[SOLVED] Possible PSU issue - SOLVED

[dominic2005]

To save others, in the unlikely event of experiencing a similar issue, the cause of this was the Corsair GPU Riser Cable (Gen 3.0). The PCI-e slot on the board was Gen 5.0 and the GPU itself was Gen 4.0, so the board, detecting a Gen 4.0 card, was trying to run the GPU at Gen 4.0 through what I'd technically downgraded to Gen 3.0 without realising, and it just caused lockups.​

​

Removing the cable solved the problem.​

​

Good evening,

I've recently upgraded my PC FROM:

Asus ROG Maximus IX Formula
Intel Core i7-7700K
Corsair Vengeance LED 32GB (2x16GB) DDR4 3000MHz Dual Channel Kit
Asus GTX3080 12GB
Corsair iCue RGB Hub + 6 RGB Fans
Corsair AIO H100I Elite
2 x Samsung M2 SSD
INWIN 900w PSU - Close to 6 years old

No issues for almost 6 years

TO:

Asus ROG Strix Z790-F Gaming W
Intel Core i9-13900KF
Corsair Vengeance RGB EXPO 64G (2x32GB) DDR5 5200MHz Dual Channel Kit
Asus GTX3080 12GB
Corsair iCue RGB Hub + 6 RGB Fans
Corsair AIO H100I Elite
2 x Samsung M2 SSD
INWIN 900w PSU - Close to 6 years old

I am experiencing random complete freezes within Windows 11. Sometimes it happens right after logon, sometimes ten or fifteen minutes after boot, sometimes during installing a driver/etc, and sometimes never.

I've not fully tested the RAM yet, but I have done several tests and there doesn't seem to be an issue there. I've stress tested the CPU whilst monitoring temps, never breached 97c nor did it lock up or freeze. (I am going to leave the RAM testing overnight with memtest86 to be completely sure).

Could this be a sign that I am exceeding the power the PSU can provide? Various calculators suggest my system requires 800w, but they don't take into account the iCue stuff. Given the age of the PSU as well?

Thanks in advance,


Dom

 

[Darkbreeze]

What is the ACTUAL model of your InWin power supply? Because InWin has several 900w models and they are entirely different in terms of quality, performance and reliability. Some of their power supplies are very decent, and a lot of them are pure junk. Being 6 years old it could be any number of issues with the PSU and none of them are actually related to it being "only" 900w, if in fact the unit can still or ever could, sustain that.

But, what IS likely is the fact that you are running AMD memory in your Intel machine. EXPO memory kits are 100% INTENDED for use with AMD systems, not Intel. My guess is that there is a problem with one of the timings or a combination of them that your Intel memory controller does not like, so even if it passes Memtest86 that doesn't mean that much. I've seen PLENTY of memory that could pass Memtest and yet had problems, that were resolved by replacing the kit with a different or more suitable one.

Kits primarily intended for use with Intel platforms tend to have issues when used with AMD or Ryzen platforms, and visa versa so first thing you need to do is get a kit that is not intended for an AMD system. But the PSU could absolutely be playing a role here as well although it's a lot less likely, but at six years old, PSU could be a contributor. I say less likely because nothing substantial has changed aside from the CPU power requirements and you had no problems before that you mentioned, so less likely, but still worth some stink eye.

 

[Bob.B]

Good evening,

I've recently upgraded my PC FROM:

Asus ROG Maximus IX Formula
Intel Core i7-7700K
Corsair Vengeance LED 32GB (2x16GB) DDR4 3000MHz Dual Channel Kit
Asus GTX3080 12GB
Corsair iCue RGB Hub + 6 RGB Fans
Corsair AIO H100I Elite
2 x Samsung M2 SSD
INWIN 900w PSU - Close to 6 years old

No issues for almost 6 years

TO:

Asus ROG Strix Z790-F Gaming W
Intel Core i9-13900KF
Corsair Vengeance RGB EXPO 64G (2x32GB) DDR5 5200MHz Dual Channel Kit
Asus GTX3080 12GB
Corsair iCue RGB Hub + 6 RGB Fans
Corsair AIO H100I Elite
2 x Samsung M2 SSD
INWIN 900w PSU - Close to 6 years old

I am experiencing random complete freezes within Windows 11. Sometimes it happens right after logon, sometimes ten or fifteen minutes after boot, sometimes during installing a driver/etc, and sometimes never.

