• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

Should I be Worried? I9 14900K

Associate
Joined
22 Jan 2024
Posts
5
Location
England
Just built a new PC and am worrying about the performance of my I9 14900k. Should I be worried about this?

During the 3D Mark CPU Profile Test, the Max Thread part ramps up my CPU temperature to a maximum of 88 degrees. This seems very good, however, I've found that just the sheer act of launching the 3D Mark application causes "P-core 2" spikes to 98 degrees very briefly (for 1 reading) and shows up "yes" that it's throttling. Occasionally some of the other P-cores throttle too during this process. I am using HWinfo for monitoring.

I am just confused, is this normal? Why does a core throttle when loading up 3D mark, but yet when I'm running the actual tests it only reaches 88 degrees maximum.

More info:
- Idle temps are 28 degrees.
- CPU is overclocked by: Multicore enhancement (Auto - let bios optimize). And Intel adaptive boost technology enabled. XMP also enabled.

(POSTED MORE INFO AND IMAGES IN MY COMMENT BELOW)
 
Last edited:
Man of Honour
Joined
12 Jul 2005
Posts
20,537
Location
Aberlour, NE Scotland
The 14900k is a ridiculously power hungry and hot running cpu and you need a really good cooler (preferably a 360mm AIO) to manage them. Many people undervolt them to manage power draw and temps and you should find out how to do this by doing a search of the forums or googling the net.
 
Soldato
Joined
25 Sep 2009
Posts
9,633
Location
Billericay, UK
Interesting post, I wonder why 3Dmark is loading itself using P-Core's and why the thread scheduler in the hardware and Windows isn't directing it through the E-cores (given it's only unpacking the software to memory)?

Anyway, from what your describing loading 3DMark is only using one CPU core as result the CPU will assign power where it's needed and if there's spare power because the rest of the CPU is idle then it will boost the core clock frequency that's are being used. In this case the CPU has boosted for as long as it can and has hit it's thermal limit and downclocks itself so it doesn't overheat, this is perfectly normal. When the actual benchmark is being run more CPU resources are being used so the power budget is more evenly distributed across the silicon, frequencies won't be as high as they were when the CPU is running more single threaded tasks hence why under a heavy load the CPU runs a bit cooler.

The 14900K is a monster when it comes to power consumption, you need a substantial cooler to keep it from thermal throttling but I guess you already known when you did you research before buying. The crappy LGA1700 retention arm doesn't help either (uneven contact between the heatsink and IHS).
 
Last edited:
Associate
OP
Joined
22 Jan 2024
Posts
5
Location
England
what cooler are you using??

I am using a Corsair 360mm AIO Capillix XT.

The 14900k is a ridiculously power hungry and hot running cpu and you need a really good cooler (preferably a 360mm AIO) to manage them. Many people undervolt them to manage power draw and temps and you should find out how to do this by doing a search of the forums or googling the net.

I will look into undervolting the chip too, thanks for the suggestion.

Interesting post, I wonder why 3Dmark is loading itself using P-Core's and why the thread scheduler in the hardware and Windows isn't directing it through the E-cores (given it's only unpacking the software to memory)?

Anyway, from what your describing loading 3DMark is only using one CPU core as result the CPU will assign power where it's needed and if there's spare power because the rest of the CPU is idle then it will boost the core clock frequency that's are being used. In this case the CPU has boosted for as long as it can and has hit it's thermal limit and downclocks itself so it doesn't overheat, this is perfectly normal. When the actual benchmark is being run more CPU resources are being used so the power budget is more evenly distributed across the silicon, frequencies won't be as high as they were when the CPU is running more single threaded tasks hence why under a heavy load the CPU runs a bit cooler.

The 14900K is a monster when it comes to power consumption, you need a substantial cooler to keep it from thermal throttling but I guess you already known when you did you research before buying. The crappy LGA1700 retention arm doesn't help either (uneven contact between the heatsink and IHS).

Not very technically inclined when it comes to overclocking, but here are some images of HWinfo when just loading it up from the shortcut on my desktop. It seems to be using all the P-cores on the first attempt at loading it, and then the top 4 P-cores on the second attempt at loading it.


e44c3uyb1zdc1.png


uu4cp2490zdc1.png



the spike is when system info is scanning your system, maybe a short high load to boost everything to get a reading for the Info web page once the test is complete.

