Speed difference FW 800 vs. FW 400 - test

Dafke

macrumors 6502
Original poster
Mar 24, 2005
261
0
Hey all, for some reason I always thought FW 800 is twice as fast as FW 400 but I just found out I'm wrong, or maybe there's something wrong with my equipment :(

Here's the situation:
I have this enclosure for a SEAGATE Barracuda7200.10 250GB HDD. Today I decided to test the difference between FW 400 and FW 800. I transfered a folder the size 7,68 GB (ripped DVD image) from my MacBook Pro to the external:
first through the FW 800 connection: 3 min 07 sec
then the FW 400 connection: 3 min 30 seconds

I expected the difference to be much bigger! Do you think something might be wrong with the harddrive or the enclosure?
 

kolax

macrumors G3
Mar 20, 2007
9,188
115
Nothing is wrong - the hard drive simply cannot output at FW800's maximum speed.

Also bear in mind, 800Mbp/s (megabits, not megabytes) is the maximum speed, and it won't reach that very often. It really depends on the speed of the device you are using it with.
 

n0d3

macrumors regular
Oct 12, 2006
163
0
Did you connect anything on the FW400 port when using the 800? Both share the same bus so connecting anything slower will slow it down to that speed.

Here was some benchies I did using Xbench quite some time back.

Hitachi T7K500 320GB (16MB) SATAII over FW800:
Disk Test 60.29
Sequential 85.07
Uncached Write 112.79 69.25 MB/sec [4K blocks]
Uncached Write 100.34 56.77 MB/sec [256K blocks]
Uncached Read 52.24 15.29 MB/sec [4K blocks]
Uncached Read 110.56 55.56 MB/sec [256K blocks]
Random 46.69
Uncached Write 17.01 1.80 MB/sec [4K blocks]
Uncached Write 121.25 38.82 MB/sec [256K blocks]
Uncached Read 93.18 0.66 MB/sec [4K blocks]
Uncached Read 126.37 23.45 MB/sec [256K blocks]

Hitachi T7K500 320GB (16MB) SATAII over FW400:
Disk Test 52.03
Sequential 57.72
Uncached Write 59.60 36.59 MB/sec [4K blocks]
Uncached Write 57.36 32.46 MB/sec [256K blocks]
Uncached Read 45.57 13.34 MB/sec [4K blocks]
Uncached Read 76.11 38.25 MB/sec [256K blocks]
Random 47.36
Uncached Write 18.19 1.93 MB/sec [4K blocks]
Uncached Write 102.47 32.80 MB/sec [256K blocks]
Uncached Read 93.78 0.66 MB/sec [4K blocks]
Uncached Read 110.26 20.46 MB/sec [256K blocks]
 

netnothing

macrumors 68040
Mar 13, 2007
3,654
284
NH
Did you connect anything on the FW400 port when using the 800? Both share the same bus so connecting anything slower will slow it down to that speed.
So are you saying that on my Mac Pro...if I have something plugged into both the FW400 and FW800 ports, I won't ever get any speed benefit from the FW800 port? It will slow down to FW400 speeds?

Is this just when both are being used at the same time?

-Kevin
 

CanadaRAM

macrumors G5
No, it means that the test results are likely limited by the performance of the MacBook internal drive, more than reflecting a difference between 400 and 800.

I have a 4 inch diameter garden hose and an 8 inch garden hose. If I hook up a 3 inch garden hose to each one, and measure the output, I could conclude there's no improvement by going to the 8 inch.
 

mkrishnan

Moderator emeritus
Jan 9, 2004
29,641
12
Grand Rapids, MI, USA
Can I ask a semi-related question, to understand the reality of bottlenecking here...

Theoretical max speeds:
FW400 = 400mbps
FW800 = 800mbps
SATA (first gen) = 1.5gbps

So now suppose you have a 7200RPM drive in a computer. If you take it out and put it in a FW400 enclosure (single drive), are you likely to notice any real degradation in performance?

I have a sad example... I think the SATA controller on one of my computers is dead. It has a 7200RPM 160GB HD (I think it's a Deskstar). I think the controller is dead and the drive is fine; if that's the case, I'm going to try at least temporarily running this computer off a firewire boot drive in an external enclosure.

