9600k hwinfo64 effective clock readings inconsistent between OC's

[Mister Woof]

1 minute ago, SGTChalk said:

I went and looked for that setting, couldn't find it for my mobo, but I'm also not using turbo for this OC. Worth noting that once Throttlestop has been run, at least for me, even if it's closed the issued remains resolved until the next reboot, so I've just been having it run at startup and then closing it to save the admittedly itsy bitsy amount of resources it uses.

I've seen a similar sentiment about the effective clocks in hwinfo64 elsewhere, but from the testing I've done it's been extremely consistent with it's pre and post Throttlestop run readings, and those readings have been very clearly tied to huge differences in bench scores in Cinebench R20. Pre running Throttlestop my 9600k running 4.7ghz all core scores anywhere from 2400-2600s in Cinebench R20 with with hwinfo64 effective clocks reading roughly 4450-4550mhz core 0, and 4550-4650ish on the remaining cores, but after I run Throttlestop hwinfo64 reads a steady 4700mhz effective all cores, and it scores consistently in the 2800's

ditto - just starting throttlestop fixes weird issues.

 

Dude deserves a medal 

[Imannudein]

On 4/4/2021 at 8:59 AM, Mister Woof said:

ditto - just starting throttlestop fixes weird issues.

 

Dude deserves a medal 

Same here. I've been running TS ever since I found this thread. Weird. I'm beginning to think that this is a Windows issue rather than hardware. 

 

[KDYZ]

Same here. I have 10700KF with MSI Z490 gaming edge wifi, temps are great, case airflow is great, power consumption is fine, no throttle flags anywhere in HWINFO64, but after 2-20 minutes of Prime95 (it seems to be random, no relation to temps or power) the CPU throttles back by cutting C0 state occupancy to around 98% (90-95% in the case of core0).

 

I tried literally every combination of bios settings, OC, stock, undervolted, disabled and re-enabled most of the CPU options one by one and in various combinations, nothing works. It always throttles back at some point. Once throttled, it stays that way until the next reboot, even if the system is left idle for several hours. I don't see how it can be CPU related, it seems a good chip. It was easy to get it stable at 5GHz, and only needs about 1.175V (maybe less, I haven't pushed it lower yet) to hit 4.7GHz on all cores.

 

I only installed TS to see if I could find out what was throttling it, but it won't tell me because (as the OP and others have said) just opening it is enough to prevent the throttling. I also don't need to keep it open, it's enough to open it once and close it immediately. I have even tried changing TS settings to see if I can make it throttle again, but I can't reproduce the behaviour with TS open.

 

I would be really interested if you ever manage to figure out what TS is doing at open that prevents whatever is going on here!

 

P. S. I just found out if I run TS straight from boot it shows yellow PL1 under Core, and yellow EDP OTHER under Ring. If I then run HWINFO64, both flags disappear. In any case, they never turn red, and C0 occupancy stays at 100%, so there is no throttling going on despite the flags. I don't really know why they are there, PL1 and PL2 are set to 4096W in BIOS.

[unclewebb]

1 hour ago, KDYZ said:

straight from boot it shows yellow PL1 under Core, and yellow EDP OTHER under Ring

It is not unusual to see a couple of yellow boxes just after booting up. Yellow boxes indicate a record of throttling. This might happen when you first push the power button before the BIOS has a chance to set the turbo power limits. Initial yellow boxes are not important.

 

1 hour ago, KDYZ said:

If I then run HWINFO64, both flags disappear.

HWiNFO continuously resets the throttle flags within the CPU. That makes it difficult / impossible for other software to track what is going on. 

 

1 hour ago, KDYZ said:

PL1 and PL2 are set to 4096W in BIOS

It is best to set PL1 and PL2 to 4095. Some motherboards are screwing up this setting. The maximum value the power limit register can hold is 4095.875. If the BIOS actually tries to set this to 4096, it will wrap around and be set to 0.

 

1 hour ago, KDYZ said:

I would be really interested if you ever manage to figure out what TS is doing

I will look into this further. No idea how many 10th Gen CPUs have this problem. It could be a bug at the CPU level. There are lots of bugs that Intel does not like to talk about.

 

[Mister Woof]

1 minute ago, unclewebb said:

It is not unusual to see a couple of yellow boxes just after booting up. Yellow boxes indicate a record of throttling. This might happen when you first push the power button before the BIOS has a chance to set the turbo power limits. Initial yellow boxes are not important.

