Randomly stuck during POST at Q-Code 65

[langusiii]

Hi there,

Bought a new rig, Z690 Formula, i9-13900KF, 32GB DDR5 6400 and GForce 3080 (see my profile for more details) for my daughter last Nov/2022 and the only reason I am not throwing it away is the $6K I spent so far.

In average 4 of 5 attempts with no pattern, it gets stuck during POST with a Q-Code 65 (CPU DXE Initialization Started). For that 1 of 5 times in which it passes that point, I am able to go into the BIOS to configure it or just let it continue into the OS without any further problem (so this is completely unrelated with the OS/drivers).

Given that the DRAM is not in the QVL (uglies Corsair Venegance and Dominator are in, but nicer Venegance RGB is not) I purchased another set of ugly Kingstons 6400 MB/s within the QVL, but the problem persists. Also purchase an i9-13900K (with video support) resulting in the same. Purchased a Z690 Hero with the same results. Finally last night arrived another Z690 Formula for recommendation of Asus Customer Service, but the outcome is still the same.

Updated the BIOS to 2204 several times, updated the ME Update to latest, ensured the CPU / Socket don't have bended pins / dirt, clear CMOS infinite times to start from scratch, tried with XMP I, II, Tweaked, AI-Manual, from Intel Extreme Tunning utility, writing manually the resulting ok values, raising a little the RAM voltage from standard but still within range, disconnecting everything but essential, etc... even Best Buy Geeks... all with the same results!! (Note: I am not trying to OC... yet... just want a normal boot 99% of the time as everyone else).

Not many resources online, just a few posts from 2012, also from other manufacturers, without any conclusive proposal. As the result, I feel there is nothing wrong with each hardware component, but I am pretty sure is some variable/s I am not able to find which are and to which non-standard-value should be set.

Anyone with some experience in how to solve this code 65 (a very few times was a 63, and lesser others I don't remember) or any tips will be really appreciated.

Thanks in advance, geniuses!

Langus

[JohnAb]

Never come across that code before, it seems to be related to loading of services for the processor, chipset and other components prior to loading the OS. Have you tried disabling fast boot (not helpful, that's a Windows setting, Doh) and turning off multi-core enhancement in BIOS for now? Also, if you haven't tested the RAM, do you get stable results in MemTest 86 (available in BIOS menu).

I don't really have any clear idea about your issue, sorry, but the above might be worth a try. BIOS 2204 seems to have positive feedback generally so probably not the issue, but you could go back to 2103 (assuming that your BIOS versions are the same as the Hero BIOSs) and try that version if you haven't done so. It also seemed good for most people. For that BIOS, ME Firmware v16.1.25.1885 was recommended, but you can use later ME firmware, so should be good just to try the older BIOS and leave the firmware where it is for now. Assume all BIOS settings are on defaults as well and you've tried with XMP off?

You might also try BIOS/Advanced/Platform Misc Configuration/ASPM = Disabled

I hope one of those ideas helps 😞 I think my money might be on the RAM at this point. XMP off is certainly worth a go.

If not, there may be settings around memory voltages and CPU LLC curves that could help. I won't try and advise on those here (also because I have no experience of changing those settings on this platform myself and don't want to advise wrongly), just saying that there are other options to try if the above is not helpful.

[langusiii]

Hi John, thanks for your reply.

However as you presume, disabling Fast Boot didn't help. Same results with ASUS Multicore Enhancement set to Disabled - Enforce All Limits. The weird thing is that this happens before the "Press F2/Del to enter BIOS" with all Standard (CMOS Cleared) Settings, once passed that point everything runs smooth!!

I read some post looking into raising the Memory Controller Voltage, so from the RAM specs of 1.35v I raised to 1.3875v (result from a previous successful POST) but is still happening, seems less frequent (let's say 1 success out of 3 times), but still happening. But given that I allow to be adjusted (I don't remember where is that variable), now the BIOS shows 1.403v after a new successful POST.

Last tries this week was to connect the case fans to the motherboard that were previously to the iCUE Commander Core, same results. Disconnecting them (not the CPU Fan) has also the same results.

Could the GeForce 3080 be causing this? I mean thru a power drainage during POST? The PSU seems pretty solid Corsair with 850w, but I never took the 3080 out due the CPU is a KF (with no video support).

Other thing I noticed when it starts, is that Q-Codes shown runs on CPU numbers 2 or 3 times starting from 00, then passes to Memory codes, then comeback to CPU 1 or 2 more times, and if it sticks thru a 55 and 31 for a little almost imperceptible longer time, then from 31 jumps to 65 and gets stuck. But when it is a little less time on those 55 and 31, instead it goes to an 86 and then flows thru the VGA, HDD, and voila... into BIOS or OS... Maybe I am starting seeing ghosts there, but there looks like a pattern... I can predict when it's going to fail into 65 (very few times a 63, just got a 03, saw a D1).