It seems that the max transfer rate from cache to controller for that drive is about 300mbps, which would seem to indicate to me that it should be nearly as fast on the FW bus as it is internally. Is that reasonable?
 

netnothing

macrumors 68040
Mar 13, 2007
3,654
284
NH
No, it means that the test results are likely limited by the performance of the MacBook internal drive, more than reflecting a difference between 400 and 800.

I have a 4 inch diameter garden hose and an 8 inch garden hose. If I hook up a 3 inch garden hose to each one, and measure the output, I could conclude there's no improvement by going to the 8 inch.
I guess I was asking whether it was like wireless routers. Where if you have a wireless router setup to accept both 802.11b and 802.11g, if you have any 'b' devices on the network....it will slow down the 'g' devices to 'b' speeds.

So in my case...I have the 2 ports on my Mac Pro. If I have something plugged into the FW400 port, does that automatically slow the FW800 port down?

I realize there are other limitaions (HD speeds, etc).

-Kevin
 

Dafke

macrumors 6502
Original poster
Mar 24, 2005
261
0
Nothing is wrong - the hard drive simply cannot output at FW800's maximum speed.

Also bear in mind, 800Mbp/s (megabits, not megabytes) is the maximum speed, and it won't reach that very often. It really depends on the speed of the device you are using it with.
Damn, if I had known this i wouldn't have invested in an expensive FW 800 enclosure. On the other hand, it is the internal hard drive of my MacBook Pro that is keeping things slow, right? So the FW 800 connection is still useful for booting from the external drive?
Thanx!
 

Dafke

macrumors 6502
Original poster
Mar 24, 2005
261
0
Did you connect anything on the FW400 port when using the 800? Both share the same bus so connecting anything slower will slow it down to that speed.
No, nothing else was connected to the FW ports when testing. I'll see if I can run some benchmarks soon.
Thank you too!
 

netnothing

macrumors 68040
Mar 13, 2007
3,654
284
NH
So if your going to use an external drive as merely a backup drive, there's no need to go FW800? I'm guessing Time Machine wouldn't take advantage of FW800 over FW400?

Why would you go FW800? I guess only if you want to run the system from that external drive?

-Kevin
 

Dafke

macrumors 6502
Original poster
Mar 24, 2005
261
0
So if your going to use an external drive as merely a backup drive, there's no need to go FW800? I'm guessing Time Machine wouldn't take advantage of FW800 over FW400?

Why would you go FW800? I guess only if you want to run the system from that external drive?

-Kevin
Well, I just thought it would be nice if backing could be done fast plus I use the drive as a scratch disk for Photoshop and I thought i would use it for booting from the external drive (which I haven't done very often; after working like that you have to be sure to sync the two drives again if you whish to work from the internal drive again. Also the external drive gets rather hot after using it for a while).
 

netnothing

macrumors 68040
Mar 13, 2007
3,654
284
NH
Well, I just thought it would be nice if backing could be done fast plus I use the drive as a scratch disk for Photoshop and I thought i would use it for booting from the external drive (which I haven't done very often; after working like that you have to be sure to sync the two drives again if you whish to work from the internal drive again. Also the external drive gets rather hot after using it for a while).
I agree....I guess your tests will tell the true answer. Let us know when you have some data.

-Kevin
 

Dafke

macrumors 6502
Original poster
Mar 24, 2005
261
0
xbench test results

Okay, I've done some testing, here we go:

The tested drive is a SEAGATE Barracuda7200.10 250GB HDD. However, this drive is split in two partitions, I tested the largest one which is 149 GB large with 38 GB of free space.

Over FW800:

Disk Test 49.10
Sequential 89.08
Uncached Write 109.94 67.50 MB/sec [4K blocks]
Uncached Write 91.87 51.98 MB/sec [256K blocks]
Uncached Read 57.87 16.94 MB/sec [4K blocks]
Uncached Read 130.84 65.76 MB/sec [256K blocks]
Random 33.89
Uncached Write 11.79 1.25 MB/sec [4K blocks]
Uncached Write 66.68 21.35 MB/sec [256K blocks]
Uncached Read 91.38 0.65 MB/sec [4K blocks]
Uncached Read 136.74 25.37 MB/sec [256K blocks]

Over FW400:

Disk Test 41.54
Sequential 55.36
Uncached Write 49.25 30.24 MB/sec [4K blocks]
Uncached Write 61.06 34.55 MB/sec [256K blocks]
Uncached Read 46.04 13.47 MB/sec [4K blocks]
Uncached Read 72.22 36.30 MB/sec [256K blocks]
Random 33.24
Uncached Write 11.96 1.27 MB/sec [4K blocks]
Uncached Write 61.57 19.71 MB/sec [256K blocks]
Uncached Read 90.20 0.64 MB/sec [4K blocks]
Uncached Read 106.14 19.70 MB/sec [256K blocks]

Over USB 2.0:

Disk Test 16.11
Sequential 12.15
Uncached Write 24.07 14.78 MB/sec [4K blocks]
Uncached Write 28.75 16.26 MB/sec [256K blocks]
Uncached Read 4.45 1.30 MB/sec [4K blocks]
Uncached Read 35.52 17.85 MB/sec [256K blocks]
Random 23.92
Uncached Write 8.81 0.93 MB/sec [4K blocks]
Uncached Write 39.89 12.77 MB/sec [256K blocks]
Uncached Read 69.04 0.49 MB/sec [4K blocks]
Uncached Read 70.78 13.13 MB/sec [256K blocks]

So, first we notice that USB2.0 is ridiculously slow! Second we notice n0d3 has a faster drive (is it faster because it's bigger?). Third, the difference between FW400 and FW800 is less than 20%.
 

netnothing

macrumors 68040
Mar 13, 2007
3,654
284
NH
Okay, I've done some testing, here we go:

The tested drive is a SEAGATE Barracuda7200.10 250GB HDD. However, this drive is split in two partitions, I tested the largest one which is 149 GB large with 38 GB of free space.

Over FW800:

Disk Test 49.10
Sequential 89.08
Uncached Write 109.94 67.50 MB/sec [4K blocks]
Uncached Write 91.87 51.98 MB/sec [256K blocks]
Uncached Read 57.87 16.94 MB/sec [4K blocks]
Uncached Read 130.84 65.76 MB/sec [256K blocks]
Random 33.89
Uncached Write 11.79 1.25 MB/sec [4K blocks]
Uncached Write 66.68 21.35 MB/sec [256K blocks]
Uncached Read 91.38 0.65 MB/sec [4K blocks]
Uncached Read 136.74 25.37 MB/sec [256K blocks]

Over FW400:

Disk Test 41.54
Sequential 55.36
Uncached Write 49.25 30.24 MB/sec [4K blocks]
Uncached Write 61.06 34.55 MB/sec [256K blocks]
Uncached Read 46.04 13.47 MB/sec [4K blocks]
Uncached Read 72.22 36.30 MB/sec [256K blocks]
Random 33.24
Uncached Write 11.96 1.27 MB/sec [4K blocks]
Uncached Write 61.57 19.71 MB/sec [256K blocks]
Uncached Read 90.20 0.64 MB/sec [4K blocks]
Uncached Read 106.14 19.70 MB/sec [256K blocks]

Over USB 2.0:

Disk Test 16.11
Sequential 12.15
Uncached Write 24.07 14.78 MB/sec [4K blocks]
Uncached Write 28.75 16.26 MB/sec [256K blocks]
Uncached Read 4.45 1.30 MB/sec [4K blocks]
Uncached Read 35.52 17.85 MB/sec [256K blocks]
Random 23.92
Uncached Write 8.81 0.93 MB/sec [4K blocks]
Uncached Write 39.89 12.77 MB/sec [256K blocks]
Uncached Read 69.04 0.49 MB/sec [4K blocks]
Uncached Read 70.78 13.13 MB/sec [256K blocks]

So, first we notice that USB2.0 is ridiculously slow! Second we notice n0d3 has a faster drive (is it faster because it's bigger?). Third, the difference between FW400 and FW800 is less than 20%.
Not sure how to read the xbench results. But I guess what you're saying is that there's no need to go FW800 over FW400 if all your doing is backing up data?

-Kevin
 

n0d3

macrumors regular
Oct 12, 2006
163
0
I was referring to the MacBook Pro ports, the FW400/800 ones share the same bus. I'm not sure about the Mac Pro.
 

SpaceJello

macrumors 6502
Dec 2, 2006
383
1
So it seems like from the benchmarks, that FW800 is really limited by the macbook pro's firewire bus that shares between FW800 and FW400.

