Intel i9-11900K is supposed to be the top one in the 11th Gen CPU family. Recently I bought a desktop computer with this CPU, but its performance is not satisfactory. Shortly after that, I bought a laptop with an i7-11800H CPU, and surprisingly found that it’s actually faster than that desktop on matrix multiply with Julia. Then I used cpu-z to benchmark these two CPUs, getting the following results:
which showed that the score of my i9-11900K was much lower than a reference i9-11900K provided by cpu-z. So, what’s the problem? Have I bought a fake i9-11900K?
But unfortunately the situation was even worse (580 single thread and 4433 multi thread). I also tried to change the active power plan from balance to high performance in XTU’s settings, but couldn’t make things better. Have I missed something important?
When benchmarking it using the XTU, the highest temperature is 72 degree, does it mean cooling is crucial for my case? As shown in the attached picture, my system has a pretty low score compared to a typical i9-11900K score on the web. Really weird. BTW, my desktop is a newly bought Dell Vostro 3890, whose cooling system should be of no problem.
it looks like it has some stats on “thermal throttling” (lower right) check to see if it was doing any thermal throttling (I’m not familiar with any of this as I don’t run Windows).
This claims the vostro 3890 comes with a B560 chipset, which tend to be the “budget” motherboards if I understand correctly (vs the Z590).
I’d follow the advice from the post above on Current and Power Limit Throttling, but there is a chance that this motherboard is relatively limited in power delivery.
One thing nobody has asked yet is what your RAM configuration is. What clock speeds does it support, perhaps you might even have to change the clock in the BIOS to take full advantage of your hardware? Something that can be quite important is that you have at least two matching memory sticks and you need to ensure that you followed the motherboard instructions on what slots they should be placed in.
I had never considered that - am currently planning to replace one of the 16GB sticks in my new Lenovo P14 with a 32GB one, are you saying I might reduce overall system performance by doing this?
I believe it should work, but all sticks will probably have to run at the most conservative speeds.
Thus, there’s no sense buying a faster 32GB stick than the 16GB sticks you already have (unless you’re replacing all the 16 GB sticks), while if you buy a slower 32GB stick, this will force you 16GB sticks to run slower to match.
Similar with voltage, you’ll probably need to run the maximum voltage any of them need.
Great, thanks - in this instance Lenovo has helpfully decided to solder on one of the two sticks, so it looks im basically limited to buying replacement RAM with the same specs
I think you are right. When benchmarking it in XTU, the Core TDP seems to be limited at 58W and the Power Limit Throttling is always yes during the benchmarking. So I think 58W is the limit of that motherboard.
Thanks, but what to talk about to the Dell support? Ask it to change a motherboard for that machine? If it is not willing to do that, is it easy for me to change a motherboard by myself?