I am completely lost and frustrated now... I like it when it works, but I cannot use it like this...

Any further help or suggestions, REALLY appreciated...

Rgds,
Langus

[langusiii]

Hi John,

I read your thread about advice to update the Intel ME and BIOS, and what I can see is that I did it in the wrong sequence. My initial thought getting this error was the BIOS, so I first update it, then I got more patience to wait for a POST and boot into windows, then installed all drivers and ME. I did the same with all 3 motherboards I tested (Z690 Formula, Z690 Hero and Z690 Formula) getting the same freaking Q-Code 65.

If the sequence might have impacted in generating my problem, is there anyway to make it right now?

My conclusion so far is that this is not OS related giving that the error is before POST is ended and handled to UEFI. Also it is not hardware related given that I have tried different motherboards, cpus, and memories within QVL with the same result. I could only be BIOS (or as you pointed Intel ME) related and as it eventually booting right (now has improved to something like 1 success out of 3 attempts), the BIOS / ME are changing / learning variable during the startup.

I have tune every variable I thought it might have an Impact in the BIOS, but I still have the same results. Haven't been able to spend much time in the last few days, but recorded some booting stats (attached).

ASUS Customer Support is useless, just recommended to RMA the 3rd motherboard (I cannot believe it :-|), any further ideas besides getting patience until the next BIOS update hopping that it will solve it? Is there any way to get in touch with the ASUS BIOS Team?

Rgds,
Langus

[JohnAb]

Hi Langus, sorry I missed your 2nd post, so just trying to catch up. As far as ME is concerned, installing in the wrong order can cause some temporary issues, but nothing permanent so if you have matched BIOS and ME firmware now, I don't think that will be it. I was going to suggest double checking all connections but if you have changed the motherboard, no reason to think that's the problem.

You seem to have some strange issues going on and I don't really have much to suggest, apart from possibly the PSU. Have you tried another one? It's unlikely I think, but you seem to have changed everything else.

Also, could there be a short somewhere? Check you support posts and make sure that you don't have an extra one which could be contacting the motherboard PCB somewhere. I saw a post on here somewhere recently where somebody was saying that a model of the 790 boards should have a post removed for that reason. I know you have a 690, but no harm in double checking, just in case you have an extra support post or one in the wrong place.

Otherwise check all of your cables, especially from the PSU. For example, could there be a bad plug or a damaged/bent pin on a connector?

Like I said, I'm struggling, just throwing out some possibilities. You might also try a total power off, kill the PSU switch and then try another BIOS reset. But again, probably not going to help. There must a be a cause somewhere... keep trying

PS - I don't think you have tried another GPU right? That's also worth a try, can you borrow one or even try an old cheap one just to eliminate your current GPU as a possible cause?

[langusiii]

Hi John,

I have not tried another GPU nor a PSU yet. But I did disconnect everything else (fans, iCUE, case connectors, USBs, keyboard & mouse, etc) but CPU fan and mobo + GPU feed. Same results.

However, I have a doubt. I am using the same cable with dual plugs to feed both 8pins connectors on the GPU. Should I use two different cables coming from the PSU?

I'll give a try to use the 13900K (with video support, that I have not returned yet) taking out the GPU and report. If also fails, I could also try an old PSU (not sure if the connectors will be still valid) or just order a new one and try.

Regarding the cables in short, I doubt it, it is a pristine brand new PSU. However it could always have a bad connector that you'll never see inside the plug. I'll try to turn the cables around, to see if the error moves the problem to another place.

For how the case is located on my desk, killing the PSU switch are 30% of my reboots, 70% are the buttons on the board. And I also tried to clear CMOS being off the wall and powered but in off (just with the mobo lights on). Figured it's exactly the same as "load optimized values and reset" setting in the BIOS exit menu.

Is there any way to connect some sort of FT232RL to open a terminal and see something additional on the Q-Code 65? Like some sort of BIOS Log (I know there is nothing openly accessible) or so to identify a more specific place where it fails?. Years ago I de-bricked a router this way.

I already entering in the "resignation phase", and learning how to live with it :D... What it puzzles me is that on those times that gets thru the-freaking-point, there is no more problems at all, even under the stress tests (within windows) heats up, speeds up the fans, but never breaks the 100 Celsius barrier.

Will be back.

Thanks,
Langus
PD: Forgot to mention that I left MemTest86 running for 5 hours with a no-errors result

[pennyboy]

langusiii wrote:
Hi John,

I have not tried another GPU nor a PSU yet. But I did disconnect everything else (fans, iCUE, case connectors, USBs, keyboard & mouse, etc) but CPU fan and mobo + GPU feed. Same results.