Curious:
If you have an expresscard for either eSATAor FW 800, and connecting another drive through the ports on that expresscard (either eSATA or FW800), would it be possible to see a higher bench mark then the ones you see with FW 800 through the build in mbp port since they are on a different bus?

Or are they on the same bus?
 

mkrishnan

Moderator emeritus
Jan 9, 2004
29,641
12
Grand Rapids, MI, USA
Anyone want to repeat that with a striped RAID? :D
So it seems like from the benchmarks, that FW800 is really limited by the macbook pro's firewire bus that shares between FW800 and FW400.
I was only being half facetious earlier... once again, the specified maximum external transfer rate (that is, cache to bus) for the Seagate 7200.10s is 300mbps (source). The drive itself is faster than that, but you can ignore the 780mbps specification because the cache cannot transfer the data out fast enough to support that rate.

So once again, if you run this test with current single HD technologies, the output mechanism of the drive assembly is the rate limiting factor, and not the USB, FW, or eSATA bus. The only way (correct me if I'm wrong) to get valid results for this kind of study is to use a system that can push data to the bus at the very least consistently faster than 400mbps (and ideally, if you want to know the difference between FW400 and FW800, faster than 800mbps).

E.G. a striped hardware RAID. Or perhaps a computer with enough memory that it can push a large (several gigabyte) file across the FW bus from a RAM disk. But not a single 7200.10 drive in an enclosure.
 

Dimwhit

macrumors 68000
Apr 10, 2007
1,968
162
I use FW800 on my MBP and I've been pleased with the speed difference compared to FW400. I transfer rather large files frequently, and there's a noticeable difference for me.
 

velocityg4

macrumors 601
Dec 19, 2004
4,615
1,190
Georgia
The speed advantages of firewire 800 really depend on your use. If you are using the external hard drive simply as a backup of your Macbooks internal hard drive there really is no advantage, even though a new external will most likely be able to outperform a firewire 400 bus the Macbooks internal will not outpace it unless it is a 7200RPM Model. Even with the 7200RPM it will operate just below the 50MBps threshold but in burst transfers that hard drive can outperform the firewire 400 bus.

When using a firewire 800 drive as a scratch disk, boot drive and/or file storage with a large modern 3.5inch drive then the firewire 800 is advantageous. Since the external drive can work independently of the Macbooks internal drive removing this bottleneck. Since the hard drive can read write faster than firewire 400 it can not show it's full potential unless being used under firewire 800.

Firewire 400's maximum transfer is 50 Megabytes per second.

Firewire 800's maximum transfer is 100 Megabytes per second.

The typical new 2.5 inch 5400 RPM Hard Drives average read transfer rate is around 35 Megabytes per second. For 160GB hard drives 250GB models are around 45 MBps.

The typical new 2.5 inch 7200 RPM Hard Drives average read transfer rate is around 45 Megabytes per second. For 160GB hard drives smaller models are slower.

The typical modern 3.5 inch 7200 RPM Hard Drives average read/write transfer rate is around 60 Megabytes per second. For drives with data densities of around 160GB per platter and features like perpendicular recording.

Curious:
If you have an expresscard for either eSATAor FW 800, and connecting another drive through the ports on that expresscard (either eSATA or FW800), would it be possible to see a higher bench mark then the ones you see with FW 800 through the build in mbp port since they are on a different bus?

Or are they on the same bus?
The expresscard would be on a seperate bus. So theoretically you could get better performance, but if the firewire 800 hard drive is the only firewire device connected to the computer the point is moot since a single hard drive would not max out that bus.

References:
2.5 inch HD speeds
http://www.tomshardware.com/2007/08/02/speed_or_capacity/page8.html
3.5 inch HD Speeds
http://www23.tomshardware.com/storage.html?modelx=33&model1=117&model2=676&chart=34
Firewire
http://en.wikipedia.org/wiki/FireWire#Standards_and_versions
 

md63

macrumors 6502
Jun 11, 2007
264
0
I just purchased a Lacie d2 quadra to use with my MBP. The transfer of 3GB of data between the internal HDD to the Lacie by both FW800 and FW400 was identical. I had expected much faster speeds. I don't understand the usefulness of FW800 if the internal drive is the limiting factor in data tranfers.