With AMD posting capacity-aware scheduling patches this weekend, I wanted to see what my own Linux laptops think their cores are. Here’s how to check, read-only, nothing to install.
1. Look at max frequency per core
lscpu -e=CPU,CORE,MAXMHZ,MINMHZ
On a hybrid chip you’ll usually see two or more groups of max clocks. If every core shows the same number, you probably don’t have a hybrid CPU.
2. Check what the scheduler is told
grep . /sys/devices/system/cpu/cpu*/cpu_capacity 2>/dev/null
On Arm laptops (Snapdragon, Apple silicon under Asahi) you’ll typically get a capacity number per CPU, with 1024 for the biggest cores. On many x86 boxes this file doesn’t exist, which is exactly the gap AMD’s patches are about.
3. On AMD, check amd-pstate
cat /sys/devices/system/cpu/amd_pstate/status
grep . /sys/devices/system/cpu/cpu*/cpufreq/amd_pstate_highest_perf
grep . /sys/devices/system/cpu/cpu*/cpufreq/amd_pstate_prefcore_ranking 2>/dev/null
The status should say active, passive or guided. Different highest-perf or ranking values across cores tell you the hardware knows some cores are stronger, even if the scheduler isn’t using capacity yet.
4. On Intel hybrid, list the core types
cat /sys/devices/cpu_core/cpus /sys/devices/cpu_atom/cpus 2>/dev/null
That splits the P-cores and E-cores by CPU number.
5. Watch where work lands
ps -eo pid,psr,comm --sort=-pcpu | head
The psr column is the CPU each busy process last ran on. Compare it with step 1.
All of this is read-only. Don’t add boot parameters like the proposed amd_case=disable until that code actually ships in your kernel.
God Bless, and May the Source Be with You!