Question i7-14700k is Current Throttled and ExternalVR Throttled ?

[Chandula]

Hey! I recently built a new PC for video editing. When I opened up Premiere Pro just a simple playback of footage took my temperature to 100C on HWiNFO. I ran the stress test with Intel XTU and it gave me Current Throttled, Power Throttled, and ExternalVR Throttled. Power Throttle went away after increasing the Power from bios but Current Throttle and ExternalVR Throttle remain. I ran the Intel XTU CPU stress test with AVX2 for 5 minutes.

I don't think this has anything to do with the cooling of the PC itself since I have an AIO cooler with a case built for good airflow, I checked all the fans to make sure everything's mounted properly as well. From digging around the web what I THINK would be the issue is with overvolting, but I can't undervault in XTU since Undervault Protection is on, I can't find it anywhere in the BIOS as well to disable it to try if undervaulting would work.

Another issue I think might be causing this is the lack of sufficient VRMs in the motherboard. Mine only has 8 VRMs. I know 14700K is for overclocking but I don't necessarily need it. I want this to last as long as possible so I don't mind not overlocking in return for more durability.

Any help would be APPRECIATED! Thank you so much!

PC Specs:

i7-14700K
ID-cooling frostflow x 240 lite AIO
Antec dp502 flux case
Asus Prime H770-plus D4
RTX 3060
Antec HCG750 gold
Crucial 32 GB 3200 RAM

 

[General_Cool]

Since you‘re sure its seated properly, are you 100% sure your pump is plugged in (and to the correct header too) and working? Dead pumps will cause this issue. Among the list of things that could also be wrong is not applying enough thermal paste or not removing the film that comes on the bottom of the heatsink (not all heatsinks have this). You should not be thermal throttling with a 14700k and an AIO if you are at stock clocks.

Careful with undervolting, just like overvolting it will cause instability if you are not careful.

 

[TerryLaze]

I know 14700K is for overclocking but I don't necessarily need it. I want this to last as long as possible so I don't mind not overlocking in return for more durability.

Any help would be APPRECIATED! Thank you so much!

If you don't need and don't want overclocking then why do you want to remove all the limits?! Increasing the power level to above 253W is already overclocking.
Throttling just means that it reached a limit and won't go above that.

100 degrees is the normal limit, any CPU will try to boost to that level.
Intel CPUs have a thermal trip of 130 degrees so 100 is fine.

 

[thestryker]

That cooler isn't sufficient for a 14700K running at full power even if you have it mounted correctly. The ID-cooling website shows max TDP of 250W and personally I wouldn't trust it with that. If you're budget limited on cooler getting a good dual tower air cooler is a much better choice than an inexpensive AIO.

The motherboard may also be a limiting factor with regards to delivering sufficient power. While VRMs have generally gotten a lot better in recent years that doesn't seem to be particularly robust and may limit peak performance.

Your power supply is plenty good enough so unless it is faulty it isn't to blame.

 

[Chandula]

Since you‘re sure its seated properly, are you 100% sure your pump is plugged in (and to the correct header too) and working? Dead pumps will cause this issue. Among the list of things that could also be wrong is not applying enough thermal paste or not removing the film that comes on the bottom of the heatsink (not all heatsinks have this). You should not be thermal throttling with a 14700k and an AIO if you are at stock clocks.

Careful with undervolting, just like overvolting it will cause instability if you are not careful.

Yes, I took out the AIO and plugged it back in with a little more thermal paste. The issue persists though.

I got a multi-core score of 1437 in Cinebench24, is that normal? While running Cinebench the temps went upto 101 c with a lot of fluctuations. It sometimes goes from 50c to 90c within a second. This is the HWiNFO LOG while running Cinebench.

What do you think?

 

[Chandula]

That cooler isn't sufficient for a 14700K running at full power even if you have it mounted correctly. The ID-cooling website shows max TDP of 250W and personally I wouldn't trust it with that. If you're budget limited on cooler getting a good dual tower air cooler is a much better choice than an inexpensive AIO.

The motherboard may also be a limiting factor with regards to delivering sufficient power. While VRMs have generally gotten a lot better in recent years that doesn't seem to be particularly robust and may limit peak performance.

Your power supply is plenty good enough so unless it is faulty it isn't to blame.

I too suspected the AIO might be the culprit but how do you explain the wild fluctuations in temps while running Cinebench? It sometimes goes from 50c to 90c within a second and back down to 50s and 80s. This is the HWiNFO LOG while running Cinebench Multi-Core which I got a score of 1437, is that a good score considering that it's not overlocked?

