Applications run >25% slower under kernel 5.15.0-raspi than 5.4.0-raspi

I am one of the maintainers of an open source project and noticed that the build times for our .debs are ~50% longer for Jammy than they are for Focal or Bionic. I have three identical Raspberry Pi model 4B (rev 1.4) build hosts. All are using an official Raspberry Pi power supply, have the same brand/model SD card, and large heat sink w/ fan that keeps the processor temperature below 42degC. I have performed various benchmarks in an attempt to determine the source of the bottle neck. In particular, I tried to rule out the SD card as the source by creating a tmpfs and running iozone in it to measure the R/W speeds of the memory. What I found was that iozone reported nearly identical values for focal and bionic by significantly lower values (>25%) for jammy. I tried to rebuild the focal kernel .deb from src on jammy but could not get that to work so eventually I just tried copying the relevant files from /boot and /lib/modules from the focal build host to the jammy build host. After booting the jammy build host using the focal kernel, iozone reports comparable I/O times tmpfs as the focal & bionic build hosts, leading me to believe that the 5.15.0-raspi kernel is being configured with some difference that causes a substantial performance degradation. I have tried diffing the two configs but nothing immediately jumped out at me.

Can anyone else confirm this behavior before I file a bug?