The math of data transfer: USB 2.0 and FW…
apple Verified ·
Imported from the Apple Support Forums via the Wayback Machine 🔗
I'm confused about data transfer speeds. I've got a USB 2.0 hard drive that is supposed to have a data transfer rate of "up to" 480 MBps, and a FW800 drive that is supposed to have a transfer rate of "up to" 800 MBps. However, I tested reading and writing an 8.5 GB folder with 5,700 files in it. The ACTUAL transfer time was 8:52 for the USB (works out to 16 MBps) and 4:02 for the FW800 drive (works out to 35 MBps) - both substantially below the 'advertised' rate.
So I looked at some Read tests for hard drives, irrespective of interface. [
http://www.barefeats.com/hard56.html ] Apparently a 5600 rpm drive from Seagate or Hitachi only has a read speed of 36-37 MBps. (And the 'maxed out' 7200 rpm drive is not that much faster, at 46MBps). That tells me that my FW800 drive is achieving nearly maximum transfer speed. But now I don't understand why the USB 2.0 drive would be half as fast, if its 'claimed' speed of 480 MBps is still 10x faster than the maximum read or write speed of a 5600 rpm hard drive?
Is something wrong with my math?
15 replies
Sign in or join Beachball Help to reply.
- 0
Peter Arnold1 ·
Imported from the Apple Support Forums via the Wayback Machine 🔗
One issue is that you are mixing up Megabits-per-second (Mbps) and MegaBytes-per-second (MBps). By convention, little "b" equals bits and capital "B" equals bytes. Firewire and USB are specified in Mbps while transfer rates of hard disk drives are typically measured in MBps. A byte is eight bits.
For example, FW800 has a maximum rate of 800 Mbps which is equivalent to 100 MBps. Similarly, USB has a maximum rate of 480 Mbps which is equivalent to 60 MBps.
On paper, both USB 2.0 and FW800 transfer rates should accommodate the maximum transfer rate of the hard disk drive. In practice, USB 2.0 is always slower because the USB protocol was never designed to accommodate long, sustained transfers. USB was designed for keyboards and mice. Firewire, on the other hand, was designed for hard drives from the get-go.
- 0
J Michael ·
Imported from the Apple Support Forums via the Wayback Machine 🔗
Thanks for the excellent explanation! So by that same logic, due to the read/write limitations of most hard drives, should I not expect any appreciable "speed" increase in transferring large amounts of data to/from drives with FW800, as opposed to FW400?
- 0
Peter Arnold1 ·
Imported from the Apple Support Forums via the Wayback Machine 🔗
Correct. FW400 has a maximum line rate of 50 MBps which is greater than the 46 MBps rating of the 7200 RPM disk drive you described. However, if you were to use a disk drive whose transfer rate exceeded 50 MBps, then FW 800 would be beneficial.
- 0
J Michael ·
Imported from the Apple Support Forums via the Wayback Machine 🔗
So, in other words, there is no reason for me to spend the extra money to get a FW800 external drive compared to the less expensive FW400 drive 🙂
- 0
Peter Arnold1 ·
Imported from the Apple Support Forums via the Wayback Machine 🔗
Yes, so long as the transfer rate of the drive is less than 50 MBps. One advantage of buying a FW 800 enclosure is that, in the future, you could install a faster hard drive into the enclosure and realize improved performance. There are hard drives available today that have sustained transfer rates specified at 100 MBps.
- 0
camoracer ·
Imported from the Apple Support Forums via the Wayback Machine 🔗
Can I jump in?
Does number of devices in Firewire chain have an effect on speeds? If so then FW800, having more capacity, might support more devices before being limited.
- 0
Rod Hagen ·
Imported from the Apple Support Forums via the Wayback Machine 🔗
One thing that I think should be emphasised here is Peter's comment concerning the less than optimal performance of USB 2.0 based drives.
The Mbps figures alone simply don't tell the story. A supposedly "480 Mbps" USB 2.0 drive connected to a Mac will a be much, much slower than a "400 Mbps" FW 400.
On older PPC based Macs the difference was quite staggering with USB 2 struggling to make even half the speed of FW 400, and operating at less than a third that of FW 800 in real world tests - see
http://www.barefeats.com/usb2.html for example.Things are a little better on the Intel based Macs, but, as the barefeats article at
http://www.barefeats.com/hard69.html points out, USB 2 drives on Macs still struggle to reach 20 MBps, while even FW400 manages 36 MBps.USB 2. performance is one area where, I hate to say it, many Windows machines beat Macs by a substantial margin, though even then it will usually be slower than FW400.
If you are looking for an external drive for your MBP then *look for one with at least FW 400, and grumble if your favourite supplier doesn't carry them.* Those cheap USB 2 drives might look like good value, and yes, you can even boot from them with current macs, but they will be as slow as the proverbial snail compared to a FW 800 , or even FW 400, drive.
The other option, these days, is an eSata drive connected through your Express Card port. With an appropriate fast drive fitted this will be the quickest option, though also the most expensive by a significant margin by the time you add in the necessary express card adaptor, and, of course, it takes up a port that you might wish to use for something else.