 

[Chandula]

If you don't need and don't want overclocking then why do you want to remove all the limits?! Increasing the power level to above 253W is already overclocking.
Throttling just means that it reached a limit and won't go above that.

100 degrees is the normal limit, any CPU will try to boost to that level.
Intel CPUs have a thermal trip of 130 degrees so 100 is fine.

I didn't remove any limits. When I ran XTU for the first time it gave me a 'Power Throttled' error, as a solution for that a lot had said that increasing the power would work, and it did. I didn't go over 253W.

I was under the impression that going to 100c degrees would be an issue, especially durability-wise. I ran cinebench24 and got 1437 for the multi-core score, Isn't that a bit low for an i7-14700K even for stock?

 

[thestryker]

I too suspected the AIO might be the culprit but how do you explain the wild fluctuations in temps while running Cinebench? It sometimes goes from 50c to 90c within a second and back down to 50s and 80s. This is the HWiNFO LOG while running Cinebench Multi-Core which I got a score of 1437, is that a good score considering that it's not overlocked?

CPU is likely throttling as a lot of motherboards have a 100c temperature limit even though that's well below CPU limit. You can see the clockspeed randomly dropping down to base clock. The Cinebench score seems low which would go hand in hand with the CPU throttling. It could also be a power delivery type issue causing it.

 

[Chandula]

CPU is likely throttling as a lot of motherboards have a 100c temperature limit even though that's well below CPU limit. You can see the clockspeed randomly dropping down to base clock. The Cinebench score seems low which would go hand in hand with the CPU throttling. It could also be a power delivery type issue causing it.

With those fluctuations in clockspeed and temps, we can rule out the AIO, even though it's a budget one, it shouldn't fluctuate like that, correct? I tried disabling Turbo Boost from BIOs and that gave no high temps at all, but a multi-core score of 1050.

How do you suggest I should troubleshoot the power delivery type? or any other methods I should try to see if the issue cancels out?

Thank you for the help btw!

 

[thestryker]

With those fluctuations in clockspeed and temps, we can rule out the AIO, even though it's a budget one, it shouldn't fluctuate like that, correct?

No you can't rule it out entirely due to how some motherboards are setup.

How do you suggest I should troubleshoot the power delivery type? or any other methods I should try to see if the issue cancels out?

There really isn't a lot you can do aside from pointing a fan directly at the VRM to see if that makes a difference.

There isn't a motherboard manual for the BIOS on that board but it would be worth looking to see if there are power limits you can manually set in it. If you're able to set a power limit you could try that to see where you run into issues. Depending on what the temperatures look like you may be able to rule the cooler out.

It would be worth checking with Throttlestop to see what it shows in regards to the throttling.

 

[General_Cool]

What do you think?

RMA the chip. There could be an issue with the temperature sensors inside the chip that are causing it to read ridiculous values, or it could have had the heatspreader installed poorly, or not enough thermal compound between the die and heatspreader, etc. Your temperature should not be fluctuating 40 celsius within milliseconds, that tells me the monitoring is incorrect or there is zero thermal protection. The higher end 14000 series Intel chips are known to have crap thermal compound under the heatspreader that ends up being the bottleneck. No matter what AIO you have you're screwed unless you're directly on the die.

You could also take the side panels off and stick your hand on the radiator to see if its getting warm. If its cold or lukewarm at best you know its your chip's sensor, since if its truly overheating from not being cooled enough the radiator would be hot to the touch, since it would indicate water being returned to the CPU block that's not cold enough to dissipate heat.

Do you have another cooler to test with?

 

[helper800]

RMA the chip. There could be an issue with the temperature sensors inside the chip that are causing it to read ridiculous values, or it could have had the heatspreader installed poorly, or not enough thermal compound between the die and heatspreader, etc. Your temperature should not be fluctuating 40 celsius within milliseconds, that tells me the monitoring is incorrect or there is zero thermal protection. The higher end 14000 series Intel chips are known to have crap thermal compound under the heatspreader that ends up being the bottleneck. No matter what AIO you have you're screwed unless you're directly on the die.

You could also take the side panels off and stick your hand on the radiator to see if its getting warm. If its cold or lukewarm at best you know its your chip's sensor, since if its truly overheating from not being cooled enough the radiator would be hot to the touch, since it would indicate water being returned to the CPU block that's not cold enough to dissipate heat.

Do you have another cooler to test with?