 

HWiNFO continuously resets the throttle flags within the CPU. That makes it difficult / impossible for other software to track what is going on. 

 

It is best to set PL1 and PL2 to 4095. Some motherboards are screwing up this setting. The maximum value the power limit register can hold is 4095.875. If the BIOS actually tries to set this to 4096, it will wrap around and be set to 0.

 

I will look into this further. No idea how many 10th Gen CPUs have this problem. It could be a bug at the CPU level. There are lots of bugs that Intel does not like to talk about.

 

Just wanted to say, thank you for your service.

[KDYZ]

1 hour ago, unclewebb said:

It is not unusual to see a couple of yellow boxes just after booting up. Yellow boxes indicate a record of throttling. This might happen when you first push the power button before the BIOS has a chance to set the turbo power limits. Initial yellow boxes are not important.

 

HWiNFO continuously resets the throttle flags within the CPU. That makes it difficult / impossible for other software to track what is going on. 

 

It is best to set PL1 and PL2 to 4095. Some motherboards are screwing up this setting. The maximum value the power limit register can hold is 4095.875. If the BIOS actually tries to set this to 4096, it will wrap around and be set to 0.

 

I will look into this further. No idea how many 10th Gen CPUs have this problem. It could be a bug at the CPU level. There are lots of bugs that Intel does not like to talk about.

 

Thanks very much for your reply! The Auto settings in my BIOS default to 4096, but TS reads 4095 so I guess the BIOS settings are ok in this case. I will set it to 4095 manually just to be sure.

[SGTChalk]

On 4/9/2021 at 6:51 AM, KDYZ said:

Same here. I have 10700KF with MSI Z490 gaming edge wifi, temps are great, case airflow is great, power consumption is fine, no throttle flags anywhere in HWINFO64, but after 2-20 minutes of Prime95 (it seems to be random, no relation to temps or power) the CPU throttles back by cutting C0 state occupancy to around 98% (90-95% in the case of core0).

I noticed this exact behavior down to the letter the first time I'd put in a new OC, but after it had happened once it would go straight to the hidden throttle after reboots (C0 state and all). Thanks for coming in, this really isn't seeming like an isolated thing with multiple people finding this thread and sharing the same problem, there's got to be a bunch out there

[Imannudein]

I just realized one thing: I begin to observer this behavior after I updated my BIOS to 2004 on the Maximus 12 Hero WiFi. This update then automatically applied the latest Intel ME Firmware Version 14.1.51.1528. 

 

I'm no expert but I cannot discount this factor as nothing changed aside from these 2 elements. 

[KDYZ]

On 4/11/2021 at 10:18 PM, SGTChalk said:

I noticed this exact behavior down to the letter the first time I'd put in a new OC, but after it had happened once it would go straight to the hidden throttle after reboots (C0 state and all). Thanks for coming in, this really isn't seeming like an isolated thing with multiple people finding this thread and sharing the same problem, there's got to be a bunch out there

In my case it is definitely intermittent. I have been testing quite a lot recently (with help from @unclewebb) and changed something in BIOS (I set Speedshift to 'On' rather than 'Auto'), after which I didn't see the reduced C0 state residency for a few days. But today it is back, and worse than before (with some cores averaging around 70%), and there was no delay at all this time - right from starting P95, no core managed to hit 100% C0, and most were well below. Previously it has always taken a while under load to reduce C0 - normally about 10 minutes of P95, but in one case, just a few loops of Heaven benchmark. And then today, it seemed to happen without any significant CPU loading at all, perhaps straight from boot.

[SGTChalk]

6 hours ago, KDYZ said:

In my case it is definitely intermittent. I have been testing quite a lot recently (with help from @unclewebb) and changed something in BIOS (I set Speedshift to 'On' rather than 'Auto'), after which I didn't see the reduced C0 state residency for a few days. But today it is back, and worse than before (with some cores averaging around 70%), and there was no delay at all this time - right from starting P95, no core managed to hit 100% C0, and most were well below. Previously it has always taken a while under load to reduce C0 - normally about 10 minutes of P95, but in one case, just a few loops of Heaven benchmark. And then today, it seemed to happen without any significant CPU loading at all, perhaps straight from boot.

