing I have noticed is in bios, I have one voltage showing in the red for some reason, however, it doesnt list what it is... I'll try to get a picture uploaded right now. Might be poor quality because I'll have to snap the picture on my phone and then upload it. I do mine with the 3950x as well. Could it be silicone degradation at this point? I dont think it would because I've never seen it before, only heard about it. Not to say it doesnt happen, but after only a year? My temps are really well controlled. Plus I've had these issues since putting the build together. I'm really leaning toward the mobo but I don't really know how to tell for sure. I guess its just speculation.

 

Edit: So Just turned off and tried to boot again and the first time I tried to boot I was greeted with code F0 "detect HDD". Turned it off, and turned it on again and was able to enter bios and then saved and exit and the computer boot up and I'm typing on it now. Also now, the voltage is no longer in the red, however, it was the 3.3v rail. Before it was showing 2.9, how its showing 3.2. Maybe just the eco mode of the psu, I dont know. I've swapped psu though. here are some pictures. Sorry for the terrible quality, I'm using my blackberry to take the pictures. They are in the spoiler since they are so large.

 

Also, yes I know the gpu isnt in the top slot. That slot is also x16. I put it there to add the ram cooler the other day because it interfered in the top slot. Hasnt effected performance, at least not enough for me to notice. This all happened in the top slot as well. 

 

 

 



 

 

[Jurrunio]

1 hour ago, Morrob95 said:

Would it be the MOBO at that point or the cpu?

Probably the motheboard, it's reading all voltages I can see (including memory, should be 1.35V) low.

 

1 hour ago, Morrob95 said:

Could it be silicone degradation at this point? I dont think it would because I've never seen it before, only heard about it. Not to say it doesnt happen, but after only a year?

AMD is very aggressive here so I won't be surprised if that is the case. Doubt, but not impossible.

 

1 hour ago, Morrob95 said:

That slot is also x16

It's x8 by connection, you're only getting half the speed here

 

1 hour ago, Morrob95 said:

Which imo is a bit ridiculous for a 600$ board.

If it weren't for the fancy and totally unnecessary VRM waterblock and measly 5G LAN (compared to 2.5G), the Hero would have saved you at least $200. Not that it's still anywhere cheap

[Frizz]

Going to second here that it is likely the MB, since you swapped out the PSU and have the same issues. 
Voltage sensors are on the MB and since multiple sensors you mentioned are reading low, I have some bad news.

You MIGHT be able to up certain voltages to kinda workaround it, but if it is shutting off under normal loads, then it isn't holding the right voltages is very likely the cause and needs replaced.

The HDD not detected is likely that port/ports not getting enough juice, this all points back to MB. Can you RMA it?

[Morrob95]

20 hours ago, Jurrunio said:

Probably the motheboard, it's reading all voltages I can see (including memory, should be 1.35V) low.

 

AMD is very aggressive here so I won't be surprised if that is the case. Doubt, but not impossible.

 

It's x8 by connection, you're only getting half the speed here

 

If it weren't for the fancy and totally unnecessary VRM waterblock and measly 5G LAN (compared to 2.5G), the Hero would have saved you at least $200. Not that it's still anywhere cheap

 

14 hours ago, Frizz said:

Going to second here that it is likely the MB, since you swapped out the PSU and have the same issues. 
Voltage sensors are on the MB and since multiple sensors you mentioned are reading low, I have some bad news.

You MIGHT be able to up certain voltages to kinda workaround it, but if it is shutting off under normal loads, then it isn't holding the right voltages is very likely the cause and needs replaced.

The HDD not detected is likely that port/ports not getting enough juice, this all points back to MB. Can you RMA it?

Damnit boys! I think I'm outside of the RMA window. How ridiculous that I need two rmas in two years with a 600$ board. Frankly I always have opted for the cheapest board with the correct socket and I've never had one fail before. I'm talking the gigabyte and asrock specials! This time since I was going with a 3950 I figured I'd go crazy. This board is a gimmick for the price. At least for the enthusiast. It doesnt seem to overclock any better since amd already has that in the bag. 

 

One more question for ya, any chance it would be the cable extensions causing issues? Also, arent rails rated for like + - 5% deviation?

 

Also, I dont need the 5g lan, my internet is only 30mb down because I live in the boonies and have to use a T-Mobile home internet hotspot LOL. 

[Jurrunio]

52 minutes ago, Morrob95 said:

One more question for ya, any chance it would be the cable extensions causing issues? Also, arent rails rated for like + - 5% deviation?

Cable extensions? Depends on whether they are well made and meet the correct specs

 

12V has 10% deviation, the rest is 5% from what I remember.

[Morrob95]

17 hours ago, Jurrunio said:

Cable extensions? Depends on whether they are well made and meet the correct specs

 

12V has 10% deviation, the rest is 5% from what I remember.

Update: Did bios flash back (updated to the newest bios, apparently they had a new one just a couple weeks ago that I hadnt seen) and undid the cable extensions and the voltage increased on the rails. Enabled dohc after I took the below picture and the voltage is readying a little under 1.35. Something like 1.33. Is that okay? Also reset bios to standard settings again just to make sure everything was back to factory. 

 

Even with the voltages seemingly okay now, I still crashed. Tried to boot and got an error code saying "detect HDD". Had to hard reset (hold power button) multiple times and then it started right back up like nothing had happened. Event log just shows that windows shut down unexpectedly still. Would my best bet still be top replace the mobo? What would you do in this situation? I don't have a ton of money to toss into it right now so if I can narrow it down to the problem and only replace that part, that would be best.

 

See pic below for voltages (at stock ddr4 speed without xmp/dohc enabled)

 

 

 

 

[Jurrunio]

2 minutes ago, Morrob95 said:

Update: Did bios flash back (updated to the newest bios, apparently they had a new one just a couple weeks ago that I hadnt seen) and undid the cable extensions and the voltage increased on the rails. Enabled dohc after I took the below picture and the voltage is readying a little under 1.35. Something like 1.33. Is that okay? 

 

Even with the voltages seemingly okay now, I still crashed. Tried to boot and got an error code saying "detect HDD". Had to hard reset (hold power button) multiple times and then it started right back up like nothing had happened. Event log just shows that windows shut down unexpectedly still. Would my best bet still be top replace the mobo? What would you do in this situation? I don't have a ton of money to toss into it right now so if I can narrow it down to the problem and only replace that part, that would be best.

 

You can try look for loose connectors and disk health, crystaldiskinfo for example and do the latter.