HP Z2 mini G9 - upgrade CPU to Raptor Lake, getting PCA not fully compatible warning

[RolloZ170]

In my use case (small number of CPU-bound tasks), the i5-14600 is 10-12% slower than the i5-13600. I believe the reason is due to the smaller caches in the 14600, which I didn't realize when I bought it.

can't find any confirmation of your theory.
13600
L2-Cache 11.5MiB (6x 1.25MiB + 2x 2MiB)
L3-Cache 24MiB
14600
L2-Cache 20MiB (6x 2MiB + 2x 4MiB)
L3-Cache 24MiB

 

[Wulfdao]

can't find any confirmation of your theory.
13600
L2-Cache 11.5MiB (6x 1.25MiB + 2x 2MiB)
L3-Cache 24MiB
14600
L2-Cache 20MiB (6x 2MiB + 2x 4MiB)
L3-Cache 24MiB

The cpubenchmark.net site shows L1 cache as 560KB for the 14600, and 1248KB for the 13600.
However, this site shows L2 cache as 2.0MB for the 14600, which looks like a typo (should be 20.0MB).

Whether there is a typo in their listed L1 sizes is unknown, but these numbers, as shown, could
reasonably explain my observation of the 14600 being measurably slower for my particular use case.

Both computers run Linux, and reports from lscpu, dmesg, dmidecode, inxi, hwinfo, nproc, and /proc/cpuinfo
did not show any obvious differences. Both BIOS's show turbo is not disabled.

 

[WhiteNoise]

In theory, the 14600 should be a bit better, since it's true raptor lake and not alder lake rebadge.

Maybe you got shitty silicon lottery.

 

[Wulfdao]

Maybe you got shitty silicon lottery.

This has crossed my mind, of course.
I don't buy CPUs regularly, so I don't have a feel for how bad a defective chip can be.
The multi-core architecture would seem to provide many ways to hide defects.

 

[RolloZ170]

Both computers run Linux, and reports from lscpu, dmesg, dmidecode, inxi, hwinfo, nproc, and /proc/cpuinfo
did not show any obvious differences. Both BIOS's show turbo is not disabled.

swap processors at check again ?

 

[Wulfdao]

swap processors at check again ?

That's a possibility down the line.
I just bought the 125W heat sink, and my first test will be to see if it makes any difference on the 14600.

 

[RolloZ170]

That's a possibility down the line.
I just bought the 125W heat sink, and my first test will be to see if it makes any difference on the 14600.

ther are important settings in the BIOS, not same as default maybe. e.g. PL1 Time Window,
on Xeon Scalables, this can allow the CPU to draw more Power than TDP for a time,
default is mostly 1 second, but if this is set to 128 seconds, it makes a bif difference in benchmark who finish under 128 seconds.