Clock Watchdog Timeout

[toby12f]

System:

- EVGA RTX 3080 TI FTW3 Ultra Gaming OC
- Asus Rog Strix Z690-F Gaming Wifi
- Corsair Dominator Platinum DDR5 5600 MHz CL36 2x16 GB
- Intel i9-12900K
- Corsair H170i AIO (420mm radiator)
- Western Digital SN850 2 TB NVMe SSD
- SeaSonic Focus 850W 80+ Platinum

Apologies in advance if this is not the right spot for this question, in which case I would appreciate if you could point me in the right direction.

I keep getting CLOCK_WATCHDOG_TIMEOUT BSODs. I have tried with and without XMP II, with and without Asus AI overclocking, and even 100% default settings. I also thought it might have been because the motherboard jumper for additional CPU voltage wasn't 'enabled', so I moved it, but I am still getting this BSOD in <10 minutes on Prime95. The thermals on the CPU are all under 100°C on each core. Even with stock settings, the cores reach 5.3 MHz. I have been told this is an issue with the voltage, and I should raise VCore slightly, however I don't know how to do this. It even BSODs occasionally on idle loads, with only 1% of CPU in use. All my drivers are up to date, as is the BIOS. If the problem is indeed the CPU voltage, how do I go about raising it safely? Thanks again, any help is appreciated.

[rjbarker]

- load defaults in BIOS and reboot then
- flash bios to latest
- once again load defaults in bios after flashing latest bios reboot

With everything at default including Memory running at 4800 (default) try running Memtest to ensure no issues with memory, if you pass
then go ahead and run a number of benchmarks etc....if you continue to get Watchdog error BSOD then the only thing I could think to try would be a clean install of Windows 11 (assuming your running W11.

Also make sure all your Drivers are correctly installed including Clean install of GPU Drivers....

Then report back.

Dont bother enabling XMP until you do all the above first and am sure you have no BSOD's

[toby12f]

I have done all of that and it still BSOD. Also, Win 11 way too unstable for me, still 10.

[toby12f]

Edit: MemTest86 passed with 0 errors.

[Chainbold]

toby12f wrote:
Edit: MemTest86 passed with 0 errors.

This is good, because it shows that your RAM is not faulty. But it does not mean that your system is stable after booting into the OS. To test memory stability you need to run the P95 "large FTT" or the "blend test". Both tests do utilize all of your RAM to run the calculations.

I have almost an identical system like you, same CPU, GPU, board, and also the same Corsair AIO. I have seen it with my own Gaming-F, and on this forum as well with at least a dozen user reports: This board seems to have memory related problems. If you do not get the system stable and you are sure it is not the RAM, at last resort consider RMA. This is what I did with my Gaming-F. The second board was finally working.

toby12f wrote:
I was told to use that to test stability. If it crashes after 15 minutes in Prime, doesn’t that make it unstable and likely to crash on normal use? If I am wrong please correct me and I will use it and see.

You are not wrong. If P95 is crashing your system or producing errors, your system is not stable, even if it does run less demanding applications or games for a while. If you can't run P95, it is just a matter of time that the system will hang or crash with other applications as well. I think the mentioned 15 minutes are the minimum time to run P95. Many ppl are deluding themselves about system stability. An alternative for testing is Intel's XTU test and oc utility. It allows to test the CPU and the RAM separately, in stages. XTU however is running the same tasks as P95.

[Silent_Scone]

Chainbold wrote:



You are not wrong. If P95 is crashing your system or producing errors, your system is not stable, even if it does run less demanding applications or games for a while. If you can't run P95, it is just a matter of time that the system will hang or crash with other applications as well. I think the mentioned 15 minutes are the minimum time to run P95. Many ppl are deluding themselves about system stability. An alternative for testing is Intel's XTU test and oc utility. It allows to test the CPU and the RAM separately, in stages. XTU however is running the same tasks as P95.

