Do I *really* have USB 2.0 support?
Like a chain, there are many links that must all be just right in
order for the high speed USB service to be provided.
USB 1.1 --> 12Mbps or 1.5MByte/s theoretical max for the interface. 
USB 2.0 --> 480Mbps or 60MByte/s theoretical max for the interface.
Actually sustained data transfer rate to any one USB 2.0 device is
likely to be limited by the device rather than the pipe.

I have a Dell Inspiron 8200 laptop.
It has the Intel Pentium 4-m @ 1.6GHz and 266MHz DDR SDRAM
The chipset is the Intel 845MP.
The external device is USB 2.0 rated, appproved. Its all over the packaging.
Windows XP Pro was pre-installed by the OEM.
I excitedly installed Sp1 because WIndows XP support for USB 2.0 was
not available in the original Windows XP base.
I added a PCcardBus into the slot for USB 2.0 support.  It had its own drivers
(Orange) and according to Windows Device Manager and confirmed by
Orange tech support their USB 2.0 drivers are installed correctly.

While the USB 2.0 external device works beautifully thru PCcardBus I
have never been able to get anything near the thruput on my external
device
that I orginally expected.  It is close to, but often slightly over
the thruput limit imposed by the USB 1.1 standard.
I came across an Intel document in pdf format where they show a block
diagram and describe the USB 1.1 component.  Thats' 1.1.  Now it has
me "worried".

So, do I really have all the pieces, all the ducks in a row, to support
USB 2.0? Is my Intel 845MP a factor, a limiting factor? It is the
missing link in the chain that prevents me from attaining the full
thruput?
USB 2.0 thruput?

---

Clarification of Question by grthumongous-ga on 12 Mar 2004 17:20 PST

BIOS firmware is revision A09 which is supposed to include USB 2.0

---

Clarification of Question by grthumongous-ga on 13 Mar 2004 00:22 PST

I swapped out the USB cable which had no particular "Hi-speed USB 2.0"
designation and replaced it with one that does have that text
stenciled onto the wire.  The terminators have nothing like the
Hi-Speed Logo.

The 10 minute test run showed no measurable difference, but it remains
an intriguing angle.   I will look for a super duper cable on Saturday
and update with the results.

---

Request for Question Clarification by aht-ga on 13 Mar 2004 00:32 PST

What is the external device? Is your external device one that can
output a continuous stream of high-bandwidth data?

---

Clarification of Question by grthumongous-ga on 13 Mar 2004 19:27 PST

The output Travan Tape device is a STT6401U2.
I bought a new "Hi-Speed" rated USB 2.0 cable with the specific
designator and gold-plated contacts.
Reran test with Write of a 122MB file in 90 sec at 1.3MB/s

Write 80MB to DVD burner on same USB cardBus. Discarded result due to
unknown formating overhead.
Use Windows Explorer to Read (copy) the 80MB from DVD+RW media to
C:\Temp on the harddrive (paste) at default 4x speed in 18 sec at
4.4MB/s

The DVD read at 4.4MB/s exceeds the USB 1.1 speed limit.
Therefore the tape write thruput of 1.3MB/s must have been gated by
native data rate at the drive head.

Do you have any specific USB 2.0 confirmation on the Intel 845MP
chipset or general info on the role of a chipset in suppporting USB
2.0?

---

Request for Question Clarification by aht-ga on 13 Mar 2004 21:04 PST

grthumongous-ga:

In light of the comment I have posted below, can you clarify how I
might be able to help you with this Question?

Regards,

aht-ga
Google Answers Researcher

---

Clarification of Question by grthumongous-ga on 14 Mar 2004 10:49 PST

aht,
I think you have helped me a lot.
To recap, the 845MP chipset is not a factor because the cardbus came
with its own NEC controller. The tape write thruput might be limited
by the need to first read the harddrive so a ramdisk to cache the data
being written would be useful for the next trial.  I followed your
lead and am considering a product called none other than Ramdisk by
www.superspeed.com

You also said that video devices are the ones that can really push the
envelope on the input output interface.

