Click for SMOKING FAST SATA 6G SSDs!
Click for SMOKING FAST SATA 6G SSDs!


A Click on this Banner shows your site support to my Sponsors


www.xlr8yourmac.com

Firewire Direct's Portable Firewire Hard Drive Kit
How it Compares to VST's Superslim Drives
by Mike
Published: 5/27/2000
(Last Updated: 5/31/2000 for tests after PB Firmware update)

Performance Tests

Important Notes on Performance:

I want to remind the reader of several things to be aware of when examining the results of performance tests on this page.

  • I used a 12GB Toshiba drive for this kit build. Different drives may show better or worse results.

  • Firewire Direct's pre-built models may perform differently, depending on the drive they use.

  • Test results with VST's 12GB slim drive (graphite model) were with rev 55 firmware/v2.1 drivers (latest as of 5/26/2000). Version 2.1 drivers significantly boosted performance in some tests over the previous driver version.

  • VST (and Firewire Direct) may use several different drive brands which can affect performance. Since you cannot specify drive brands on the pre-assembled drive, performance may vary even if you were to order the same VST 12GB drive I tested. Also note for example that VST's specs on the 6GB drives show much lower rates (about 1/2) than the 12GB models (assuming due to drive used).

    Update: For an example of how each retail drive can vary, see the update below with tests from another new 12GB VST portable drive.

  • For owners of older Macs with Firewire PCI cards, the choice of card can make a dramatic impact on performance. This is shown in some of the tests below comparing a RocketFire PCI card and a Western Digital PCI card in the S900. The TI firewire chip on the RocketFire card in some cases delivered dramatic performance gains over the NEC chip used on the Western Digital card. (Note: Other Firewire cards also use TI chips, including the $69 FirewireDirect model I tested after this article which had similar performance to the RocketFire card.)

  • Remember that a stopwatch was used for the Finder file copy timed tests, therefore human error in repeatability of starting and stopping the timer could easily make a fraction of a second difference in run/run times. Consider times with less than 1 second differences in these tests as practically equal.

Benchmark Tests:

The first test results shown are from ATTO's popular benchmark. The graph below shows the results of ATTO Tools v2.3 tests set to 8MB sample size (0-8MB test file size):

FW Direct drive with Pismo

The next graph shows the drive's performance in an S900 w/XLR8 G3/300 cpu upgrade using a Rocketfire Firewire PCI controller:

FW Direct drive with S900/RocketFire card

To show how much faster the RocketFire card is than the Western Digital Firewire PCI card in the S900, take a look at the write rates in the following graph (same system, same drivers, same drive)

Firewire Direct drive with WD firewire card

For comparison, here's the ATTO results from the more expensive VST 12GB drive (using Pismo G3/500 Powerbook onboard firewire):

Atto results - VST 12GB portable drive


Update: Tests After PowerBook Firmware 2.04 Update: On 5/30/2000, Apple released a firmware update for the PowerBook Firewire models to offer improved Firewire and USB drive support. I repeated the tests 3 times after seeing the lower write rates. (Tests run after a clean reboot and fresh format of the drive. Identical system/extensions/drive was used, only the PB firmware had changed.)

Before/After PB Firmware Update
Using Firewire v2.3.3
(Sustained Rates - VST Superslim 12GB drive)
(ATTO Benchmark Performance)
Test
Before Update
After Update
Writes
11.9 MB/sec
9.92 MB/sec
Reads
13.53 MB/sec
13.09 MB/sec

Below are graphs of ATTO benchmark results after the firmware update. Included to illustrate my original comment on how assembled drives can vary in performance (since there's no way to tell the brand/model of drive used), results of tests with another brand new VST 12GB drive are shown. Note the dramatic differences in the write curves. The original drive has not only slightly higher rates, but much flatter write curve. Perhaps a different brand drive, different HD cache size is the reason why. The same VST firmware revision and identical system was used to test. Tests were repeated 3 times after a clean reboot to ensure results were repeatable.

First VST 12GB drive after Firmware update
First VST 12GB Drive Sample

2nd VST 12GB Sample after firmware udpate
Second VST 12GB Drive Sample

Bottom line is there's no way to ensure what drive brand/model you will be getting with most retail (completed) drives and you can't assume review tests results will be repeatable with the drive you buy.


Other Post-Firmware Update Comments: One other oddity I noticed after the firmware update was the fact ATTO Tools utility lists four 0.0 MB volumes (unmounted) in the drive list on the PB. (Note - Apple's drive setup does not show these 0MB volumes, only ATTO's tool utility.) I don't remember seeing this before the firmware update but I'm curious if other PB owners see the same thing. Clicking the info icon next to the 0MB volumes doesn't report any useful info. The 18GB Powerbook internal drive was formatted as a single partition, using Apple's drivers/drive setup from OS 9.

ATTO Volumes oddity

BTW - I see *one* 0.0 MB volume like this on my G4/AGP system as well (firmware update applied yesterday). Again it could have been there before, maybe I just never noticed it. I booted the G4/AGP system with my 2nd IDE drive with OS 8.6 and even a previous version of ATTO Tools showed the one 0MB volume. (Perhaps I just never noticed this oddity in ATTO Tools before.)


[Note: all the tests results below were run *before* the firmware update. I've not had time to repeat tests after the update was applied.]

