II have teh 13900KS CPU with a ROG RYUJIN 360 rad AIO and no matter what I do the CPU gets hot under load up to 0C and it throttles. I am not over clocking.
I’ll play a game like Starship troopers extermination or run BOINC or other heavyloads and it just gets so hot. the back and side of the chassis is very hot to the touch.
I am running the 13900KS on an asus z790 proart creators with the Asus Kyuin 360, 64Gb DDR 5800 RAM, dual samsung 990 Pro NVME and a ASUS TUFF 4090RTX with a seasoniic ATX 3.0 1300W PSU. the case is a fractal 7XL its the one with no side window.
I have no idea what to do to keep the CPU cooler and from throttling.
Thank you for reading this post.
For the new Intel CPUs we’ve increased the temperature limit in BIOS from 100°C to 115°C, to get a better feel for temperatures without thermal throttle getting in the way.
Ensure this is set in the bios to properly ramp the pump up, linked to the CPU temperature. If in doubt set it to 100% and see what temps are like.
Would also ensure case fans are also set in the BIOS to ramp up properly, and are all facing the correct way.
That said it’s perfectly normal for a metal case to get pretty warm to the touch when a PC is under load, especially one with a 13th gen CPU and a 4090 in it - under load you’re looking at a good 600w of heat even without tweaking.
I’ve recently run the 13900k and 14900k and they will hit 100ºC and throttle without much provocation.
Looking at the individual core temperatures can be useful e.g. if under full load with all cores loaded, if some cores are significantly hotter than others it can indicate poor thermal paste application.
I would:
Check the thermal paste application, probably redo it ensuring adequate coverage of the whole CPU.
Use a quality thermal paste.
Ensure the pump cold plate is correctly seated to the CPU.
Ensure the pump cold plate isn’t damaged.
Good advice above to run the pump and fans at full speed to test max cooling. Maybe check the pump and fan curves in both the BIOS and the controller software if installed.
With the k series i find it necessary to undervolt and set PL1/PL2 power levels to control temps. The ks series are even hotter. Try installing Intel XTU to easily experiment with undervolting and power limits.
My understanding of the 13900KS is that you cant keep it from thermal throttling.
Short of TEC/Heatpump to get the block down to sub ambient temperatures the 13900KS simply will generate more heat than can be dissipated.
The only thing I have seen work is undervolting. Going into the bios and backing off your voltage can reduce the amount of power it pulls at full turbo giving your AIO a chance to keep up with the load.
check your bio settings you probably have multi-core enhancement turned on and it’s trying to overclock everything at all core which isn’t that useful.
that AIO is definitely outclassed especially depending on what your ambient heat levels are. Go look at a mora3
The thing about consumer Intel CPU is that it will run as fast as possible (i.e. consuming/dissipating heat as much power as possible) until it hits one of the throttling thresholds. Throttling is not the problem here if you’re getting the performance number you expected.
To get more into details, Intel CPU has multiple limits either fused or configurable via BIOS. Namely, long power limit (PL1=125W), short power limit (PL2=253W), transient peak limit (PL3), overcurrent limit (PL4), thermal limit or TjMax (100c), Fast Throttle or Per-Core Thermal Limit (107c), IccMax (307A). The frequency will scale based on V/F curve for how much voltage required at a certain frequency as regulated by the By Core Turbo Ratio.
The thermal throttling issue here comes from motherboard manufacturers ignoring the base spec, and setting PL1=PL2=4096W by default (and some even set IccMax to unlimited). This means the CPU will boost until it hits one of the other limits. Typically, this is the thermal limit of 100c.
As to why this is happening, despite there’s also a clock limit in place: it is because the By Core Turbo Ratio on a client CPU is very coarse. The CPU cannot really boost to max all cores clock under the spec PL2=253W (on a 13900K, 253W can sustain around 5.1 GHz when the fused all cores boost is 5.5 GHz). So when the PL limit is removed, it will try to boost all cores up to the By Core Ratio clock, and consume more power as a result.
If you do not have a performance problem (indicating that the throttle was expected) and want to just run CPU cooler, set the PL1 and PL2 to Intel’s base spec. In BIOS, there should be a toggle called “Load Intel Base Power Limit”. You’ll likely see 5-10% loss in performance. If Intel Base Power Limit is not your thing, either undervolt or set PL1=PL2 to some reasonable number that is not 4096W (PL1=PL2=253W). (Really, performance becomes a diminishing return after 200W or so, I personally don’t really see any reason to set this to anything above 253W).
Drop the max boost by 100-200mhz and undervolt as much as possible without losing stability or performance. You’ll still be ahead of 99%+ other consumer systems.
one of ou of the top of the line gaming MB (auto OC and insane default setting are almost given)
ASUS => there will be shaeningans galore
XMP memory
least power efficient cpu from its generation ( and ± 3 gens)
Result? You will have trouble cooling this on stock setting and custom water loop, much less if any OC is happening by default. And stock setting is extreme OC
That said, these modern high end cpus will boost as much as possible within their thermal constrains to maximise clocks, to the point of diminishing returns. If it bothers you you can impose limits yourself. But if clocks and benchmark scores look similar to reviews everything is probably installed fine.
(meant in best way possible)
