Re: TI-H: special component
[Prev][Next][Index][Thread]
Re: TI-H: special component
Don't you think that the chip would however, get backlogged with data after
a while? i mean...it could in theory bring in the data and send it back out
that quickly, however if it's planning on doing anything real with the
data, i don't think that this will work at all well. however, if you can
find a way to make it work on a 10 Mhz network. Please tell me. I've been
considdering building a little device that connects to a modem and to my
dad's office network and allows me to dial in and execute various tasks
(i.e. reboot my servers there, connect via telnet to some of the firewalled
workstations, that kind of thing).
At 08:33 AM 6/18/98 -0700, you wrote:
>
>On Thu, 18 Jun 1998, Jon Olson wrote:
>
>> Seeing as the PIC only operates at 10 Mhz and the network operates at 100
>
>Just a minor technicality, but I think the high-end PICs max at 20MHz..
>
>They take 4 clock cycles to make an instruction cycle, so it's getting 5
>MIPS. A 10 Mb/s (megabit) network (in theory) pushes 1.25 MB/s (megabyte)
>of data. So if you had some sort of a shift register taking the data from
>the network line and presenting it a byte at a time, then I think a 20
>MHz PIC would be able to switch 10 Mb/s network traffic. (Did I calculate
>all that correctly?) A 100 Mb/s network (also in theory) moves 12.5 MB/s
>down the line. Since a PIC can read its I/O ports in a single instruction
>cycle, it can take in 5 MB/s at 5 MIPS. So it has enough room left over to
>push 10 Mb/s traffic back out another port, but it doesn't have enough
>power to even read in a 100 Mb/s source without losing frames.
>
>So anyway, like Jon said, 100 Mb/s traffic is definitely out. But I think
>10 Mb/s might be possible... unless I don't have a clue what I'm doing,
>and that is also possible.
>
>--
>Greg Hill
>greg-hill@bigfoot.com
>www.comports.com/link
>
>
Follow-Ups:
References: