More 7900XTX oddities

[fizzy]

Hi,

I posted previously with issues with what I thought was my GFX card playing up (Saphire Nitro+ 7900XTX), but tested out fine in another pc and my main pc worked fine when using my previous card back in (Radeon VII). Ended up having to do a fresh install of Windows 11 too due to corruption possibly due to all the blind restarts etc.

Anyhow, thought it was all sorted - memtest86 showed no errors, 3dmark stress tests all ok etc.

It has been running fine for a few weeks. Away for a couple of days and on coming back when I switched pc on, fans ramped up on GFX card and pc wouldn't complete post. Looking at the on board bios diagnostics display it was continuously cycling through various codes but never failing on anything, just going through the same codes over and over. Tried a full power off cycle etc, no effect.

Took out the 7900XTX, tried booting - and board completed post but with a display error code as expected, and looked to load windows (plenty of disk activity). Once settled, quick press of power button to trigger windows shutdown and all shutdown as expected.

Put card back in, tried again.... and all ok? I'm baffled....

System :-

ASRock X570 Taichi, latest bios installed.
Sapphire Nitro+ 7900XTX
AMD 5950X
8 Pack Dark 8GB 3600mhz DDR4 x 4
Seasonic 1000W platinum PSU
No overclock in place.

Proving hard to pinout the exact issue with this one. Might have to be a case of waiting until something fails "properly" to find cause?

 

[champ222]

Any riser cables in use?

Either way, you could try and specify a PCIE version, in Bios, rather than Auto. try selection PCIE 3 and see if that helps. though if its an intermittent problem, its difficult to know if its fixed or not.

 

[Raelgun]

Hi,

I posted previously with issues with what I thought was my GFX card playing up (Saphire Nitro+ 7900XTX), but tested out fine in another pc and my main pc worked fine when using my previous card back in (Radeon VII). Ended up having to do a fresh install of Windows 11 too due to corruption possibly due to all the blind restarts etc.

Anyhow, thought it was all sorted - memtest86 showed no errors, 3dmark stress tests all ok etc.

It has been running fine for a few weeks. Away for a couple of days and on coming back when I switched pc on, fans ramped up on GFX card and pc wouldn't complete post. Looking at the on board bios diagnostics display it was continuously cycling through various codes but never failing on anything, just going through the same codes over and over. Tried a full power off cycle etc, no effect.

Took out the 7900XTX, tried booting - and board completed post but with a display error code as expected, and looked to load windows (plenty of disk activity). Once settled, quick press of power button to trigger windows shutdown and all shutdown as expected.

Put card back in, tried again.... and all ok? I'm baffled....

System :-

ASRock X570 Taichi, latest bios installed.
Sapphire Nitro+ 7900XTX
AMD 5950X
8 Pack Dark 8GB 3600mhz DDR4 x 4
Seasonic 1000W platinum PSU
No overclock in place.

Proving hard to pinout the exact issue with this one. Might have to be a case of waiting until something fails "properly" to find cause?

Can I ask any issues in games?

For almost a year I was having graphical glitches in certain games which I thought was a GPU issue but turns out a faulty CPU was the cause when I replaced it.

Didn't think the CPU would be the offender for something like this

 

[fizzy]

Any riser cables in use?

Either way, you could try and specify a PCIE version, in Bios, rather than Auto. try selection PCIE 3 and see if that helps. though if its an intermittent problem, its difficult to know if its fixed or not.

No, no riser cables - fitting direct to board. Being an intermittant issue I think this is unlikely, but might try forcing PCIE 3 to see if that makes any difference. However, being a PCIE4 chipset, and a PCIE4 card I would think if things are not working as intended this would indicate a fault with one or the other.

 

[champ222]

No, no riser cables - fitting direct to board. Being an intermittant issue I think this is unlikely, but might try forcing PCIE 3 to see if that makes any difference. However, being a PCIE4 chipset, and a PCIE4 card I would think if things are working as intended this would indicate a fault with one or the other.

Indeed, might be worth trying.

 

[fizzy]

Can I ask any issues in games?

For almost a year I was having graphical glitches in certain games which I thought was a GPU issue but turns out a faulty CPU was the cause when I replaced it.