Yea. It seems to be spiking my temperatures during this part of the loading screen (image below). I was thinking the same because it is as if the spike only last for one sensor reading and then back down it goes.

ckadypu42zdc1.png
 
Last edited:
Soldato
Joined
15 May 2012
Posts
5,812
Location
Louth, lincs
Not sure what board you have(if ASUS) but you can set Loadline calibration to level 4, then IA DC LL to 1.02 then IA AC LL to 0.2 and finally core/global volage to adaptive. This should lower voltage requirements, run some tests, if you crash increase AC LL only to say 0.22 and re test, or lower if you can 0.18??
this will get your temps down assuming you have the cooler mounted correctly
 
Last edited:
Associate
OP
Joined
22 Jan 2024
Posts
5
Location
England
Also, I have another question, how is it that any throttling is occuring? I thought the trigger was 99 or 100 degrees? In the second attempt (image above), P-Core 2 hit a maximum temperature of 89 degrees and it still somehow managed to throttle?

@Nickolp1974 @Daytrader I just disabled Multicore Enhancement, but kept Adaptive Boost Technology on. The temperature results were very similar to the 2nd attempt (image above)
 
Soldato
Joined
15 May 2012
Posts
5,812
Location
Louth, lincs
Also, I have another question, how is it that any throttling is occuring? I thought the trigger was 99 or 100 degrees? In the second attempt (image above), P-Core 2 hit a maximum temperature of 89 degrees and it still somehow managed to throttle?

@Nickolp1974 @Daytrader I just disabled Multicore Enhancement, but kept Adaptive Boost Technology on. The temperature results were very similar to the 2nd attempt (image above)
the throttling trigger can occur because your cpu is trying to boost, it cant because of the temps it should still run at its base clocks though.
 
Soldato
Joined
3 Feb 2008
Posts
5,483
Also, I have another question, how is it that any throttling is occuring? I thought the trigger was 99 or 100 degrees? In the second attempt (image above), P-Core 2 hit a maximum temperature of 89 degrees and it still somehow managed to throttle?

@Nickolp1974 @Daytrader I just disabled Multicore Enhancement, but kept Adaptive Boost Technology on. The temperature results were very similar to the 2nd attempt (image above)

I just tried to load 3dmark out of curiosity (I have enforce limits enabled + Adaptive boost on) and the package power only peaked at 178w, so enforcing limits likely isn't doing anything in this scenario as the limit is 253w. My hottest P Core hit 79C during the system scan, average probably 73ish, idling around 27C at the moment.

Check your CPU package power in HWINFO, and if it's similar, I'd be looking at the cooler mount and maybe a contact frame if you're comfortable installing one (It's dead easy). First though get hold of Cinebench R23 where you can more reliably test temperatures under consistent full load. If you're still seeing mid to high 90s with enforce limits on then my money would definitely be on a cooler/mount issue.
 
Associate
OP
Joined
22 Jan 2024
Posts
5
Location
England
I just tried to load 3dmark out of curiosity (I have enforce limits enabled + Adaptive boost on) and the package power only peaked at 178w, so enforcing limits likely isn't doing anything in this scenario as the limit is 253w. My hottest P Core hit 79C during the system scan, average probably 73ish, idling around 27C at the moment.

Check your CPU package power in HWINFO, and if it's similar, I'd be looking at the cooler mount and maybe a contact frame if you're comfortable installing one (It's dead easy). First though get hold of Cinebench R23 where you can more reliably test temperatures under consistent full load. If you're still seeing mid to high 90s with enforce limits on then my money would definitely be on a cooler/mount issue.
Really appreciate you spending the time to try it out. CPU package hit 180W. I could have swore I put the Corsair AIO on right, but it must be the issue. Thanks to everyone else who was helping too!
 
Soldato
Joined
14 Aug 2018
Posts
3,393
Really appreciate you spending the time to try it out. CPU package hit 180W. I could have swore I put the Corsair AIO on right, but it must be the issue. Thanks to everyone else who was helping too!
For less than £10 including postage you should really use a Thermalright contact frame to help keep temps down and prevent throttling, especially as you have the 14900k which should almost come with them in the box. I suspect the difference for a 14900k the temperature drop is up to near to ~10c

It's really simple to do and takes me less than 3mins to remove the old mechanism and install the contact frame.

 
Back
Top Bottom