Artem Kachitchkine wrote:
>
>> There is probably a simple explanation for the low throughput (says
>> he, somewhat hopefully) and the best way to check
>> this out would be to eliminate the WWAN network altogether and talk
>> directly to an echo server on the USB serial link.
>> I was hoping to do VoIP on 3G. Guess that will have to wait until the
>> driver is sorted out.
>
> Someone from the Sun's USB group (or a member of the larger
> opensolaris community, says he, hopelessly) needs to look into this,
> yes. When I was working on the usbsacm prototype several years ago
> with a loaned Novatel V620 card on Verizon 1xEV-DO, I tried some
> broadband speed tests and was getting from 700Kbps to 1Mbps depending
> on signal quality. Same card on the same laptop produced up to 1.5Mbps
> on Windows. But that was then, when skies were higher and bytes were
> bigger :)
>
> -Artem
>
Yes, the Good Old Days ! Funny thing is they are always better now than
they were then :)
I should be able to get an echo server going on a USB client to test
this out, but time to do low-level hardware development
is a luxury I no longer have, I am afraid.
But come to think of it, doing the same 3G throughput tests on Windows
and Linux would give some sort of baseline.
And not trying to denigrate Windows, but my impression is that their
scheduler would allow the communication application
to hammer the CPU to the exclusion of everything else, hence the better
throughput.
P.A.