Please make it "official" so you can claim your well earned remuneration.

---

Clarification of Question by grthumongous-ga on 14 Mar 2004 10:56 PST

PS, I did verify the drivers are Orange and thanks for reminding me.

grthumongous-ga:

Thanks for this opportunity to help you with your question. The
approach I have taken (originally in the Comment below) was to
address:

- whether the Intel 845MP/ICH3-M chipset in your Dell Inspiron 8200 is
interfering with your ability to achieve the maximum USB 2.0 speeds;

- whether there is a device or hardware problem that leads to the
speeds you are seeing;

- possible trials that can help optimize your transfer speeds.


CHIPSET
-------

(quoting from my comment below)

The Intel 845MP on your laptop's motherboard should not be involved in
this case, as the Orange Micro USB 2.0 Cardbus adapter you bought
incorporates an NEC chipset with full internal support for USB 2.0. In
fact, one of the biggest knocks against the Inspiron 8200 when it
first came out was its lack of USB 2.0 support, due to the
845MP/ICH3-M chipset. The ICH3-M is the I/O controller hub that
supports, among other things, the USB 1.1 ports built into your
Inspiron 8200. The Orange Micro Cardbus device you purchased
effectively works in parallel to the 845MP/ICH3-M, enabling you to
have USB 2.0 capabilities in spite of the 845MP/ICH3-M chipset running
your motherboard.

Intel 845MP Datasheet (see page 14 of 157 for block diagram)
http://www.intel.com/design/chipsets/datashts/25068702.pdf

Intel ICH3-M Datasheet (see chapter 11 for confirmation of USB1.1 only)
http://www.intel.com/design/chipsets/datashts/29071601.pdf

Orange Micro Cardbus USB2.0 datasheet (confirmation of NEC chipset and drivers)
http://www.orangemicro.com/bin/USB2/OrangeUSBDriverWinXPOM.pdf

DEVICE/HARDWARE ISSUES
----------------------

The first item I recommended checking was the rating on your USB
cable. After you confirmed that your cable is indeed USB2.0 rated, I
suggested turning your attention to the external hardware and the
possibility of speed limitations from your hard-drive:

(quoting from my comment below)

The challenge with the devices you mention is that they all are
dependent on your hard drives and hard drive controllers because you
are reading/writing data files. The truly high-speed USB 2.0 devices
tend to be video input devices. That said, the bit rates you report
are not very good if everything is supposed to be designed for USB
2.0. USB 2.0 is supposed to support serial communication at three
nominal speeds: 480 Mbps, 12 Mbps, and 1.5 Mbps. A device can be rated
as USB 2.0 compliant if it uses a controller that is certified as USB
2.0, even if the device itself is not capable of making use of the 480
Mbps speed. In a situation like that, the controller will drop down to
12 Mbps to conserve bandwidth.

In your trials, it looks like the best the combination of your PC and
the Travan can do is around 12 Mbps (roughly 1.5 MB/s). The only way
to truly know what the cause is, is to try the Travan unit on another
PC with USB 2.0 support, or eliminate any hard drive overhead by
setting up a RAM disk and reading/writing from there to the Travan.
The information available from Seagate RSS (now called Certance) is
conflicting: while they claim "up to 120 MB/min uncompressed" data
transfer rate, they also say that the transfer rate will be optimized
by FastSense to one of "1000, 733, or 489 kilobytes per second" to
reduce the amount of tape repositioning required. Accounting for
overhead and compression ratios, it is difficult to ascertain how this
FastSense feature will allow them to actually hit the maximum transfer
rate of 120 MB/min uncompressed. More likely, you will see between 60
MB/min to 120 MB/min performance, which is exactly what you observed
(122 MB/1.5 min = 81.33 MB/min).