I've not fully tested the RAM yet, but I have done several tests and there doesn't seem to be an issue there. I've stress tested the CPU whilst monitoring temps, never breached 97c nor did it lock up or freeze. (I am going to leave the RAM testing overnight with memtest86 to be completely sure).

Could this be a sign that I am exceeding the power the PSU can provide? Various calculators suggest my system requires 800w, but they don't take into account the iCue stuff. Given the age of the PSU as well?

Thanks in advance,


Dom

Just a test.....look through the bios and see if you can disable turbo.

 

[Darkbreeze]

Just a test.....look through the bios and see if you can disable turbo.

Ok, so, humor me. I had posted something else but then decided, no, let's don't take that tact. What are you trying to get to with this? I'm just wondering because I'm not too sure what turbo being enabled could have to do with this specific type of problem since thermal compliance does not seem to be an issue?

 

[dominic2005]

What is the ACTUAL model of your InWin power supply? Because InWin has several 900w models and they are entirely different in terms of quality, performance and reliability. Some of their power supplies are very decent, and a lot of them are pure junk. Being 6 years old it could be any number of issues with the PSU and none of them are actually related to it being "only" 900w, if in fact the unit can still or ever could, sustain that.

But, what IS likely is the fact that you are running AMD memory in your Intel machine. EXPO memory kits are 100% INTENDED for use with AMD systems, not Intel. My guess is that there is a problem with one of the timings or a combination of them that your Intel memory controller does not like, so even if it passes Memtest86 that doesn't mean that much. I've seen PLENTY of memory that could pass Memtest and yet had problems, that were resolved by replacing the kit with a different or more suitable one.

Kits primarily intended for use with Intel platforms tend to have issues when used with AMD or Ryzen platforms, and visa versa so first thing you need to do is get a kit that is not intended for an AMD system. But the PSU could absolutely be playing a role here as well although it's a lot less likely, but at six years old, PSU could be a contributor. I say less likely because nothing substantial has changed aside from the CPU power requirements and you had no problems before that you mentioned, so less likely, but still worth some stink eye.

Good morning,

Thanks for that very detailed reply. I honestly hadn't realised that memory could be tailored to a specific CPU/Chipset type.

It would possibly explain some other oddities I experienced, for instance, at the default settings the BIOS would say the memory is 4000MHz and others 4800MHz. For a time I did run with just one of the modules and didn't experience any problems for quite a while, and then suddenly out of the blue the system did lock up.

The PSU is In-Win C900 Classic Series 900W 80 Plus Platinum Modular Power Supply, and has been in use since 3rd March 2017. For a lot of that time my PC was never shut down as I used it overnight for other things as well, so in all that time I'd probably say it's been running 75% to 85% of that time, so it's had extensive use.

What would be the symptoms of the memory issue? The only issue I experience is a complete freeze of the OS. I am not seeing any corrupted files or damaged installs etc, just a random and complete freeze. And now the RESET button won't always reset the PC, that is a new development since I made this post.

Edit:

I've also noticed if I disable the DOCP profile in the BIOS, the problem is much worse/frequent. If I enable it, then it's not so much of a problem and back to being pretty random, except if I try running 3D mark on my GPU, it always locks up. (I know 100% it isn't the GPU as it's been tested on a different board/CPU/Memory and doesn't cause problems).

I've ordered a RAM kit that is listed as being compatible with the board/CPU, so will see if that solves it. In the meantime, I've checked the TDP of my current GPU and its over 350w, I have an older GPU that is 180w, IF there is a supply issue swapping those over would in theory solve it? And if not, then its 100% the RAM ?

 

[Bob.B]

Ok, so, humor me. I had posted something else but then decided, no, let's don't take that tact. What are you trying to get to with this? I'm just wondering because I'm not too sure what turbo being enabled could have to do with this specific type of problem since thermal compliance does not seem to be an issue?

It's just a test the ops call if they want to try it.
Never say never.

 

[dominic2005]

I think I've narrowed this down to the PSU. I installed my old 1080 back in, which had half the TDP of the 3080, and I had no problems.

Running a bench on the GPU which was 100% certain to trigger a lock up no longer did. I am 100% certain the issue ISN'T the GPU as over the weekend I put my previous board/CPU/Memory in and as it has been for years, all worked perfectly.

That leaves the PSU.

 

[dominic2005]

It's just a test the ops call if they want to try it.
Never say never.

This wasn't me replying to you.

I did consider your suggestion but in the face of other evidence didn't feel it would help, but thanks for your input.

 

[Bob.B]

This wasn't me replying to you.

