Short stroking bigger disks for higher IOPS and storage speed was a de-facto method in some HPC centers. Do this to a sufficiently large array and you can see unbelievable IOPS numbers for that generation of hardware.