Interesting, the whole time I've seen it I've had speedshift, speedstep and turbo off. I only noticed the delay on the first boot after fresh bios changes, usually it has been immediate, but I also wasn't tracking that closely

[lockzye]

I know that it's an old thread by now but I had the same problem as you guys, effective clock throttling. I tried everything as you guys did with no happy ending and settled in for throttle stop. Recently I think I found the solution, at least for me. I've been using for most of my overclocks a modded version of bios with BE microcodes (which are apparently are the fastest for my 9900k) it was showing that I was using DE microcodes in windows though. I've read that I had to actually uninstall a windows update to get the older microcodes to work In my case it was this update KB4589212, after uninstalling it I haven't seen any throttling on effective clock so far. Try uninstalling the update KB4589212 by going to control panel - uninstall apps and there should be a separate page with only windows updates. This helped me. Not sure if it's the BE microcodes that stop the throttling or if it was that update that was causing it. 

[KDYZ]

21 hours ago, lockzye said:

I know that it's an old thread by now but I had the same problem as you guys, effective clock throttling. I tried everything as you guys did with no happy ending and settled in for throttle stop. Recently I think I found the solution, at least for me. I've been using for most of my overclocks a modded version of bios with BE microcodes (which are apparently are the fastest for my 9900k) it was showing that I was using DE microcodes in windows though. I've read that I had to actually uninstall a windows update to get the older microcodes to work In my case it was this update KB4589212, after uninstalling it I haven't seen any throttling on effective clock so far. Try uninstalling the update KB4589212 by going to control panel - uninstall apps and there should be a separate page with only windows updates. This helped me. Not sure if it's the BE microcodes that stop the throttling or if it was that update that was causing it. 

Thanks for the info, it definitely seems to be a CPU bug of some kind. In my case I don't think it can be caused by KB4589212, because my system shows that as having been installed on April 20th, but the throttling problem goes back further than that. 

[lockzye]

20 hours ago, KDYZ said:

Thanks for the info, it definitely seems to be a CPU bug of some kind. In my case I don't think it can be caused by KB4589212, because my system shows that as having been installed on April 20th, but the throttling problem goes back further than that. 

Did you try uninstalling that update? I didn't change anything anywhere except uninstall that and the problem is gone. That update forces some changes to microcodes https://support.microsoft.com/en-us/topic/kb4589212-intel-microcode-updates-for-windows-10-version-2004-and-20h2-and-windows-server-version-2004-and-20h2-9f8641b3-bd8e-6fe7-ab4d-4603847c4030. On this list theres several updates with microcode in mind, check if you have any of them. I only had the one I mentioned.

[KDYZ]

1 hour ago, lockzye said:

Did you try uninstalling that update? I didn't change anything anywhere except uninstall that and the problem is gone. That update forces some changes to microcodes https://support.microsoft.com/en-us/topic/kb4589212-intel-microcode-updates-for-windows-10-version-2004-and-20h2-and-windows-server-version-2004-and-20h2-9f8641b3-bd8e-6fe7-ab4d-4603847c4030. On this list theres several updates with microcode in mind, check if you have any of them. I only had the one I mentioned.

I'm a little hesitant to do that, because like I said the update seems to have been installed well after the throttling problem had already started. It also doesn't seem to contain any update for my processor specifically (10700KF desktop). In my case the current microcode is shown as E2 by the Intel identification tool, which is not on the list. I think your experience suggests that it is a microcode related bug, but I think I just need an update, rather than needing to uninstall that particular one.

[Imannudein]

After the latest windows preview update and update for the BIOS at 2103, the issue for me has completely gone. I hope this will stick. 

[KDYZ]

36 minutes ago, Imannudein said:

After the latest windows preview update and update for the BIOS at 2103, the issue for me has completely gone. I hope this will stick. 

Yeah you know what, I have been trying for a few days to reproduce the issue, but so far it's working normally. I wasn't really paying enough attention to know what might have fixed it, but I know it wasn't a BIOS update in my case. I also don't really know if it is fixed, because it comes and goes. I'll post back if it seems to have gone permanently.

[Imannudein]

On 5/4/2021 at 11:12 PM, KDYZ said:

Yeah you know what, I have been trying for a few days to reproduce the issue, but so far it's working normally. I wasn't really paying enough attention to know what might have fixed it, but I know it wasn't a BIOS update in my case. I also don't really know if it is fixed, because it comes and goes. I'll post back if it seems to have gone permanently.

It's happening again! Effective clock was all over the place. So back to TS being run at login. 

[KDYZ]

7 hours ago, Imannudein said:

It's happening again! Effective clock was all over the place. So back to TS being run at login. 

