Intel’s NASPT (NAS Performance Toolkit ) is a benchmark tool designed to enable direct measurement of home network attached storage (NAS) performance. NASPT uses a set of real world workload traces (high definition video playback and recording, video rendering/content creation and office productivity) gathered from typical digital home applications to emulate the behavior of an actual application.
We’ve used some of the video and office apps results to highlight a NAS device’s performance.
HD Video Playback
This trace represents the playback of a 1.3GB HD video file at 720p using Windows Media Player. The files are accessed sequentially with 256kB user level reads.
4x HD Playback
This trace is built from four copies of the Video Playback test with around 11% sequential accesses.
HD Video Record
Trace writes a 720p MPEG-2 video file to the NAS. The single 1.6GB file is written sequentially using 256kB accesses.
HD Playback and Record
Tests the NAS with simultaneous reads and writes of a 1GB HD Video file in the 720p format.
This trace simulates the creation of a video file using both video and photo editing software using a mix of file types and sizes. 90% of the operations are writes to the NAS with around 40% of these being sequential.
A trace of typical workday operations. 2.8GB of data made up of 600 files of varying lengths is divided equally between read and writes. 80% of the accesses are sequential.
This simulates the opening and viewing of 169 photos (approx 1.2GB). It tests how the NAS deals with a multitude of small files.
The W5810 doesn’t break the 100MB/s mark for any of the NASPT video tests although it does it very close in a couple of the 4x HD Playback tests. While the results from the HD Video Playback, 4x HD Playback and HD Playback and Record are reasonable enough, the HD Video Record results are disappointing.
In the NASPT Office tests, the W5810 shows a good degree of consistency between the various disk setups, averaging 38MB/s for the Office Productivity tests and 26MB/s for the Photo Album tests. However, when it came to dealing with the mix of file types in the Content Creation test, the results were very disappointing.