Should I RMA a Ryzen 5900x that CTD's with certain games, but fine with CPB off?

[Akolonus]

Hello,

I bought a Ryzan 5900x from OCUK last November and have had problems with it since the beginning.

Firstly it wouldn't Post with an MSI MAG x570 Tomahawk until the BIOS was updated, which then allowed me to Post but was getting multiple BSOD errors. So I dusted off my trusty 3800x Ryzen and have been waiting for newer BIOS to stabilize the CPU.

During this time I switched motherboards to a Asus TUF x570 Gaming Plus and updated my Bios to version 3602 just recently. So I put in the 5900x again to see if it was any better now. Since Thursday of last week (25/3/21) I've still recieved two BSOD of (IRQL_NOT_LESS_OR_EQUAL) whilst the CPU has been powering down from activity.

Strangely, when I try Prime95, or OCCT and even Intel Burner, the tests go through with flying colours. But it is a different story with gaming, especially with Total War: Warhammer 2, where the game would crash on campaign map or specifically when finishing manual combat.

However, turning off Core Performance Boost in the BIOS, stops the crashes from happening fully but of course makes the CPU only run at its minimum 3700Ghz. The game was also stable with the 3800x installed even with Core Performance Boost on.

I'm concerned that since I've never really had much stability with this chip that it might be a dud chip and that I should RMA it.

I've already spoke to a Customer Service Agent from OCUK who says I should RMA it back and they'll test it but worry that because Prime95 and other stability checks maybe successful, that the chip will be classified "ok" and be sent back to me, but knowing myself that something is wrong with it, due to the CPB test mentioned above.

Temperatures on Stability checks have risien to 82 degress, but no more than that and generally stays in the 60-62 on full load and 33 when on Idle. Power settings are set to High Performance in Windows too.

Can anyone please advice me on whether I should RMA the chip or not.

Specs as follows:

AMD Ryzen 5900x
Asus x570 TUF Gaming-Plus (BIOS Version 3602)
Corsair H115 Water Cooling
GameMAX 850w PSU
T-Force Darkz Memory (16x2)
NVME Corsair 980 EVO 1T
Nvidia Cereberus 1070ti
Windows 10 Pro

 

[Alex h]

Rma it and try to capture message on screen when it crashes so hopefully ocuk can sort it for you

 

[RichL]

RMA to AMD directly

 

[Grim5]

What's your memory settings for the 5900x? I do find the 5000 series are less stable with memory clocks over 3200mhz

 

[Journey]

Can anyone please advice me on whether I should RMA the chip or not.

I'll echo as above, just use AMD's RMA service it will be faster, cheaper and more likely to result in you getting a new CPU.

 

[Ree]

currently going through a amd rma and man its been very quick and as straightforward as you get. cpu arrived at amd in holland on friday and my replacement cpu was shipped the same day after my processor was checked. cpu should be arriving shortly.

 

[Nicholas Eagle]

Shame that AMD doesn't do some sort of advanced replacement option, in such times it's hard to take a pc out of play while there's such a reliance on them.

I'd quite happily have a holding charge on my cc to have one shipped and then I send the faulty one back.

 

[HangTime]

If it only happens in games could it be indicative of a PSU problem? i.e. system power draw is probably higher when GPU is being heavily utilised in addition to the power boost for the CPU. If you have another PSU to hand it might be worth testing with that.

 

[OMGsideboob]

You've definitely got a dud chip. There wasn't a QC process in place for the 5000 series and many 5900X CPUs are suffering catastrophic instability under stock factory settings. Your CPU should have never made it past the production line, let alone into retail distribution.

If you don't mind relinquishing your CPU for some time (or have access to a spare), there's no reason not to RMA. Due to the peculiar and erratic way in which the 5900X exhibits this instability (it can remain perfectly stable under most test conditions but crashes randomly on idle or during gaming loads, or it could go for days without crashing but then begin to fail every 30 minutes), the instability might be difficult for them to replicate in their lab and they might very well send it back saying there's nothing wrong with it.

They could also very easily just tell you to handicap the performance and offer that as a viable solution, rather than acknowledge that you've clearly got a defective product. Remember, if it doesn't work at stock settings and requires its performance to be reduced, then it's totally defective.

(Not that this makes any difference to your CPU being defective, but just for fun, try this out: keep CPB on but have a little play around with curve optimiser to overvolt your voltage-frequency curve. This seems to improve the stability in some cases. The entire purpose of curve optimiser is to undervolt the CPU and allow it to reach higher frequencies, so this fix isn't really a fix at all because it hurts performance. Here's how to do it: in the "advanced" section of PBO/curve optimiser, set an all-core positive offset of +5 or +10 (as an initial test) and see if that does anything. I have what seems to be the exact same problem as you with my 5900X and this has improved the stability a little. If you can identify which specific core is defective in your CPU, you can try to set a positive offset on that but try a negative offset, or undervolt, on all other cores.)

 

[LtMatt]

Hello,

I bought a Ryzan 5900x from OCUK last November and have had problems with it since the beginning.