Correct me if I am wrong, but the 14, 13, and 12 series CPUs from intel are soldered to the IHS, which would be the best possible way of doing it for heat transference while still keeping an IHS. I disagree that an RMA on the CPU is in order. The likelihood of the sensors on the CPU being bad versus the likelihood of him having a problematic cooler are very slim. When synthetic tests are started it is very common for a CPU to jump 40-50C in temperature as quickly as was previously mentioned.

To the OP, have you double and triple checked that there is not a plastic film on the cold plate of the AIO? if you don't peel it you can get some very whacky temperatures readings. Personally, I would get a known good cooler and retest in cinebench. If your score is considerably higher with the new cooler, then you know the problem was related to the AIO. Do you have a budget on what you could spend on test cooler?

 

[General_Cool]

Correct me if I am wrong, but the 14, 13, and 12 series CPUs from intel are soldered to the IHS, which would be the best possible way of doing it for heat transference while still keeping an IHS.

This is just simply not true, where did you hear this?

The likelihood of the sensors on the CPU being bad versus the likelihood of him having a problematic cooler are very slim.

I agree, but it gets to a certain point where the OP has said multiple times and insists the heatsink is installed correctly with adequate thermal paste and everything is plugged in correctly. Whether this is true or not is not something we can determine from the internet. From the way this thread has been going, it sounds like the last options now are to RMA the CPU and/or the cooler or bring it to a shop.

 

[thestryker]

How do you suggest I should troubleshoot the power delivery type? or any other methods I should try to see if the issue cancels out?

I looked through the log again (yay computer rather than phone viewing) and assuming it's accurate regarding the VR VCC temperatures they're going to >100C which leads me to believe it is indeed power delivery (at the very least cooling) which is the throttling problem. This doesn't change the fact that the cooler isn't sufficient for that CPU, but would indicate this specific issue lies there.

If you end up getting a better quality board you might consider moving to DDR5 as there are enough cores on your CPU to benefit from the extra bandwidth.

 

[floorstar]

I didn't remove any limits. When I ran XTU for the first time it gave me a 'Power Throttled' error, as a solution for that a lot had said that increasing the power would work, and it did. I didn't go over 253W.

I was under the impression that going to 100c degrees would be an issue, especially durability-wise. I ran cinebench24 and got 1437 for the multi-core score, Isn't that a bit low for an i7-14700K even for stock?

Just downloaded cinebench24 and ran it on my 14700k and got 1828 at all stock settings, high temp of 80.
I'm using a 360mm AIO, I think you need better cooling

 

[Chandula]

RMA the chip. There could be an issue with the temperature sensors inside the chip that are causing it to read ridiculous values, or it could have had the heatspreader installed poorly, or not enough thermal compound between the die and heatspreader, etc. Your temperature should not be fluctuating 40 celsius within milliseconds, that tells me the monitoring is incorrect or there is zero thermal protection. The higher end 14000 series Intel chips are known to have crap thermal compound under the heatspreader that ends up being the bottleneck. No matter what AIO you have you're screwed unless you're directly on the die.

You could also take the side panels off and stick your hand on the radiator to see if its getting warm. If its cold or lukewarm at best you know its your chip's sensor, since if its truly overheating from not being cooled enough the radiator would be hot to the touch, since it would indicate water being returned to the CPU block that's not cold enough to dissipate heat.

Do you have another cooler to test with?

The radiator gets hot and the outflowing air is HOT as well when running Cinebench, so I'm guessing RMA is not needed, I certainly hope so. I do not have another cooler to test with at the moment, but I'm going to try and find one.

To the OP, have you double and triple checked that there is not a plastic film on the cold plate of the AIO? if you don't peel it you can get some very whacky temperatures readings. Personally, I would get a known good cooler and retest in cinebench. If your score is considerably higher with the new cooler, then you know the problem was related to the AIO. Do you have a budget on what you could spend on test cooler?

Yes, I unmounted and mounted it back in 3 times, Maybe I'll do it again with some better thermal paste to make sure again. I do not have an extra budget but if that's what's needed, there's no other choice right, I'm trying to find another cooler to test first. Will let you know!

which leads me to believe it is indeed power delivery

This is what I initially believed as well, at least on the Intel website they mention that Current throttling happens when the 'The motherboard is not able to provide high enough current for the given processor.' I will test with a test cooler first and see what happens.

