Latest Linux Code Smashes 14M IOPS Per-Core With Intel Core i9 12900K + Optane

Back in the day, 2.5 million IOPS per core was an impressive feat... That day was little more than one year ago. With faster hardware and relentless optimizations by Linux kernel developers, 14 million IOPS per core is the new record now achieved.

Thanks to the work of Linux block subsystem maintainer Jens Axboe and other upstream kernel developers, over the past year particularly record-after-record benchmarks were being reached when measuring the IOPS per-core potential for further lowering the Linux I/O overhead. The 2~3 million range was impressive a year ago while now it's north of ten million and closing in on fifteen million.

Fast hardware with never-ending Linux performance optimizations mean new records.

Axboe is still using the Intel Core i9 12900K "Alder Lake" with Intel Optane P5800X NVMe storage for his flagship hardware setup. Breaking the 14M IOPS threshold was achieved when he was using the very latest development code, including his "perf-wip" Git branch that carries his various performance optimizations until mainlined along with "a few pending goodies."

14M IOPS out of a single CPU core? Yes please. Usual test case, 12900K box. This is current -git + perf-wip, with a few pending goodies.

Keep pushing forward, next stop 15M/core. When we optimize, we don't to it for benchmarking, but rather to reduce overhead in general. pic.twitter.com/iHIegsZ9fB

— Jens Axboe (@axboe) March 22, 2022

The next milestone is aiming for 15M IOPS per-core. While not everyone has this class hardware and this is just benchmarking off the raw block device, ultimately all of these Linux I/O optimizations allow for reducing overhead and yielding more efficient code paths that come to benefit all Linux users.