See below ... :-)
> -----Original Message-----
> From: Clarence Verge [SMTP:[EMAIL PROTECTED]]
> Sent: Friday, 11 August 2000 12:00
> To: [EMAIL PROTECTED]
> Subject: Re: Comm's performance & FIFO settings & epppd(d)
>
> Da Silva, Joe wrote:
> >
> > After trying various things, I have found that setting the FIFO receive
> > interrupt threshold to 1 (epppd sets this to 8, according to the doc's),
> > comm's works like a charm!!! Nice and fast, with packet driver stat's
> > reporting zero errors and zero dropped packets (on bad pages, I was
> > previously getting about 10 receive errors and 18 dropped packets,
> > when loading a 20-25K page).
> >
> > So, can anybody think of a *good* reason why the FIFO receive
> > interrupt threshold should be set to 8??? I know this theoretically
> > reduces the comm's overhead a bit, but in practice, this seems to
> > be the cause of poor comm's performance (I know I am not the
> > only one to have this trouble!).
>
> What speed is the computer this 9600 modem is on ?
> I think a 1 or 2 byte threshold at 9600 baud is a lot more appropriate
> with
> a reasonable computer, but would it also waste a lot of computer power if
> you have either a faster modem or a very slow computer.
>
[da Silva, Joe]
Various 486 machines (typically 486-66).
Just to clarify baud rate : 9600bps is more or less satisfactory as
is,
however anything *higher* is very problematic with the ISP's pages
and some other local web sites (very fast response from ISP). So,
this proposal is specifically aimed at getting faster modems to work
suitably with Arachne.
I agree that slow computers may notice the extra comm's overhead
more, however these computers are even *more* likely to need this
FIFO change in order to work at > 9600bps. Also, BTW, the next
FIFO threshold value after 1 is 4 (which works much better than 8
at > 9600bps but still has a bit of trouble ...).
> > I understand than Bernie is rewriting "epppd(d)" to reduce it's memory
> > needs (just out of curiosity, how is this going to be achieved?). So,
> > I would like to propose that this change to the FIFO receive interrupt
> > threshold also be implemented. Any thoughts on this ... ?
>
> As long as it is user configurable, why not ?
>
[da Silva, Joe]
Sure, user configurable would be ideal. ;-)
> - Clarence Verge
> --
> - Help stamp out FATWARE. As a start visit: http://home.arachne.cz/
> --
>
Best regards,
Joe.