Firstly it wouldn't Post with an MSI MAG x570 Tomahawk until the BIOS was updated, which then allowed me to Post but was getting multiple BSOD errors. So I dusted off my trusty 3800x Ryzen and have been waiting for newer BIOS to stabilize the CPU.

During this time I switched motherboards to a Asus TUF x570 Gaming Plus and updated my Bios to version 3602 just recently. So I put in the 5900x again to see if it was any better now. Since Thursday of last week (25/3/21) I've still recieved two BSOD of (IRQL_NOT_LESS_OR_EQUAL) whilst the CPU has been powering down from activity.

Strangely, when I try Prime95, or OCCT and even Intel Burner, the tests go through with flying colours. But it is a different story with gaming, especially with Total War: Warhammer 2, where the game would crash on campaign map or specifically when finishing manual combat.

However, turning off Core Performance Boost in the BIOS, stops the crashes from happening fully but of course makes the CPU only run at its minimum 3700Ghz. The game was also stable with the 3800x installed even with Core Performance Boost on.

I'm concerned that since I've never really had much stability with this chip that it might be a dud chip and that I should RMA it.

I've already spoke to a Customer Service Agent from OCUK who says I should RMA it back and they'll test it but worry that because Prime95 and other stability checks maybe successful, that the chip will be classified "ok" and be sent back to me, but knowing myself that something is wrong with it, due to the CPB test mentioned above.

Temperatures on Stability checks have risien to 82 degress, but no more than that and generally stays in the 60-62 on full load and 33 when on Idle. Power settings are set to High Performance in Windows too.

Can anyone please advice me on whether I should RMA the chip or not.

Specs as follows:

AMD Ryzen 5900x
Asus x570 TUF Gaming-Plus (BIOS Version 3602)
Corsair H115 Water Cooling
GameMAX 850w PSU
T-Force Darkz Memory (16x2)
NVME Corsair 980 EVO 1T
Nvidia Cereberus 1070ti
Windows 10 Pro

Does it happen with the system at BIOS default settings with the memory at 2133Mhz (default auto speeds)? If so then yes RMA. If no, then something is not stable and the CPU is unlikely to be faulty.

 

[Joxeon]

Can anyone please advice me on whether I should RMA the chip or not.

RMA it direct to AMD not OCUK as they will test it with stuff like prime etc and it will likely pass so they will just return it to you.

 

[Destination]

(IRQL_NOT_LESS_OR_EQUAL)

Have you any other cards in the machine?
IRQL is an odd one, used to be when you had cards in a funny slot, or in slots that didn't like one another.

 

[welshrat]

Pbo off should still allow boost to 4.8ghz. I have had that crash before and it's usually a driver issue. Mine was a sound blaster issue which I resolved with later driver

 

[pastymuncher]

I wouldn't put it past the psu being at fault. Any 850w psu that only has a pair of 6+2 pci-e connections should raise questions about it's quality. Even my 650w has four of them! Most places that I can find selling it only state that it has a 1 year warranty which is the minimum required by law. Any decent quality psu should have 5, 7, 10 or more years warranty. There are no proper reviews and very little information available on these psu's which should flag up a big warning. That is not a psu that I would ever buy or recommend.

 

[kitfit1]

I wouldn't put it past the psu being at fault. Any 850w psu that only has a pair of 6+2 pci-e connections should raise questions about it's quality. Even my 650w has four of them! Most places that I can find selling it only state that it has a 1 year warranty which is the minimum required by law. Any decent quality psu should have 5, 7, 10 or more years warranty. There are no proper reviews and very little information available on these psu's which should flag up a big warning. That is not a psu that I would ever buy or recommend.

+1

I think both you and me @pastymuncher still have vivid memories of cheap "high powered" psu's from the early 2000's. They actually never were any good and certainly didn't last very long...............................the 1 year warranty might even be optimistic when paired with a 5900x.

 

[jim_britton]

this is quite an old thread now so hopefully it's all resolved, but I had crashes in similar circumstances with my 5800x. Just gaming - OCCT, Prime95 etc worked fine. I RMA'ed it back to OCUK and they shipped a replacement the same day and I've not had a crash since.

Although, my crashes were the more typical WHEA errors. IRQL_NOT_LESS_EQUAL errors are usually related to some driver or other, aren't they?

 

[pastymuncher]

+1

I think both you and me @pastymuncher still have vivid memories of cheap "high powered" psu's from the early 2000's. They actually never were any good and certainly didn't last very long...............................the 1 year warranty might even be optimistic when paired with a 5900x.

(Cough) Hiper Type R 580w (Cough). Biggest mistake I ever made when choosing pc components and it cost me dearly!!

 

[StephenJ]

(Cough) Hiper Type R 580w (Cough). Biggest mistake I ever made when choosing pc components and it cost me dearly!!

I bought the same PSU but luckily had no issues

 

[HangTime]

Haha Hiper PSUs that brings back memories, they were getting recommended left right and centre on here then a few months later all the tales of woe. Up there with the legendary IBM DeathStar hard drives!

 

[nlel1975]

I wouldn’t let one bad experience with a PSU rule the rest of your life.