This page displays the latest benchmark numbers for seL4 from the publicly available sel4bench repository.

BenchmarkSabre I.mx6 (Cortex A9 @1GHz)Jetson TX1 AARCH64 (Cortex A57 @1.9GHz)ia32 SkyLake @3.4GHzx64 SkyLake @3.4GHzx64 No spectre/meltdown mitigations SkyLake @3.4GHz
IRQ path cycle count6738478641396943
IPC microbenchmark, client->server301374398687468
IPC microbenchmark, server->client324378384704484

GitHub sel4bench-manifest72feeea70bfd7b98398ad1fbe73d7d56da9425db
IRQ path cycle count
Time in cycles to deliver an interrupt to user space, where a context switch is not needed.
IPC benchmark, client->server
Time in cycles for half an IPC between address spaces.
IPC Benchmark, server->client
Time in cycles for the IPC reply between address spaces.