Just downloaded cinebench24 and ran it on my 14700k and got 1828 at all stock settings, high temp of 80.
I'm using a 360mm AIO, I think you need better cooling

What's your motherboard and the AIO model? Maybe it's the cooler after all, I thought a 240mm would be enough.

Thank you everyone for your help!

 

[Chandula]

No you can't rule it out entirely due to how some motherboards are setup.

There really isn't a lot you can do aside from pointing a fan directly at the VRM to see if that makes a difference.

There isn't a motherboard manual for the BIOS on that board but it would be worth looking to see if there are power limits you can manually set in it. If you're able to set a power limit you could try that to see where you run into issues. Depending on what the temperatures look like you may be able to rule the cooler out.

It would be worth checking with Throttlestop to see what it shows in regards to the throttling.

I opened up the case and put on my AC so it directly hit the computer, same issue. I have set the power limits to 253W through XTU which I think is the maximum needed for base clocks. Also, When I turn off the turbo power option through bios, temps are SO GOOD, it doesn't go above 75 when running Cinebench, but the scores are poor, 1050 only.

I have downloaded Throttestop but I don't know how to use it. Would love to have some guide!

 

[thestryker]

What's your motherboard and the AIO model? Maybe it's the cooler after all, I thought a 240mm would be enough.

For a 240mm AIO on that CPU you'd need to buy one a little higher end. The only one I'd look at is the Arctic Liquid Freezer II because they use a thicker radiator and aren't super expensive. Deepcool has a really good midrange line now, but I haven't seen much on the 240mm LS520 SE with regards to maximum power in actual use.

I opened up the case and put on my AC so it directly hit the computer, same issue. I have set the power limits to 253W through XTU which I think is the maximum needed for base clocks. Also, When I turn off the turbo power option through bios, temps are SO GOOD, it doesn't go above 75 when running Cinebench, but the scores are poor, 1050 only.

You need turbo on or else the CPU isn't going to clock up at all and should stay around 125W (this is for base clock, 253W is stock boost) no matter what your power limit is set to.

I have downloaded Throttestop but I don't know how to use it. Would love to have some guide!

Since you're just trying to figure out throttling all you'd need to do is hit the limits button and check it out when you run Cinebench.

 

[Chandula]

Since you're just trying to figure out throttling all you'd need to do is hit the limits button and check it out when you run Cinebench.

It would be worth checking with Throttlestop to see what it shows in regards to the throttling.

I just ran Cinebench with Throttlestop, here's what I got. What do you think?

 

[floorstar]

What's your motherboard and the AIO model? Maybe it's the cooler after all, I thought a 240mm would be enough.

​

Z790 Aorus elite "16 phase vrm", and the AIO is a Lian Li Galahad II Trinity Performance 360

I ran cinebench again and checked vr vcc it maxed at 43c

 

[thestryker]

I just ran Cinebench with Throttlestop, here's what I got. What do you think?

PL1/2 is unconcerning as those are power limits, and EDP under ring isn't particularly concerning, but EDP under core indicates current problem. That points at the motherboard power delivery.

 

[helper800]

This is just simply not true, where did you hear this?

This is my source directly from Intel.

 

[General_Cool]

This is my source directly from Intel.

I misunderstood what you meant. That is completely different from it actually being soldered. STIM is a type of solder that is soft and is made entirely of indium. Normal "hard" solder is made from tin and lead, not indium.

Still, that doesn't change my original statement. That "thermal compound", AKA the STIM, is still not adequate to transfer enough heat to the IHS in the high end i7 and i9 14000 series.

 

[helper800]

I misunderstood what you meant. That is completely different from it actually being soldered. STIM is a type of solder that is soft and is made entirely of indium. Normal "hard" solder is made from tin and lead, not indium.

Still, that doesn't change my original statement. That "thermal compound", AKA the STIM, is still not adequate to transfer enough heat to the IHS in the high end i7 and i9 14000 series.

It is definitely adequate for complete function, I think what you mean is that if it is replaced you can get better performance. Point being it is solder with the sameish thermal transfer of traditional solder.

 

[General_Cool]

It is definitely adequate for complete function

Function, yes. Complete function, far, far from it. As an example, i9-13900Ks hit 100 degrees celsius regularly during stress tests with a 360mm radiator with fans on full blast. The only way to lower those temps is to delid or underclock/undervolt, and that is not an acceptable solution for most consumers. Sure, stress tests aren't exactly identical to actual operating conditions, but its still not acceptable that a chip that costs several hundreds to thousands of dollars is thermal throttling out of the box in any application at all.