Hi all,

I've been recently working getting the mscv6 port to work so that I can tryout 
patches at home where I have no embedded hardware to test.

Getting the current code to run was not that hard. I have added an OS layer to 
be able to test socket api based applications.

I have implemented a TCP server test which waits for a client to send some test 
parameters. After receiving the parameters (mainly number of packets to send), 
the server sends X 'packets' with 1460 bytes data each (with 'packets' I mean 
it calls lwip_send X times with 1460 bytes each time).

Now, I have the following question: On my embedded system (altera NIOS running 
at 60 MHz), I achieve lwIP send rate of about 2MByte/s. However, on my PC (1.3 
GHz Windows XP SP2), I only get ~4-8 packets per second!!! (While processor is 
always around 100%)

And no, the socket interface it not the bottleneck here. I've had a look at the 
processor consumation of each thread: tcpip_thread is the highest with about 60 
%, then comes the input thread (winpcap->pbufs) with about 40%, after that the 
socket thread with nearly 0% (it does not do really much for a 1.3 GHz PC, 
anyway...)

Does anyone have experience with the winpcap network driver? Did I make some 
mistake here? Or is there any problem with that XP service pack network 
limitations I heard of somewhere?

Simon
_______________________________________________
lwip-users mailing list
[email protected]
http://lists.nongnu.org/mailman/listinfo/lwip-users

Reply via email to