Hi so when starting my PC up I have a solid red light at the GPU. And then when in the desktop making mouse movements and opening programs leads to random GPU spikes as well as audio stutters and mouse stutters. Whenever loading GPU intensive tasks such as games the PC will crash and restart and via latencyMon I am also getting very high latency.  I have tried a full PC reset, uninstalling drivers with DDU, BIOS update...
 

There is no way for the GPU to detect that the power supply is inadequate at startup, there is no communication between them. The way they monitor this is to test that both cables are plugged into the sockets provided when we used to have 2 sockets on GPU instead of the new connector. This could indicate a cable has not seated correctly or shaken out of contact during flight. You may not believe it but I have seen first hand parts unplug after a long flight.

However at this point I am guessing you have tried to reseat the cables for GPU. One thing that could be happening in a power supply is that 2 sockets are bonded together and then another 2 sockets are bonded together on a separate circuit, its far fetched but you could try moving the one of the gpu cables across to a spare gpu socket.

Hey, I have tried that and still getting the red light. The red light does go off when powering the system on though so I am not sure the light just indicates it is getting power? Should I try and use something like OCCT and HWMonitor to check the power delivery?

Also looks like none of my DisplayPorts and HDMI ports are working, the PC will now also just shut off after a minute or two 
Also forgot to mention that earlier when I used the Memory Diagnostic Tool it did detect memory issues as well

Edited February 16, 20251 yr by Strider20A

I plugged the HDMI into the motherboard and I am now getting and image, and the lag/stuttering is completely gone compared to when plugged into the GPU

 

Based off these clear that GPU not getting enough power, saw around max 110w consumption but typically fluctuates between 60 - 90w with spikes in utilisation ranging between 17 -100%. 

The thing that can catch us out is the riser cable. I have had this issue with the BBC running a 4060, At first everything was ok and then it started to glitch out. Main issue with riser cables is that they allow interference on high speed data transfer witch then causes crashes and nasty lag, The way we validated this was to put the GPU in a separate computer, everything worked, then added the riser card and boom.

The Riser card should be 4.0 standard, I have checked Azzar site but it doesn't spec either. I have had PC's run on 3.0 standard before but it may have been under favorable conditions. I can't really see the large screen diagnostics you posted.

Would the riser cable lead to the lower power consumption for the GPU though? These should be easier to see, this is shortly after booting the system.

Based off of these is there any indication that maybe the PSU got damaged somehow and now isn't delivering the power it should? Or is it more likely a GPU failure?

 

But I am guessing that maybe there is something wrong with the riser cable since under load the bus interface is x16 @ 2.0 instead of 4.0? 

Hey so I went to Bios and changed the Pcie slot speed to gen 3 from auto and this seems to have fixed the issue! I am guessing the riser cable is a gen 3 then, and if so would it be worthwhile to go to a gen 4 or not really? Either way thanks for the assistance, I really appreciate it! 🙂

 

No worries, Do you want to run a gpu benchmark, I can authenticate my steam account if you send me the QR code. we just need to coordinate at time. In most cases pci3 is sufficient. this one being TI overclocked could be a bottleneck. Im in sydney time and 8am till 12pm.

Edited February 18, 20251 yr by 011010010110

Currently don't have it on hand, but don't want to bother you further so it is fine thanks. Will let you know if I need any further assistance though 🙂

Hey so after changing the Pcie link slot speed to gen 3 in the bios I was able to remove the stutters but then I started getting random BSOD, so I decided to test the RAM via memtest86 and found that one of the RAM sticks were faulty. Does the RAM still have its warranty and would I be able to replace it here or would I need to send it back to you guys? Thanks!

I would swap the ram sticks arround, make sure the bios has re-profiled them and run the test. this will do 2 timgs, ID if the Ram stick is actually faulty since now the opposite stick should fail and secondly you are re-inserting and profiling the ram. Let me know the outcome.

Hi so have done that. I did the test running both, I then swapped them around and was still getting errors, I then tried doing one stick at a time and one was able to finish 4 passes without error but the moment I swapped the other one in (the faulty one), I started getting errors immediately. I also tried using different dimm slots at a time and was still only getting errors on the one RAM stick. Currently only running one stick (the working one) and haven't had a single crash yet

Edited February 27, 20251 yr by Strider20A

Turns out the RAM stick was faulty, resolved with a new RAM stick ordered!