I know... This issue has been beaten to death. I have read article after article about how Firewire is faster in real-world applications.... its performance due mostly to the fact that it is a Peer-to-Peer protocol allowing for intelligent bus negotiation. In contrast, USB2 is a Master-Slave protocol with high administrative overhead for the CPU and with bandwidth shared between in-channel devices. I believe it to be true, however, I just haven't observed it yet. I want to believe... Make me believe.
My issue is this: In the real-world situations I deal with, USB2-attached hard drives _always_ give me better performance than the same model drive attached with Firewire. This has been nagging at me for a long time now, and I would like some insight to help me track down why I am seeing this trend.
Here is the scenario:
I use 300GB Maxtor One Touch III hard drives with USB/Firewire interfaces. I have to repeatedly setup a portable wired network using all laptops. The central laptop (server) is usually (but not always) a Dell Inspiron 5150 - sometimes an Averatec 2370, or other 4-pin Firewire-equiped laptop. If using USB, I have the hard drive connected directly to a USB port on the laptop (not thru a hub). If using Firewire, I use a 6-pin to 4-pin cable to attach the drive to the laptop's 4-pin Firewire port. A managed switch connects the server with 8-10 other computers. There is a network share on the hard drive with thousands of image files. Customers at all 8-10 client stations simultaneously access image files (2-5MB in size) in rapid succession. That's where I see the difference.
With nothing else changing, if I switch between Firewire and USB, there is a noticeably larger lag in loading each picture if the hard drive is connected with Firewire.
- With the USB connection, the images load almost instantly but can take up to 2 seconds maximum with an average of <1 second.
- With the Firewire connection, every image takes from 1.5 - 5 seconds to load with an average of about 3.5 seconds.
Any thoughts on why that is? I'm very curious.
- Are the purported advantages of Firewire over USB restricted to certain situations? For example, where multiple devices are attached to the computer thru the same host port (i.e. via daisy-chain for Firewire, or a port-expander for USB) and the flow of data is from one of these devices to another in the same chain/tree?
- Are the benefits of Firewire partially dependent on the host/device chipset? I've seen some peripherals advertise an "Oxford" Firewire chipset as a selling point; does chipset make that much of a difference?
- Is filesize a determining factor? Do files have to be some minimum size for Firewire to really stretch its legs (i.e. video files vs image files)?
- Hypothetically, would I see any different real-world performance using a Desktop and a PCI add-in card, as opposed to a laptop? Is the laptop southbridge somehow a limiting factor here for the Firewire?
- It shouldn't make any difference, but ... Could the 4-pin/6-pin cabling have anything to do with it (as opposed to 6-pin/6-pin)... I know in theory it shouldn't matter because the extra pins are just for power, but, has anyone ever seen a real-world difference in using 4-pin connections instead of straight 6-pin?
Start Free Trial