That is the strangest thing...I literally have not seen this for a week or two now, but when I saw your post I went to test again, and it is throttling. I even tested it earlier today, and it wasn't! But now sure enough, it is back.

[Mister Woof]

Haha I left it on the whole time

[Imannudein]

8 hours ago, KDYZ said:

That is the strangest thing...I literally have not seen this for a week or two now, but when I saw your post I went to test again, and it is throttling. I even tested it earlier today, and it wasn't! But now sure enough, it is back.

Haha. I feel you man. From now on, TS auto start all the way. 

[lockzye]

On 5/6/2021 at 5:05 PM, Imannudein said:

It's happening again! Effective clock was all over the place. So back to TS being run at login. 

Yup, did you download any windows updates or anything? It was fine for me for a week and now I noticed that throttling again. It has to be software side then? I can't think of anything that would do that on the hardware side.

 

17 hours ago, KDYZ said:

That is the strangest thing...I literally have not seen this for a week or two now, but when I saw your post I went to test again, and it is throttling. I even tested it earlier today, and it wasn't! But now sure enough, it is back.

 

I also noticed it yesterday, 2 days ago it was fine as I remember stress testing for a few hours and it was rock steady on the effective clock.

[Imannudein]

22 hours ago, lockzye said:

Yup, did you download any windows updates or anything? It was fine for me for a week and now I noticed that throttling again. It has to be software side then? I can't think of anything that would do that on the hardware side.

 

 

I also noticed it yesterday, 2 days ago it was fine as I remember stress testing for a few hours and it was rock steady on the effective clock.

Just normal windows update i guess. I can't pinpoint it to anything particular. 

[trowbot]

I am writing to say I have the same problem with an Intel 9700K on a Gigabyte Z390 Aorus Master. As others have mentioned, simply opening ThrottleStop seems to magically "fix" the issue until next boot. Out of curiosity, I tried the ThrottleStop 9.3.1 beta and found that it no longer has the same magic "fixing" power that the 9.3 Stable version had. @unclewebb, based on the 9.3.1 beta changelog, would you be able to speculate or make an educated guess as to what ThrottleStop 9.3 is doing to magically fix this versus the 9.3.1 beta, which seems to have no effect at all?

 

Thanks

[unclewebb]

@trowbot - I was working with KDYZ for quite a while trying to figure out exactly why this type of throttling happens but we ran into a brick wall. Without access to hardware that has this issue, it has so far been impossible for me to troubleshoot what is going on. 

 

The difference between TS 9.3.1 and TS 9.3 is that the new beta version does not apply all of its settings automatically like TS 9.3 does. My thought was that if some of the automatic features were turned off, a user with this problem could use TS 9.3.1 and slowly adjust it to try to find out what feature disables the throttling. 

 

In theory, TS 9.3.1 and TS 9.3 should work exactly the same. KDYZ tried using one of the beta versions just before TS 9.3.1 and he could not find a way to solve the throttling problem no matter what he clicked on. That makes no logical sense to me since all versions are accessing the CPU in the same way.

 

Maybe try copying TS 9.3.1 into your TS 9.3 folder and the next time you get throttling, try running the newer version.

 

You might have to continue using TS 9.3 forever.

[trowbot]

Yeah, I couldn't get TS 9.3.1 to work either, no matter what I clicked on. I tried copying TS 9.3.1 into my TS 9.3 folder and that didn't help either.

 

That makes me suspect that the note about changing the system timers used is where the magic may have been happening. I suppose its plausible that probing one of those timers gave things a swift kick or cleared out some kind of flag that caused the throttling, since the throttling happens roughly 10 minutes after booting.

 

Using TS 9.3 isn't a big deal, I'm just grasping at straws at this point trying to understand why this is happening in the first place.

 

I'm not sure if this is related, but I'm seeing the events below in the event viewer. I'm relatively certain that the Maximum performance percentage should cap out at 100, but these events are showing absurd numbers like 708%.

 

Kernel-Processor-Power (Microsoft-Windows-Kernel-Processor-Power)    Event Id: 55    Task Category: 47

 

Processor 7 in group 0 exposes the following power management capabilities:

Idle state type: ACPI Idle (C) States (3 state(s))

Performance state type: ACPI Collaborative Processor Performance Control
Nominal Frequency (MHz): 3600
Maximum performance percentage: 708
Minimum performance percentage: 38
Minimum throttle percentage: 2