Simple Speed Tests
A quick comparison of storage device performance using the hdparm utility to measure both cached reads and buffered disk reads across different drive types and models.
OCZ 30GB SSD
$ hdparm -Tt /dev/sdb
/dev/sdb:
Timing cached reads: 3358 MB in 2.00 seconds = 1679.46 MB/sec
Timing buffered disk reads: 452 MB in 3.01 seconds = 150.05 MB/sec
Western Digital WD10EADS 1TB
$ hdparm -Tt /dev/sdb
/dev/sdb:
Timing cached reads: 3372 MB in 2.00 seconds = 1685.85 MB/sec
Timing buffered disk reads: 274 MB in 3.00 seconds = 91.23 MB/sec
Hitachi HDT721010SLA360 1TB
$ hdparm -Tt /dev/sdb
/dev/sdb:
Timing cached reads: 3360 MB in 2.00 seconds = 1680.70 MB/sec
Timing buffered disk reads: 324 MB in 3.01 seconds = 107.63 MB/sec
Hitachi Deskstar HDP725032GLA360 320GB
$ hdparm -Tt /dev/sdb
/dev/sdb:
Timing cached reads: 3304 MB in 2.00 seconds = 1652.26 MB/sec
Timing buffered disk reads: 252 MB in 3.01 seconds = 83.62 MB/sec
Performance Analysis
The SSD clearly outperforms the traditional hard drives in buffered disk reads (150.05 MB/sec vs 83-107 MB/sec), while cached read performance remains relatively consistent across all devices due to RAM caching.
The Cutler.sg Newsletter
Weekly notes on AI, engineering leadership, and building in Singapore. No fluff.
Open MPI Master & Servant Example - BogoMips
Extend MPI master-servant pattern to pass numerical data between nodes using BogoMips CPU calibration across distributed cluster processors
MacBook Pro 17" 2011 - Intel SSD woes solved (kinda)
Intel S510 SSD causes sleep/wake crashes on MacBook Pro 17", but OCZ Vertex 3 provides a stable SATA III solution
Supermicro AOC-SASLP-MV8: DRIVER_TIMEOUT
Troubleshooting DRIVER_TIMEOUT errors with Supermicro AOC-SASLP-MV8 storage controller on modern Linux kernels and finding alternatives