Okay so I'm reading this article in the June issue of Videomaker and they're giving a review of one of the video workstations. When they get to the part where they're rating the operation of the drives the SSD (Kingston Hyper X) its rated at a Read/Write speed both over 500. But in the review, its only managed just slightly better than the 375 mark for read and in the lower one teens on the write.
What could account for such a huge discrepancy in ratings like this? And how does one guard against spending the money for a premium drive only to get real world speed of less than that of a RAIDed HDD?
-
I assume the workload is not sequential then? It is not a huge discrepancy but more on how they advertised it.
-
NotEnoughMinerals Notebook Deity
What review testing methods are they using?
-
Performance
SSDs are by far the fastest hard drives available on the market. The Kingston HyperX SSD boasts 555MB/s read and 510MB/s write speeds. When bench-marked, however, it only achieved 375Mb/s read and 113 MB/s write, which is still plenty fast. The biggest advantage to an SSD however is their access time, which is a mere 0.2ms.
The complete review:
http://www.videomaker.com/article/15500/ -
One problem is that they do not mention which benchmark they used. The HyperX uses the infamous SF-2281 controller which means that to have any chance of getting those speeds, you need compressible data as well as sequential reads and writes. On top of that the write speeds advertised are for the larger capacity drives.
Here's a SF-2281 SSD that i benchmarked with incompressible data and it has much better speeds at double the capacity. That is a worse case scenario too. Also, the advertised speeds usually reflect the best possible conditions which rarely happen. Mushkin Chronos Deluxe 240GB by the way (the drive is currently in RMA).
EDIT: On top of that, the type of NAND does have an impact too, asynchronous, synchronous or toggle.
SSD Speed: Posted vs Actual?
Discussion in 'Hardware Components and Aftermarket Upgrades' started by Krane, Jul 1, 2012.