Hello,

This philosophy is so outdated now that it's borderline flat-out incorrect. So much so that by leading users up this path you're potentially impeding their overclocking potential for no good reason. Unless it was built to run Prime. By endorsing this method other people will always be able to clock higher than you and be as stable as they need to be.

Remember, no overclock is ever 100% stable - that's something some users fail to wrap their heads around. For some reason, there are still Pro-Prime camps of users who like to subject their CPUs to copious amounts current that otherwise, their normal workload won't ever see. Depending on the workload, all you are doing is impeding your own overclocking range in real-world tasks. The fact Intel quickly acknowledged this by implementing the AVX offset function speaks volumes.


Even with all that aside, as far as stress testing memory is concerned there are far better tests than running large FFT (or small) Prime that isolate the memory subsystem. I and others in-house haven't used Prime to test stability since Haswell-E and use our systems crash-free on a daily basis. Of course, what you choose to take away from that depends on your mindset.

[rjbarker]

Silent Scone@ROG wrote:
Hello,

This philosophy is so outdated now that it's borderline flat-out incorrect. So much so that by leading users up this path you're potentially impeding their overclocking potential for no good reason. Unless it was built to run Prime. By endorsing this method other people will always be able to clock higher than you and be as stable as they need to be.

Remember, no overclock is ever 100% stable - that's something some users fail to wrap their heads around. For some reason, there are still Pro-Prime camps of users who like to subject their CPUs to copious amounts current that otherwise, their normal workload won't ever see. Depending on the workload, all you are doing is impeding your own overclocking range in real-world tasks. The fact Intel quickly acknowledged this by implementing the AVX offset function speaks volumes.


Even with all that aside, as far as stress testing memory is concerned there are far better tests than running large FFT (or small) Prime that isolate the memory subsystem. I and others in-house haven't used Prime to test stability since Haswell-E and use our systems crash-free on a daily basis. Of course, what you choose to take away from that depends on your mindset.

This perfectly said...I have been OC'ing for many years and my previous Gen stopped using P95 and Intel Burn Test, its archaic .

Besides...I game primarily, I certainly dont "play" Prime...the advice above looks like it was copied n pasted from an OC'ing thread in 2015 - 2017 😉

There are far better n faster methods to check for "General Stability"....get thru a few of these then game or carry on with whatever tasks you normally do and watch for instability.
Looping R23 for an hour...Realbench...IXTU ...and some Memory tests ....is fine and will catch instability pretty quick.

[Chainbold]

Silent Scone@ROG wrote:
Hello,

This philosophy is so outdated now that it's borderline flat-out incorrect. So much so that by leading users up this path you're potentially impeding their overclocking potential for no good reason. Unless it was built to run Prime. By endorsing this method other people will always be able to clock higher than you and be as stable as they need to be.

I disagree. 🙂 Intel itself is using P95 for their overclocking and stressing utility. They know why.

There are of course different ways to test stability, even with a hardcore utility like P95 - depending on the purpose of your system and the stability level you want to achieve. And there is no doubt that P95 (or similar y-cruncher) in particular running with AVX-512 enabled is stressing the system in a way that is far above regular usage. If you want to go soft with P95, disable AVX instructions and run small FFT. But one thing is for sure: No matter how "soft" you intend to test your system, if the P95 blend test is producing errors, freezes, or crashes after only a few minutes running (with AVX instructions disabled) it is 100% for sure you are not stable and will eventually crash. It's a guaranteed event - and telling users here a different story may create unrealistic expectations and eventually frustration.

Also, as I said before: A passed Memtest is not indicating system stability under OS conditions. It only shows that that memory is running fault free under the given frequency and timings. E.g. Memtest is running my DDR5 at 6000MHz 32-32-32 for hours. If I run the P95 memory test I get errors within a few minutes. And I suppose this is the reason why, for example, when gaming I got occasional freezing or crashes with the memory at 6000MHz.

Silent Scone@ROG wrote:

The fact Intel quickly acknowledged this by implementing the AVX offset function speaks volumes.