POSSIBLE TRIALS FOR FURTHER INVESTIGATION
-----------------------------------------

Seagate RSS (now called Certance), the manufacturer of the Travan
TapeStor 40 USB2.0 device you are using, claims a maximum 'raw'
throughput of 120 MB/min, or 2 MB/sec. However, as I mentioned above,
they also state on their website that the actual throughput rate will
be optimized by their 'Fastsense' feature, to help reduce the need for
tape repositioning during write operations. It is possible that they
have not updated their website to account for the USB2.0 products, but
unfortunately there is no additional information available from them.

To help you get to the bottom of this, I suggested that you either
find a USB2.0 video device to truly push the bandwidth envelope and
see if your USB2.0 Cardbus device is working as advertised, or try a
RAM disk to eliminate your hard drive read/write overhead as a
variable in the experiment.


I'd be interested in finding out how the results look for you. While
USB2.0 is capable of high speeds, not all USB2.0 devices meet our
expectations as users.

Regards,

aht-ga
Google Answers Researcher

aht, thanks.  I won't be trying the ram disk this week but I will let
you know  the outcome.

Have you confirmed that your USB cable is 2.0-rated? If not, it will
indeed choke your speed down.

Regards,

aht-ga
Google Answers Researcher

grthumongous-ga:

The Intel 845MP on your laptop's motherboard should not be involved in
this case, as the Orange Micro USB 2.0 Cardbus adapter you bought
incorporates an NEC chipset with full internal support for USB 2.0. In
fact, one of the biggest knocks against the Inspiron 8200 when it
first came out was its lack of USB 2.0 support, due to the
845MP/ICH3-M chipset. The ICH3-M is the I/O controller hub that
supports, among other things, the USB 1.1 ports built into your
Inspiron 8200. The Orange Micro Cardbus device you purchased
effectively works in parallel to the 845MP/ICH3-M, enabling you to
have USB 2.0 capabilities in spite of the 845MP/ICH3-M chipset running
your motherboard.

I presume you have already gone through all of the driver confirmation
steps mentioned here:

http://www.orangemicro.com/bin/USB2/OrangeUSBDriverWinXPOM.pdf

especially confirming that you have the Orange drivers installed and
not the Microsoft USB 2.0 drivers?

The challenge with the devices you mention is that they all are
dependent on your hard drives and hard drive controllers because you
are reading/writing data files. The truly high-speed USB 2.0 devices
tend to be video input devices. That said, the bit rates you report
are not very good if everything is supposed to be designed for USB
2.0. USB 2.0 is supposed to support serial communication at three
nominal speeds: 480 Mbps, 12 Mbps, and 1.5 Mbps. A device can be rated
as USB 2.0 compliant if it uses a controller that is certified as USB
2.0, even if the device itself is not capable of making use of the 480
Mbps speed. In a situation like that, the controller will drop down to
12 Mbps to conserve bandwidth.

In your trials, it looks like the best the combination of your PC and
the Travan can do is around 12 Mbps (roughly 1.5 MB/s). The only way
to truly know what the cause is, is to try the Travan unit on another
PC with USB 2.0 support, or eliminate any hard drive overhead by
setting up a RAM disk and reading/writing from there to the Travan.
The information available from Seagate RSS (now called Certance) is
conflicting: while they claim "up to 120 MB/min uncompressed" data
transfer rate, they also say that the transfer rate will be optimized
by FastSense to one of "1000, 733, or 489 kilobytes per second" to
reduce the amount of tape repositioning required. Accounting for
overhead and compression ratios, it is difficult to ascertain how this
FastSense feature will allow them to actually hit the maximum transfer
rate of 120 MB/min uncompressed. More likely, you will see between 60
MB/min to 120 MB/min performance, which is exactly what you observed
(122 MB/1.5 min = 81.33 MB/min).

The only real way to test your PC's USB 2.0 capabilities would be to
find a USB 2.0 video device and seeing how much throughput you can get
when dumping the feed directly into a raw-format file on your
defragmented hard drive.

Regards,

aht-ga
Google Answers Researcher

grthumongous-ga:

Thank you for the extremely generous tip!

Regards,

aht-ga
Google Answers Researcher