Didn't think the CPU would be the offender for something like this

I have been playing Baldurs Gate 3 a lot and have been having crashes to desktop with that, especially into the later acts. However reading online this is a common issue so not an indicator in itself - and as stated various benchmark and stability tests seem to pass ok, hence adding to the confusion. Other games previously have been fine, including Cyberpunk etc.

CPU issue is a possibility - I was getting beep error codes on post sometimes before which I thought were GFX related but on reading up may actually be CPU. Things settled again for a while though before this issue, so all very frustrating trying to track down exact cause.

 

[JediFragger]

Didn't think the CPU would be the offender for something like this

Let me guess, AMD. I hear a lot of it tbh.

 

[Raelgun]

I have been playing Baldurs Gate 3 a lot and have been having crashes to desktop with that, especially into the later acts. However reading online this is a common issue so not an indicator in itself - and as stated various benchmark and stability tests seem to pass ok, hence adding to the confusion. Other games previously have been fine, including Cyberpunk etc.

CPU issue is a possibility - I was getting beep error codes on post sometimes before which I thought were GFX related but on reading up may actually be CPU. Things settled again for a while though before this issue, so all very frustrating trying to track down exact cause.

This was the thing, running stability tests never drew it out.

It was very specific games and a number of games just did not have issues at all, like they ran perfectly.

The only game that crashed to desktop was Starfield, but once I replaced the CPU it even fixed that.

Not saying its your exact issue but if you know of a space CPU, you can try that. ( the faulty CPU in question was a Ryzen 5900X )

- YouTube

Enjoy the videos and music that you love, upload original content and share it all with friends, family and the world on YouTube.

www.youtube.com

This is an example of what my CPU was doing ( you wouldnt think a CPU was doing this )

 

[Raelgun]

Let me guess, AMD. I hear a lot of it tbh.

Sadly yes, at first I thought it was the Asus ROG Mobo I had that was at fault ( however im not ruling out that the Asus mobo might of been the source of the problem and inflicted the issue but replaced that mobo anyways )

Next build will be intel but thats for the USB ports lol, AMD dropped the ball hard for not including USB4

 

[fizzy]

This was the thing, running stability tests never drew it out.

It was very specific games and a number of games just did not have issues at all, like they ran perfectly.

The only game that crashed to desktop was Starfield, but once I replaced the CPU it even fixed that.

Not saying its your exact issue but if you know of a space CPU, you can try that. ( the faulty CPU in question was a Ryzen 5900X )

- YouTube

Enjoy the videos and music that you love, upload original content and share it all with friends, family and the world on YouTube.

www.youtube.com

This is an example of what my CPU was doing ( you wouldnt think a CPU was doing this )

Interesting - not seeing anything as in your vid though.

If I had a spare AMD board and Chip I would be parts swapping in my testing. Unfortunately, I don't so it would require buying parts to test unless I can isolate the fault by some other means or it becomes a "proper" fault.

Only other system I have is an old intel system which is what I was testing the gfx card in before. It's also odd that when I was having my previous spate of issues, everything ran fine using my old Radeon VII - but then started working with the 7900XTX again. At the moment there's no definitive descernible cause, which is always a pain when trying to figure out pc problems.

 

[VonScar]

Interesting - not seeing anything as in your vid though.

If I had a spare AMD board and Chip I would be parts swapping in my testing. Unfortunately, I don't so it would require buying parts to test unless I can isolate the fault by some other means or it becomes a "proper" fault.

Only other system I have is an old intel system which is what I was testing the gfx card in before. It's also odd that when I was having my previous spate of issues, everything ran fine using my old Radeon VII - but then started working with the 7900XTX again. At the moment there's no definitive descernible cause, which is always a pain when trying to figure out pc problems.

How old is the power supply? Have you got a spare PSU to test PC?

Sometimes, random problems with errors, which cannot be tracked down may relate to the PSU not coping with transient spikes.

Also, I assume you are not daisy chaining the cables to it. I mean separate power cables from the PSU to GPU. Not one cable to both sockets on GPU.

Just a thought.

 

[fizzy]

How old is the power supply? Have you got a spare PSU to test PC?

Sometimes, random problems with errors, which cannot be tracked down may relate to the PSU not coping with transient spikes.

