Skip to main content

Workstation Storage: Modeling, CAD, Programming, And Virtualization

MATLAB: Loading Data

Overall Statistics
Elapsed Time33.9 s
Read Operations2127
Write Operations488
Data Read42.37 MB
Data Written4.62 MB
Disk Busy Time0.43 s
Average Data Rate109.29 MB/s

Data analysis is another facet of workstation performance, often performed by those in an engineering- or life science-related occupation. This is a lot more complex than what most of us working in Excel do on a daily basis.

Consequently, we set our sights on more serious data analysis. In this specific trace, we took an 8 MB CSV file from the United States Census Bureau, which contains statistics on over 80 000 cities and towns, and loaded the data set into MATLAB. Overall, importing data appears to be a very lightweight task, since the vast majority of operations occur at a queue depth of one, and most of the data transferred is random in nature. Finally, two-thirds of all operations are 4 KB in size.

I/O Trends:

  • 75% of all operations occur at a queue depth of one
  • 82% of all data transferred is random
  • 66% of all operations are 4 KB in transfer size
  • 10% of all operations are 8 KB in transfer size
  • 9% of all operations are smaller than 512 KB in transfer size