Benchmarking usb video capture transfer rates

Hi,

I'm looking for something that will run under windows or linux that
measure exactly the data transfer speed of my external usb video
capture box. I have a plextor m402u and I want to know exactly how
many of them I can attach to one usb channel before the bandwidth is
saturated. I also want to be able to benchmark what the transfer rate
is at various resolutions, compressions, etc.

The tool I use would preferably be gpl'ed under linux, but I'll take a
windows tool as long as it's some kind of free or free for personal/
educational use. ;)

Thanks,

On 11 Mar 2007 13:13:20 -0700, "alfrodull@gmail.com"
<alfrodull@gmail.com> wrote:

>Hi,
>
>I'm looking for something that will run under windows or linux that
>measure exactly the data transfer speed of my external usb video
>capture box. I have a plextor m402u and I want to know exactly how
>many of them I can attach to one usb channel before the bandwidth is
>saturated. I also want to be able to benchmark what the transfer rate
>is at various resolutions, compressions, etc.

It's making MPEG2 streams, right? Just look at the bitrate
(per file, the MPEG2 settings used) of those and pad it by
about 10-15%. It would not be good to try to put multiple
on the same USB hub/port-pair, if you need more devices and
have ran out of hub pairs then add a PCI USB2 card or two.

>
>The tool I use would preferably be gpl'ed under linux, but I'll take a
>windows tool as long as it's some kind of free or free for personal/
>educational use. ;)
>
>Thanks,

A tool that tells you this still won't tell you want the
actual limit is on your specific ports using the data sizes
output by the capture device. It certainly won't be close
to USB's theoretical performance limit.

On Mar 11, 10:49 pm, kony <s...@spam.com> wrote:

> It's making MPEG2 streams, right?

It does mpeg4 plain, Divx or Xvid, but different kinds at different
resolutions and compression settings. Also, because of what we'll be
taping, essentially talking heads, a high compression ratio with a low
audio stream. We'll be recording in a couple of different
resolutions, 640x480 and 320x240. I haven't been able to find the
technical details on the m402u, but I'm assuming that if the software
tells it we want 320x240 at high compression the data stream will be
less than 640x480 at high resolution. Of course, that's an assumption
I'm making. This is a hardware based decoder so I'm assuming that the
hardware is taking care of converting the analog into the right kind
of data stream. If it isn't and it is relying on the software then I
might as well get a cheaper unit. ;)

And the possibility is strong that I simply don't know enough about
how this works behind the scenes. These units are basically intended
to take vcr input and convert it to digital. We're going to be using
them in a more complicated way to tape various classes.



> A tool that tells you this still won't tell you want the
> actual limit is on your specific ports using the data sizes
> output by the capture device. It certainly won't be close
> to USB's theoretical performance limit.

I'm assuming that the theoretical input limit is 480 mbits/sec on each
usb channel. So if I have 4 devices on 4 ports on the same channel,
each pumping data at 100 mbits/sec I won't saturate the channel. In
theory. In practice 3 may be the limit due to overhead, etc. I'd
like to know what the limits are before hand.

Thanks!

On 12 Mar 2007 07:24:27 -0700, "alfrodull@gmail.com"
<alfrodull@gmail.com> wrote:

>On Mar 11, 10:49 pm, kony <s...@spam.com> wrote:

>> A tool that tells you this still won't tell you want the
>> actual limit is on your specific ports using the data sizes
>> output by the capture device. It certainly won't be close
>> to USB's theoretical performance limit.
>
>I'm assuming that the theoretical input limit is 480 mbits/sec on each
>usb channel. So if I have 4 devices on 4 ports on the same channel,
>each pumping data at 100 mbits/sec I won't saturate the channel. In
>theory. In practice 3 may be the limit due to overhead, etc. I'd
>like to know what the limits are before hand.


In practice you may not even be able to use 3.

The best practice is not to try to push any bus to it's
limits, to add hub pairs with addt'l PCI cards.