I also ran Macbench 5.0 disk tests on the portable firewire drive. The graph includes the following comparisons:

  • Firewire Direct Kit w/12GB Toshiba drive connected to PB G3/500 onboard firewire
  • Firewire Direct Kit w/12GB Toshiba drive connected to S900 w/RocketFire (RP) PCI firewire
  • Firewire Direct Kit w/12GB Toshiba drive connected to S900 w/Western Digital (WD) PCI firewire
  • PB G3/500 internal 18GB Toshiba drive
  • PB G3/500 expansion bay kit w/18GB Toshiba drive
  • VST 12GB portable firewire drive (latest firmware/driver)

The expansion bay drive results were from my previous how-to article. To see how the OEM 12GB Toshiba drive included with some PB G3s, see my original PowerBook G3/500 Firewire review benchmarks page.

Macbench 5.0 Disk/PubDisk results

The graph below shows the results of MacBench 5.0's disk inspection tests. Note the dramatic difference in some tests between the Rocketfire (RF) PCI card versus the Western Digital (WD) card in the S900.

Disk Inspection tests - portable </center>drives


Finder File Copy Tests:

I ran tests on the PowerBook G3/500 to time how long it took to copy the system folder from the internal 18GB Toshiba drive to the portable firewire drive. System was on AC power and all conditions were identical except for the extension loaded (VST drive requires its own extension to mount, Firewire Direct drive would mount with either extension, but theirs was used of course for these tests).

In the real 'through-the-file-system' copy tests I calculate the MB/sec rates (rounded to 1 decimal place). Capture rates from DV apps would be much higher than these rates, based on tests I've run in the past with MotoDV. (Due to the patch for MotoDV not working properly I was not able to report its capture rates on the drives. Digital Origin is to answer my question on why my registration number will not take after the update. I'll post results of MotoDV capture rate tests when this issue is resolved.)


Copy 256MB System Folder (mix of file sizes): The table below shows the time to complete a file copy of a 256MB (2,596 files) from the internal drive to the firewire drive. (Again remember that a stopwatch was used for all these tests, so the start/stop human error/repeatability issue makes times that are less than a second difference to be practically identical.) The effective MB/sec rates look low, but capture rates from an application would be far higher, as there is a lot of overhead from copying in the Finder (especially with thousands of files involved).

Time to Copy 256MB System Folder
(From PB G3/500 internal 18GB drive to Portable Firewire Drive)
FW Direct Kit
(12GB Toshiba)
VST Portable
(12GB)
2:01
2.12 MB/sec
1:34
2.72 MB/sec


Single Movie File Copy/Duplicate Tests: The next tests I ran were times to copy a 51.6MB DV movie file from the PB G3's hard drive to the firewire drive. I then timed how long it took to duplicate the same file from the firewire drive. (The test file was Clip 05 from the iMovie Tutorial.)

Time to Copy 51.6MB DV Movie
(Copies were from PB G3/500 internal 18GB drive to FW drive)
Test
FW Direct Kit
(12GB Toshiba)
VST Portable
(12GB)
Copy File
11.06 Sec.
(4.67 MB/sec)
11.26 Sec.
(4.58 MB/sec)
Dupl. File
11.65 Sec.
(4.43 MB/sec)
10.94 Sec.
(4.72 MB/sec)

Either of these drives should be more than adequate for digital video capture (appx. 3.8MB/sec rates) since capturing video from an application doesn't have all the file I/O overhead of the Finder. On my old Wallstreet PB G3/250 with OEM 4GB drive for instance, MotoDV reports capture rates of 6.6MB/sec (many times faster than typical Finder file copy rates).


Movie Playback Tests: I also tested the framerates from playing back a movie from the portable firewire drive. The test movie was the 44KHz audio HQ version of the Phantom Menace trailer (25.2MB file size, 24fps rate). This test was a wash basically, with both drives playing back the movie at at 24fps, dipping to a minimum of 23.3fps for a second at a few points in the movie. Performance was literally identical with both drives.

After playback tests were run, I performed a 'save as' of the movie to the firewire drive, selecting a self-contained movie. The table below shows the times to complete:

Save Self-Contained QT Movie 25.2MB
(Phantom Menace 44KHz HQ Movie)
FW Direct Kit
(12GB Toshiba)
VST Portable
(12GB)
19.40 Sec.
19.78 Sec.

Note the above times include processing, and are not limited by file I/O speeds.

Test Systems Used:

  • PowerBook G3/500 Firewire
  • OS 9.04 with all updates applied (Multimedia update, QT 4.1, etc.)
  • Internal 18GB Toshiba drive
  • Firewire 2.3.3 drivers/Onboard Firewire ports used
  • Resolution Used: 1024x768, thousands colors
  • Latest drivers for each disk used (VST v2.1/FW Direct 1.8FWd)
  • No other firewire devices connected during tests

  • UMax S900
  • XLR8 G3/300 CPU upgrade with 1MB cache @200MHz
  • OS 9.04
  • Acard ATA/66 PCI IDE controller w/3.01 firmware
  • Maxtor 27GB DiamondMax Plus & Quantum 6GB IDE drives
  • Voodoo3 3000 PCI graphics card (B11 drivers)
  • Firewire 2.3.3 drivers
  • Both Western Digital Firewire & RocketFire Firewire/USB PCI cards tested
    (both cards were tested in PCI slot 5)
  • Resolution Used: 1024x768, thousands colors
  • No other firewire devices connected during tests



For More Drive Related Articles & Reviews - see our Storage Topics page


Index of Portable Firewire Drive Kit Project

Intro | Case Assembly | Software Install | Benchmarks

= or =
Back to
www.XLR8YOURMAC.com


Copyright © 2000, All Rights Reserved
All brand or product names mentioned here are properties of their respective companies.

Site terms and conditions of use.