Hi,
The original question reminds me that I suffer from a following
problem which possibly could be solved by the shared memory mechanism
as implemented in recent RTL/RTAI.
I have some Linux machines (2*PIII/600 each) with Gigabit ethernet
connected via a Gigabit switch. The problem is that I get max. 35 MB/s
data transfer between two machines if I write to /dev/null. If I write
to ramdisk I get only max. 24 MB/s.
If possible I would like to get more (then 50 MB/s ?).
It is perfectly visible that the limitation is the receiving machine.
I am thinking about the following : instead of using the ramdisk I
could use shared memory visible directly from both the user's space
and the kernel's space. The problem is - can I somehow convince the
Gigabit ethernet driver to write directly into this shared memory
(from kernel's space) ? So that any unnecessary moving of data is
shortcut and I get more throughput ?
(Another problem is - can I use this shared memory module, by Tomasz
Motylewski, with a standard kernel 2.2.x, without any real time
extensions ?)
Thanks in advance,
Jacek.
-- [rtl] ---
To unsubscribe:
echo "unsubscribe rtl" | mail [EMAIL PROTECTED] OR
echo "unsubscribe rtl <Your_email>" | mail [EMAIL PROTECTED]
---
For more information on Real-Time Linux see:
http://www.rtlinux.org/rtlinux/
