PCPer File Copy Test
PC Perspective File-Copy Test v0.3
Our custom PCPer-FC test does some fairly simple file creation and copy routines in order to test the storage system for speed. The script creates a set of files of varying sizes, times the creation process, then copies the same files to another partition on the same hard drive and times the copy process as well. There are four file patterns that we used to try and find any strong or weak points in the hardware: 10 files @ 1000 MB each, 100 files @ 100 MB each, 500 files @ 10 MB each and 1000 files at 1 MB each.
We named the new unit “WD CB 1TB” for these charts. The top three entries show the same drive under the three different controller / driver combinations.
For starters we notice an immediate trend – the faster interface had negligible impact on bulk file creation and transfer for spinning disks. This was somewhat expected, as the disk itself remains the bottleneck and doubling interface throughput does little to help the matter.
As for how this new drive performs compared to its brothers, it does quite well. File creation was comparable to the 2TB models, even close to the enterprise class RE4. File copy operations fell a bit further behind, likely due to changes to the caching algorithm and possibly slower seeks as compared to the fire breathing 2TB models.
I find it odd that the author
I find it odd that the author chides Marvell for the interface performance: in effect doubling the Windows O.S. for transference.
I am bewildered he then goes on to say that the 2TB Drives are faster, but perhaps, should be equalled by the WD1002FAEX.
If the Marvel controller were saturating on the Motherboard, it probably would similarly saturate for the 2TB Hard Disks. Assuming that the OEM PC and the Windows O.S. have not changed, this particular difference must lie within the testing of the WD1002FAEX.
[While it is alleged that] Marvell’s 88SE9123 or 88SE9128 controllers have not been able to deliver improvements to 3GB/sec as suggested; this is partly and largely due to Marvell’s SE91XX drivers having flaws.
Some software requires
Some software requires caching to be disabled. For example, SQL Server. When the driver ignores convention, you have no choice on having a reliable system.
1.5 and 2 TB have a major and
1.5 and 2 TB have a major and minor arm whereas 1 TB has only 1
1.5 and 2 TB have a major and
1.5 and 2 TB have a major and minor arm whereas 1 TB has only 1
Have had two of these fail –
Have had two of these fail – from completely different systems. Steer clear!!