IOPerf results

Results were obtained using a 8 kByte buffer, 5000 IO 1 pass. The test order in that table is : Table indicate 110% for fread on rplay21 (cannot be right and is due to a miss-calculation of clock versus CPU time. Same usual note i.e. under Linux, the manpages indicates that clock() returns an approximation of processor time used by the program and this issue was not adressed since identified here). Some observation however

Block IO Character Random
fwrite fread putc getc seek
Machine/FSKB KB/sec RealKB/sec CPU%CPU KB/sec RealKB/sec CPU%CPU KB/sec RealKB/sec CPU%CPU KB/sec RealKB/sec CPU%CPU KB/sec RealKB/sec CPU%CPU
_rplay21r_40000 9216.59148148.156.22 400000.00363636.36110.00 6768.1922222.2230.46 7168.4623255.8130.82 1775.571953.1290.91
_rplay22r_40000 9478.67153846.156.16 363636.36400000.0090.91 6980.8022471.9131.06 24096.3923952.10100.60 1953.121953.12100.00
NPV_rplay22r_40000 931.81111338.470.88 289890.69319340.6689.96 1487.1120869.667.12 16081.0521437.8674.85 1566.201791.3588.71
NPV_rplay21r_40000 967.74102636.660.99 207801.16296034.6968.37 1557.7021219.147.36 15566.3221874.9171.45 1546.541799.2485.59

IOzone profiling

The profiling gives a gross performance view for several kByte buffer and several file size.
The Read/Write tests are unbuffered while FRead/FWrite tests are buffered.