Cheers
Rod
PS , Camoracer, as I understand it FW is able to handle chained devices with less speed loss. With USB especially, slow downs are likely to be very dramatic (regardless of platform) if you combine "low speed" devices on the same USB bus as your HD and use them simultaneously.
Message was edited by: Rod Hagen
- 0
J Michael ·
Imported from the Apple Support Forums via the Wayback Machine 🔗
Thanks, all. Very helpful discussion. I also have the question about daisy-chaining, and in particular, whether it is "faster" to copy files from one external drive to another directly, or via the computer.
I do recall that I ran r/w a test on a previous FW800 drive using the FW800, FW400, and USB2.0 connections and the times were very similar. I found that surprising. The little test I ran for this question was on two separate drives, albeit both are Western Digital ones. I believe the 3.5" FW800 drive is a 7200 rpm one, and the 2.5" USB one is 500 rpm.
So now, out of curiosity, I'll run a test with just the FW800 drive to see whether there is any difference in the r/w speeds using different interfaces with the same drive. I'll let you know.
- 0
Peter Arnold1 ·
Imported from the Apple Support Forums via the Wayback Machine 🔗
The Firewire protocol was designed to allow autonomous transfers between devices where each device gets its own slice of the bandwidth. USB protocol does not allow for daisy-chaining and works in master/slave mode where multiple devices must be hubbed with the CPU organizing the data flow.
So, in theory, Firewire is better at sharing bandwidth across multiple devices and will provide better throughput than USB devices sharing a datalink. I don't have any personal experience with multiple, daisy-chained Firewire drives. I've read that, in practice, the performance of daisy-chained FW drives is quite a bit less than the ideal. Regardless, I would expect it to be superior to hubbed USB drives at least to some extent. If nothing else, using Firewire doesn't put as big a load on the CPU.
You might want to Google on daisy-chained FW hard drive performance. I know some testing results have been published.
- 0
camoracer ·
Imported from the Apple Support Forums via the Wayback Machine 🔗
Thanks Peter and Rod. This has been really interesting. I have been using FW400 external drive for backup for many years on the belief they were faster. Never really understood why though.
- 0
J Michael ·
Imported from the Apple Support Forums via the Wayback Machine 🔗
Okay, I did a simple test. 1,400 files, total 2.7GB, transfering from my MacBook to an external drive with all three interfaces. Transfer to = Write, Transfer back = Right.
I did the test twice, and took the faster figures from each. Here are the results, in Mbps
Write Read
USB 2.0 166 225
FW400 232 245
FW800 300 348Given that this is a 7200 RPM drive with a maximum r/w speed of about 360 Mbps (45 MBps), one would assume that all three of the connection methods should be able to easily achieve the full 360 Mbps, since that is below their 'maximum' ratings. However, none of them did.
The other interesting point is that while FW400 and 800 were both similar for read and write, USB 2.0 was much faster reading (i.e. transfering data from the external drive to the computer) than writing.
This also tells me that for real-life usage, USB2.0 is only about 55~60% as fast as FW800, whereas FW400 is only 70~75%. So despite the fact that none of them are as fast as their 'max,' the FW800 comes closer.
- 0
Kappy ·
Imported from the Apple Support Forums via the Wayback Machine 🔗
Do bear in mind that the speed ratings for those ports are their interface transfer rates, not throughput rates. So although FW400 has an interface rate of 400 Mbps (50 MBps) that isn't the throughput rate which is usually much slower. Your hard drive has an interface rate of 150 or 300 MBps (SATA 1 or SATA 2) but throughput rates are usually closer to 50-80 MBps.
- 0
J Michael ·
Imported from the Apple Support Forums via the Wayback Machine 🔗
Do bear in mind that the speed ratings for those ports are their interface transfer rates, not throughput rates. So although FW400 has an interface rate of 400 Mbps (50 MBps) that isn't the throughput rate which is usually much slower. Your hard drive has an interface rate of 150 or 300 MBps (SATA 1 or SATA 2) but throughput rates are usually closer to 50-80 MBps.
Yes but it's confusing because my HDD throughput rate is in the area of 45~50 MBps, so that would lead one to believe that because the FW400 and FW800 interface rates are both much faster (50MB/s and 100 MB/s, respectively) therefore they should both have the same transfer times. One would expect transfer rates of 45 MB/s for both FW400 and FW800, unless there is a significant difference in their architecture.
- 0
Kappy ·
Imported from the Apple Support Forums via the Wayback Machine 🔗
The maximum interface rate for FW400 is 50 MBps, and FW800 is 100 MBps. But throughput rates are a function of other things such as the bridge card used in the enclosure and the drivers as well as the size of the data packets. Throughput rates are therefore considerably slower. More in the order of 10-40 MBps for sequential read/writes.
- 0
J Michael ·
Imported from the Apple Support Forums via the Wayback Machine 🔗
That does not explain why the FW400 would be slower than the FW800, particularly if the lowest common denominator is the HDD r/w speed. Bridge card, driver, data packets, are all the same.