I did consider your suggestion but in the face of other evidence didn't feel it would help, but thanks for your input.

Turbo brings on a big bump in power draw which may have been just enough to push things over the edge.

 

[dominic2005]

Turbo brings on a big bump in power draw which may have been just enough to push things over the edge.

Yes of course I understand that, I just don't see what would be achieved now?

I've fairly conclusively narrowed my issue down to the PSU, all disabling Turbo would do, if anything, would allow the system to run stable albeit at significantly reduced performance.

Replacement PSU, and RAM will be here tomorrow (just in case I am wrong), so I should know more by then.

Again, I appreciate all input.

 

[Bob.B]

Yes of course I understand that, I just don't see what would be achieved now?

I've fairly conclusively narrowed my issue down to the PSU, all disabling Turbo would do, if anything, would allow the system to run stable albeit at significantly reduced performance.

Replacement PSU, and RAM will be here tomorrow (just in case I am wrong), so I should know more by then.

Again, I appreciate all input.

Now?.....nothing.
At the time I suggested it it was just a test to see if it made diff.

 

[Darkbreeze]

So, yeah, a new PSU is definitely looking like it is in your future, especially since you've ordered one. What, EXACTLY, did you order?

Same for the memory which definitely needs to be replaced because I guarantee using an EXPO kit on an Intel platform WILL cause some kind of problem either regularly or intermittently/periodically, whether minor or major is variable. What did you order on that as well as far as the model of the kit?

 

[dominic2005]

So, yeah, a new PSU is definitely looking like it is in your future, especially since you've ordered one. What, EXACTLY, did you order?

Same for the memory which definitely needs to be replaced because I guarantee using an EXPO kit on an Intel platform WILL cause some kind of problem either regularly or intermittently/periodically, whether minor or major is variable. What did you order on that as well as far as the model of the kit?

Re the RAM, just a temporary replacement to get the machine usable until I can invest in better.

TeamGroup Vulcan 16GB (2X8GB) DDR5 PC5-41600C40 5200MHz Dual Channel Kit

Re the PSU:

Corsair HX1500i 1500W Modular 80+ Platinum PSU

 

[dominic2005]

Sooooo, turns out it was none of the above.

It was the riser cable I used to vertically mount my GPU. For some reason, the combination of that board/GPU/riser just didn't want to play, even though it would work fine with an older 1080, and always worked fine with my older board and the same 3080.

I can't for the life of me imagine why that might cause a problem, but apparently it does.

Edit: Gen 3.0 riser cable on a gen 4.0 GPU on a gen 5.0 board side socket is just too much?

 

[Darkbreeze]

Maybe, it's possible you might need to have a Gen 4 riser cable, but I'm doubtful. All PCIe 3.0 through 5.0 devices, slots and hardware are supposed to be backwards and forwards compatible. I guess the only way to know for sure is to try it as I've not heard any such thing at any time in the past or ever for that matter. But, the majority of systems aren't using riser cables either.

 

[dominic2005]

Maybe, it's possible you might need to have a Gen 4 riser cable, but I'm doubtful. All PCIe 3.0 through 5.0 devices, slots and hardware are supposed to be backwards and forwards compatible. I guess the only way to know for sure is to try it as I've not heard any such thing at any time in the past or ever for that matter. But, the majority of systems aren't using riser cables either.

As with you, it's not something I've ever heard of, but then again, it's not something I've gone looking for to hear of either. All I know is removing the riser solved the issue. I am 99% certain the card itself isn't the issue as it would work on my previous board with the riser fine. Strangely enough though, on the new board the riser didn't cause an issue when paired with an older 1080, that only leaves the combination of the 3080/riser/new board being the issue.

Obviously the only component removable from that is the riser.

I will keep a close eye on the RAM as well though, although the board does seem to recognise it is EXPO as the bios loads a DOCP profile when it's fitted, and when I tried the replacement RAM it loaded an XMP profile. And the only information I can find regarding compatibility from Asus and Corsair is when running on an Intel platform it may not give the absolute max performance it would on an AMD platform, which does seem to mirror the fact that the BIOS recognises it as 5200, as does Windows, yet CPU-z says it's 4800.

Either way, it's an upgrade over my previous, so as long as it continues to run stable, I can live with it. I will eventually replace it, once I've decided if my goal of 128GB is worth the tradeoff of the board/CPU only able to run that amount at substantially slower speeds. I'm leaning towards it not being worth it, but that's another discussion.