In message <20010502160125.A1339@noc.untraceable.net>, Andrew Brown writes:
>>Unless you got `silo overflow' or `ibuf flood' messages, that's not the
>>problem, or the hardware isn't reporting overflows.  It also seems a
>>little more suspicious since I *know* that we can keep up with a UART
>>running at 115200 baud, and Ricochet maxes at not much more than that.
>
>when my ricochet is plugged into my serial port (db9), yes, it maxes
>at that.  well...i have pppd opening the port at 115200, so maybe
>that's it.
>
>then i plug my ricochet into my udb port, i get much more than that.
>at least twice as much more (~25 kilobytes/sec reported from ftp,
>whatever that works out to), although the modem always reports CONNECT
>460800 when using usb.
>
>steve's ricochet device is a pccard, so i'm not sure what that would
>be affected by, although i sort of assumed it would have its own uart
>on the card that would be capable of really high data rates.  steve?
>what speed does ftp tell you you're getting when you use it?

I don't have enough experience yet to say.  I've only had two 
opportunities to use it, and only got 40K bps and 80K bps, in very 
brief tests.

A Linux-oriented Web page says that the card has a 576 byte buffer 
(i.e., one packet), which the host will need to pick up more or less 
all at once.  The card appears to the kernel to be a 16550A.

All I know for sure is that this morning, with a large ring buffer, I 
didn't get any "ibuf flood" messages; yesterday, with the default 
buffer, I did.

On the other hand, yesterday I got a higher data rate than I did 
today...
>
>i've also heard from several sources that ricochet will be upgrading
>their network from 128k to 256k sometime in the next year or so.
>
I've heard the same thing.

		--Steve Bellovin, http://www.research.att.com/~smb