HDTach 2.61 Summary
We can sum up the HDTach results in a couple of extra graphs. Let's plot CPU utilisation and random access time on the same graph to save space, sharing an axis, so we can talk about those facets of performance.
The graphs tells us a couple of things. Firstly, USB2.0 results in higher (apparent) CPU usage than disks connected to the FireWire interface. I could bleat on about packet overheads and related stuff, but I don't trust HDTach enough to do so. It's definitely higher CPU usage than a native IDE or SATA controller would exhibit though, that much we can confirm.
Secondly, both interfaces do no favours for random access time on any of the disks, doubling their rated figures for the most part. There's always going to be some overhead when you have to bridge the ATA interface and squirt the command set and data over a non-native interface, adding an extra layer to things.
So external drives on non-native interfaces perform poorer than their natively run equivalents. A big shock to you all I'm sure.
With the Freecom on the above graph, you can instantly see it's not the fastest drive in the world. 4200rpm, 2MB, 2.5" notebook drive on the USB2.0 interface means we're a bit tortoise-like. Ignore it, it's not meant to be a Raptor. It's the other three 3.5" disks that surprise.
Look here. Notice the 111MB/sec burst speed of the natively run DiamondMax 9? Notice how it barely squirts one third of that in this case, run over a FireWire connection? The FireWire400 spec gives us 400Mbit/sec maximum bus bandwidth. 50MB/sec in usual terms. With overheads, we're hitting the bus limit here, none of the faster disks able to shine.
Let's see if we can confirm that with ATTO Disk Benchmark and spot our rough 35MB/sec limit again.