Also, I assume you are not daisy chaining the cables to it. I mean separate power cables from the PSU to GPU. Not one cable to both sockets on GPU.

Just a thought.

It's probably a few years old now - believe it had a 7 year warranty with it when I bought it and cost a fair bit, and is a good quality seasonic platinum unit. I do have a Corsair HX 700W, but that is at least 10 years old and only really used in emrgencies. At this rate if I'm looking to try replacing psu, try another cpu and motherboard... I'm heading towards getting a new pc! May well be an option if things start playing up again.

As for transient spikes - I would imagine that would be while running, not an issue with initial posting. It was something effecting the post process that cleared after I removed the gfx card then re-inserted later.

edit: And no I'm not daisy chaining power connections. Three seperate leads from psu, and it's a modular unit.

 

[GhostDog1981]

It's probably a few years old now - believe it had a 7 year warranty with it when I bought it and cost a fair bit, and is a good quality seasonic platinum unit. I do have a Corsair HX 700W, but that is at least 10 years old and only really used in emrgencies. At this rate if I'm looking to try replacing psu, try another cpu and motherboard... I'm heading towards getting a new pc! May well be an option if things start playing up again.

As for transient spikes - I would imagine that would be while running, not an issue with initial posting. It was something effecting the post process that cleared after I removed the gfx card then re-inserted later.

edit: And no I'm not daisy chaining power connections. Three seperate leads from psu, and it's a modular unit.

I tend to agree with @VonScar it’s worth trying another PSU as I had a similar problem with my RM850x and a 7900XT MBA which kept crashing on load or certain games. The 3060ti I had before never crashed. At first I thought it was AMD drivers (can’t really blame me with all the negativity in the social media forums) but someone suggested trying another PSU and it did the trick. No crashes and stable since April23.

 

[fizzy]

I tend to agree with @VonScar it’s worth trying another PSU as I had a similar problem with my RM850x and a 7900XT MBA which kept crashing on load or certain games. The 3060ti I had before never crashed. At first I thought it was AMD drivers (can’t really blame me with all the negativity in the social media forums) but someone suggested trying another PSU and it did the trick. No crashes and stable since April23.

As stated main issue was issues on post so didn't even get as far as running games or even loading windows. Large capacity quality psu's aren't cheap but may be able to borrow one off a friend, although don't know how old it is.

 

[SpudMaster]

As stated main issue was issues on post so didn't even get as far as running games or even loading windows. Large capacity quality psu's aren't cheap but may be able to borrow one off a friend, although don't know how old it is.

Are you running everything at stock? Lastest MB BIOS? Remove XMP and try running RAM slower or testing two stick at a time, 4 sticks can stress the CPU's IMC more

 

[Armageus]

CPU Cooler overtightened? PCI-E issue especially can often be caused by it.

 

[fizzy]

Are you running everything at stock? Lastest MB BIOS? Remove XMP and try running RAM slower or testing two stick at a time, 4 sticks can stress the CPU's IMC more

Everything at stock, no overclocking. Latest bios (as detailed in my initial post). XMP profile is on and passes memtest86 testing.

As stated, issue was at post... and seemed to resolve itself on removal of GFX card. Putting card back in and everything ok again.

 

[fizzy]

CPU Cooler overtightened? PCI-E issue especially can often be caused by it.

I'm using a custom water loop for the cpu. Attachments can't go too tight on the cpu waterblock due to spring/fitment. Double checked anyway and they seem fine.

Not had any such issues running this setup with my radeon VII for a couple of years - only seems to have started being tempremental with the 7900XTX card - which is why I was originally thinking it was a card issue but now not sure and hard to pinpoint exact problem.

 

[eeii]

Is the PC permanently on at the wall switch. With mine I turn on at the wall switch and wait about 10 seconds for all the capacitors to fill up. Then turn on at the pc. If I don’t wait 10 seconds I’ll get post errors with loads of beeps and it just won’t post.

 

[SpudMaster]

Everything at stock, no overclocking. Latest bios (as detailed in my initial post). XMP profile is on and passes memtest86 testing.

As stated, issue was at post... and seemed to resolve itself on removal of GFX card. Putting card back in and everything ok again.

Has it done it again since you reinstalled the card? Could just have been poorly seated in the PCI-E slot and worked its way out to the point it couldn't function?