However, I have a doubt. I am using the same cable with dual plugs to feed both 8pins connectors on the GPU. Should I use two different cables coming from the PSU?

I'll give a try to use the 13900K (with video support, that I have not returned yet) taking out the GPU and report. If also fails, I could also try an old PSU (not sure if the connectors will be still valid) or just order a new one and try.

Regarding the cables in short, I doubt it, it is a pristine brand new PSU. However it could always have a bad connector that you'll never see inside the plug. I'll try to turn the cables around, to see if the error moves the problem to another place.

For how the case is located on my desk, killing the PSU switch are 30% of my reboots, 70% are the buttons on the board. And I also tried to clear CMOS being off the wall and powered but in off (just with the mobo lights on). Figured it's exactly the same as "load optimized values and reset" setting in the BIOS exit menu.

Is there any way to connect some sort of FT232RL to open a terminal and see something additional on the Q-Code 65? Like some sort of BIOS Log (I know there is nothing openly accessible) or so to identify a more specific place where it fails?. Years ago I de-bricked a router this way.

I already entering in the "resignation phase", and learning how to live with it :D... What it puzzles me is that on those times that gets thru the-freaking-point, there is no more problems at all, even under the stress tests (within windows) heats up, speeds up the fans, but never breaks the 100 Celsius barrier.

Will be back.

Thanks,
Langus
PD: Forgot to mention that I left MemTest86 running for 5 hours with a no-errors result

I would use separate rails for the video card. However, given that your PC is stable when it does POST I doubt that is the problem. Have you tried running your memory at the native max speed for the Z690 chipset, which I think is 4800Mhz.
You mentioned "tried with XMP I, II, Tweaked, AI-Manual, from Intel Extreme Tunning utility", however all these settings are going to Overclock your memory, anything over 4800mhz is technically an overclock on the Z690 chipset. Try setting Ai Overclock tuner to 'Auto', which should run your RAM at 4800.

[langusiii]

Thanks JohnAB and pennyboy!

Just want to report that I finally found the problem :cool:. Short story, it was the i9-13900KF CPU :eek:.

Using an i9-13900K processor in my original Z690 Formula, it actually showed the Q-Code 65 in a couple of tests. However, it did not showed when I used the same i9-13900K in a Z690 Hero, neither when I used that same i9-13900K in the replaced Z690 Formula. So, I think there were two overlapping problems related with the original Z690 Formula and the KF CPU. Not sure if the BIOS could be influencing, but I would suggest to responsible team to take a look at this setup and contact me if you have further questions.

In the meantime, discarded the PSU and PSU Cables using a new unit. Also discarded the Cooler using a new cheap cooler, and finally discarded the Thermal Paste using a new one (yes, a freak, I didn't know what else to try). And found no difference using one cable with two plugs or two cables to feed the 3080.

Also given that I lost the window to return it to the retailer (Amz), I opened a claim with Intel Customer Support, and upon studying my case, they decided to fully refund the item. With fully I meant what I paid to Amz including taxes plus the label for the shipping. We choose the refund option given that they were temporally out of stock of KF CPUs so I kept the K CPU.

Finally, I can report that Corsair Vengeance RGB (CMH32GX5M2X6400C38) which IS NOT in the QVL, WORKS FINE with this Z690 Formula and i9-13900K, even using XMP1 at 6400Mb/T.

Hope this helps somebody else to troubleshoot their system, it took me almost 3 months to get it working properly!

Rgds,
Langus

[JohnAb]

Well I would certainly try both cables on the GPU. Splitters are not recommended on modern cards and the 3080 can draw 350W at max power. That's high for a single cable. Whilst I don't think it would draw anything like that much power at boot, having just a single cable plugged in could have caused a pin to have been overloaded at some point, resulting in a poor electrical (high resistance) contact. Could that be the cause? I don't know, but I think it's worth further investigation for sure.

In addition if you have that alternative CPU with iGPU then yes, also a good idea to try that and unplug the main GPU. At least it might eliminate it as a cause.

I know nothing about getting extra info from the BIOS via a terminal I'm afraid.

I'd certainly try the GPU power leads and if that fails, then swap the processor and remove the GPU. I don't think the PSU is very likely to be honest, so I'd suggest you try that option last. At least with another PSU, you'd get a fresh set of leads as well. Expensive way to trouble shoot though. Good luck....

[JohnAb]

That's excellent news - good job. People say that CPU faults are pretty rare, and I guess they are, but I think you're the 2nd or 3rd person with a faulty CPU, at least on these forums. Very good of Intel to give you a complete refund too. That's good service 🙂 It really shouldn't take that much trouble to get everything working, so well done you!