Benchmarking BFQ

Introduction

Various tests are took to benchmark the performance of BFQ. They are run on two different machines:
Machine 1 Machine 2
CPU Intel Atom(tm) N270@1.60GHz Phenom II X4 955
RAM 1.0GB 8.0GB
Harddrive WD1600-BEVT, 5400rpm, 160GB, SATA WD5000AACS on AHCI
Architecture i386 x86_64
The tests includes:

interbench

Tests of interbench are taken on machine 2. Using 1528020 loops per ms.
Average Latency
Max Latency

blogbench

Tests of blogbench are taken on machine 1.
Read Score
Write Score

IOZONE - throughput

Tests of tiobench are taken on machine 1. The invoke command is:

./iozone -s 20m -r 64k -l 32 -u 32

The command means: Create 32 threads and each thread reads/writes 320(=20M/64K) 64KB-size records. The total reading/writing speed (KB/s) is measured.

DragonFly kernel making

Tests of DragonFly kernel making are taken on machine 1. The invoke command is:

time make nativekernel -j 2

The total time spent is measured.