AVX instructions have been implemented more than 10 years ago, with Sandy Bridge. They are used by many applications. If you want to test without AVX instructions because you think they are stressing your system too much, you can run P95 or Intel XTU without them.

Silent Scone@ROG wrote:
Of course, what you choose to take away from that depends on your mindset.

Exactly.

For real, hard core overlocking, testing and validating see here, how they tested:

https://www.igorslab.de/en/supercool-computers-direct-the-water-block-and-delid-tool-for-alder-lake/...

They used y-cruncher with AVX-512 enabled. Their target was to demonstrate highest possible CPU frequency with their direct-die waterblock. They got an 12900 up to 5.5 GHz. For bragging, they certainly could have shown a higher value with a softer stability test.

rjbarker wrote:

Looping R23 for an hour...Realbench...IXTU ...and some Memory tests ....is fine and will catch instability pretty quick.

Intel's XTU stress test is by P95. In the same way as you can do with P95, XTU let you disable AVX instructions, if you want to run a softer test.

[Jimbo93]

Chainbold wrote:
I disagree. 🙂 Part of the oc community was always over-optimism and self delusion. Each to its own. Intel itself is using P95 for their overclocking and stressing utility. They know why.

There are of course different ways to test stability, even with a hardcore utility like P95 - depending on the purpose of your system and the stability level you want to achieve. And there is no doubt that P95 (or similar y-cruncher) in particular running with AVX-512 enabled is stressing the system in a way that is far above regular usage. If you want to go soft with P95, disable AVX instructions and run small FFT. But one thing is for sure: No matter how "soft" you intend to test your system, if the P95 blend test is producing errors, freezes, or crashes after only a few minutes running (with AVX instructions disabled) it is 100% for sure you are not stable and will eventually crash. It's a guaranteed event - and telling users here a different story is only creating unrealistic expectations and eventually frustration.

Also, as I said before: A passed Memtest is not indicating system stability under OS conditions. It only shows that that memory is running fault free under the given frequency and timings. E.g. Memtest is running my DDR5 at 6000MHz 32-32-32 for hours. If I run the P95 memory test I get errors within a few minutes. And I suppose this is the reason why, for example, when gaming I got occasional freezing or crashes with the memory at 6000MHz.



AVX instructions have been implemented more than 10 years ago, with Sandy Bridge. They are used by many applications. If you want to test without AVX instructions because you think they are stressing your system too much, you can run P95 or Intel XTU without them.



Exactly.

For real, hard core overlocking, testing and validating see here, how they tested:

https://www.igorslab.de/en/supercool-computers-direct-the-water-block-and-delid-tool-for-alder-lake/...

They used y-cruncher with AVX-512 enabled. Their target was to demonstrate highest possible CPU frequency with the with the direct-die waterblock. They got the 12900 up to 5.5 GHz. For bragging, they certainly could have shown a higher value with a softer stability test.

I actually agree with Chainbold here. But that is because I demand stability. Im still not sure I am going to keep the memory at XMP1, although it has now been set that way for maybe 2 months or so. I do occasional important work on my system, I cant afford corruption of any kind when I do. If I cant get good reliability scores in windows the system maybe cannot do what I need when I need it. But I can certainly see people with gaming systems that do not need that. Now if its just a software bug causing crashes that is different, of course. Just dont run that software. Sometimes the only way to be sure is to run the system on bios defaults.

Ive been using pc s for business and pleasure a long time and started using them in the early days of micro computers.

[Chainbold]

Jimbo93 wrote:
I do occasional important work on my system, I cant afford corruption of any kind when I do. If I cant get good reliability scores in windows the system maybe cannot do what I need when I need it. But I can certainly see people with gaming systems that do not need that.

This is also my point. I'm willing to forsake some additional MHz in return for higher stability. In addition to gaming, I use my system constantly for video editing for business purposes, I cannot afford freezing or crashes. But do not underestimate the stability requirement for gaming. Some of the most popular games, like Far Cry 5 or Rainbow 6 are very CPU demanding.