IgH 1.6.9 MEASURE_TIMING 1 kHz on Debian 13 RT, N100, I226-V + ec_igc (isolcpus

Hi, first post. This is IgH only (not LinuxCNC), 1 kHz servo/press application.

Hardware / software
- CPU: Intel N100 (4 cores)
- Distro: Debian 13 (Trixie)
- Kernel: 6.12.101+deb13-rt-amd64, PREEMPT_RT
- IgH: 1.6.9, native driver ec_igc
- NIC: Intel I226-V (8086:125c), dedicated EtherCAT port
- Slave idle in PREOP during the test (application stopped)
- Governor: performance (cores locked 700-800 MHz)

CPU isolation (cmdline)
isolcpus=managed_irq,domain,3 nohz_full=3 rcu_nocbs=3 irqaffinity=0-2
intel_idle.max_cstate=0 processor.max_cstate=0 idle=poll

RT thread is pinned to isolated CPU3 with SCHED_FIFO.

Measurement
Same method as examples/dc_user MEASURE_TIMING:
clock_nanosleep(CLOCK_MONOTONIC, 1 kHz) wakeup latency, then only
receive / process / queue / send. No PDO mapping, no CiA402.
SCHED_FIFO 90, taskset -c 3, 60 s / 60000 cycles.

Wakeup latency (µs)
min / p50 / p90 / p95 / p99 / p999 / max / avg
= 6.311 / 6.901 / 10.717 / 10.898 / 11.143 / 11.592 / 29.024 / 7.741

Question: for I226-V + ec_igc on 6.12-rt, is this in the expected
range for 1 ms, or do people still prefer I210/igb for the worst-case
tail?

Thanks.