Hard Drive Benchmarks
 

The main purpose of this section, is to show the benefits, or deficits, of various interfaces and drive types.
By testing each interface, the tables will organize the numbers that can help determine the answers, at a minimum, to the questions:
 Is SATA of benefit in a G4?
 How much of a benefit?
 Is Firewire 400 really better than USB 2.0?

__________________________________________

In the following benchmark tables, the equipment used to test is:

G4 Sawtooth, 100MHz bus
ATA/66 bus drive: Seagate 160GB 7200.7
SATA PCI card: Firmtek Seritek  1V2E+2
SATA 3Gb/s Drives: WD6400AAKS, internal connection
                                 WD6400AAKS, eSATA
External SATA Housing: OWC Mercury Elite Pro Dual Bay
FW400/USB 2.0 Hard Drive: Seagate 500GB 7200.10
FW400/USB 2.0 Housing: OWC Mercury Elite Pro


QuickBench 2.1 Small Block Test: 4KB-32KB Average




QuickBench 2.1 Large Block Test: 64KB-1024KB Average




QuickBench 2.1 Extended Test: 2MB-10MB



From the information in the above tables, we can say:
Native ATA/66 is superior to both FW400 and USB 2.0
SATA and eSATA are far superior to the native ATA/66 bus
FW400 is superior to USB 2.0
eSATA is a superior interface for external drives.
I was surprised to see the write speed of the FW400 interfaced drive to be slower than the USB 2.0 interface. I thought that it may be due to the drive, and repeated the tests with a LaCie D2 drive, which houses a 160GB Maxtor ATA drive. The scores of the LaCie drive were virtual mirrors to the results from the OWC housed drive, in both FW400 and USB 2.0 connections.
However, the read speed of the FW400 interface is up to 1.8x faster than that of USB 2.0. The slow read speed on USB 2.0 would partly explain the increased time required to boot to a USB 2.0 startup drive, and the slow data read processes would be a hinderance in the loading of software, further decreasing system performance.
Beyond this, the nature of how the system interacts with USB devices and the interrupt action that is normal on the USB bus as processes occur in the logic board provide further substantiation for advice to use other than USB for mass storage devices; external hard drives should be connected to firewire or dedicated PCI hard drive controllers.
Regardless of the other architectural reasons for not using USB 2.0 as such, we can say:
FW connected hard drives are preferred over USB 2.0 connected hard drives for purposes of an external boot volume due to better read performance over USB 2.0.
A PCI SATA controller which supports boot to connected drives, makes external eSATA drives the best external performance choice. 
Yes, I know this isn’t anything new or earth shattering, but it is surprising how many people ask about FW vs USB 2.0, and eSATA is a common topic as well.

__________________________________________

Head to Head Shootout:   Western Digital 10K Raptor vs. WD6400AAKS Caviar

In the following table, I tested internally connected WD 10K Raptor and WD6400AAKS Caviar drives to determine which may really be the better boot volume in my test system.

QuickBench 2.1:   Western Digital 10K Raptor vs. WD6400AAKS



In virtually all read tests, the WD6400AAKS handily defeated the WD 10K Raptor. In the >2MB tests, the Raptor just fell apart, unable to keep up with the Caviar at all. The smaller block reads were closer, but large file reads appear to be a problem for the Raptor. The Raptors reading ability appears strongest with 128KB and 256KB blocks.
Write speeds, however, show the Raptor to have the upper hand in the test.
So, maybe a WD6400AAKS as a boot drive, and a WD 10K Raptor as a scratch drive, may be the optimal setup for these drives in this system.
At any rate, overall, the WD6400AAKS is the faster of the two, and at less than half the price for triple the capacity, looks like the ol’ Raptor is not a performance value.
PCI SATA rev 1 and SATA rev 2 drives, on the other hand, are clearly far superior to the native ATA/66 bus and Ultra ATA drives, and provide great performance increases even in a 100MHz machine.
So, let’s answer the questions:
Is SATA of benefit in a 100 MHz G4? Sure it is!
How much of a benefit? Truckloads.... 2-3x increase in read and write performance.
Is Firewire 400 really better than USB 2.0? Do I really have to answer..... OK, of course FW is better than USB 2.0, and not just for the data rates.

But that topic will have to be another post in the Blog......

__________________________________________

UPDATE

I have added some performance testing, pitting the WD Raptor and AAKS drive against an OWC SSD drive.
The results of these tests can be seen on the page:    SSD in a PPC Mac


Blog/F93FC8E5-EB42-4287-BEC2-B50A6BEC8DF1.htmlshapeimage_2_link_0