Dell 8940, 11900k Intel Performance Maximizer not working

@debilone I could be wrong but I think you need more than a CPU and a chipset that supports overclocking, you need most importantly a motherboard and a BIOS that supports overclocking. I think that the XPS 8940, like other XPS systems, while it supports K processors, lacks those features. Here are the release notes for Intel Performance Maximizer but I don't think it is helpful.

Turbo Boost only kicks in when NOT under heavy workloads with 3 cores Idle.

This is not an on demand feature you can force with software its marketing nonsense.

XMP is not a feature available at all with your system.

@debilone by default there is a time limit for Intel Turbo Boost. On my 10900 it was 28 seconds, after which the motherboard brings the CPU back down to its TDP (65W on my non-K CPU, 125W on your K CPU).

The good news is that it's rather easy to remove that limit. I tried the Intel tools too unsuccessfully, but then I used ThrottleStop and managed to remove both the time and the power limit, so my 10900 was able to maintain an indefinite ~200W boost in stress tests.

Here are my ThrottleStop settings. The first step was to go into the FIVR window and check the "Disable and lock Turbo power limits" checkbox, then reboot. After that, I went into the TPL window and set both PL1 and PL2 values to 250, as well as maxed out both time limit sliders.

One thing I'm not sure you'll be able to enable on your 11900K is the new "Adaptive Boost Technology" feature. AFAIK, that can only be enabled in BIOS and from what I've heard the Dell BIOS don't have that yet. The ABT feature basically brings your all-core boost to the same high level as the "preferred core" boost if the temps and power allow it. ThrottleStop doesn't (yet?) have a way to set that.

Thank you for this, exactly what i was looking for

PC shutsdown while running blender or cinebench r20. Not a thermal issue, all cores stays below 85c.

Happends 15 - 20 sec into the benchmark.

seems like the VRM on the dell motherboard is insufficient for the 11900k   Core temp reads 235w + during workload.

Could also be the PSU, but since there is no load on the GPU i doubt it. 

Dell should have updated the motherboards for rocketlake, not just bios updated the older z490 and called it a day..

@debilone try to limit the turbo power in ThrottleStop to 220W, on both PL1 and PL2. Or even start at 200W and raise it slowly in 5W increments to find the max that it can take. 

I don't know if CoreTemp has a metric for the VRM temp, HWiNFO 64 does. That is another factor to keep an eye on. VRMs usually thermal shut down close to 120 degrees.

I assume your XPS has the 500W PSU?

@debilone  Does this system have the optional VRM heat sink?

Hi again, thanks for the help so far!

HWiNFO 64 worked great to check vram temps, and it seems fine. It diden't go past 74c. 

I have 500w PSU and heatsink on VRAM btw. I'll attach a picture of my cooling setup.

It's the turbo power that makes PC shutdown, anything above 205w and it won't complete cinebench r20.

A 11900k should score around 5850 pts in Cinebench r20. I get around 5300 pts with PL set to 200 in ThrottleStop 9.3..

It's not a big issue for me, i mostly play games. But for content creators and others who do CPU heavy workloads the XPS 8940 with 11900k dosen't seem like a good choice. If i had known this pre purchase i would have choosen another PC.

@debilone oh, man... I wonder if Dell reused the same old gen 10 motherboards for these gen 11 CPUs. It was a known fact early on that the 11900k specifically was very power-hungry, with the 11700k not far behind. I would have expected a VRM redesign + bigger PSU to cope with the higher current demand. I've seen youtube reviews where the 11900 went way over 250W with ease. 

That being said, normal VRM behavior should be EDP / Current throttling before abrupt shutdown. If you open the "Limits" window in ThrottleStop while running the benchmark, do you see "EDP" and/or "Current" flash in yellow/red? Normally, the MB should not let the CPU draw current that the VRM can't deliver. It should throttle it down.

Abrupt poweroff makes me think more of either:

• VRM overheating past its limit of 115-120 degrees, which would mean that the sensor showing you 74 in HWiNFO is wildly inaccurate and you could use an actual temp probe to get a more accurate reading directly off the mosfets. Or:
  
  
• PSU getting overwhelmed, which could mean that the GPU may be also kicking in for some reason. Have you monitored GPU power draw while running the benchmark? (either via HWiNFO or GPU-Z).
  
  If you have a Kill-a-watt meter (or know someone who has one), it might be worth using that to see how much your computer draws from the wall during the benchmarks. Our 500W 80+ Platinum PSUs would draw a max of ~550W from the wall to deliver 500W to components (~90% efficiency). If you run your benchmark and kill-a-watt shows your wall power draw close to that number right before it shuts down, then it's the PSU. If it shuts down with the power draw being significantly below that, then it's likely the VRM. Either way, Dell should have put these CPUs only in machines that have the VRM and PSU to actually run it properly.