Simon, your remark about netif/winpcap seems to be right. For lwIP stack itself (using NETCONN on WIN32)it doesn't take too much to prepare the ethernet packet before the pcap_sendpacket is being called from pcapif.cThis time is about 1sec on my computer for 100Mbyte of data. So, it seems like most of the time is being spend inside pcapif.c low_level_output with the call to pcap_sendpacket.Windows WinSock is taking 2.5 sec to prepare and send the same amount of data, so lwIP's time of 1sec for packet preparation only seems to be adequate as opposite to total 25sec for total preparation and send with pcap. Seems a bit ugly, since pcap is used in many applications and should be really optimized for the performance, but may be it is more optimized for the receiving(capturing) rather sending. Anyway, tomorrow i will try to send the same load through the pcap directly (using this same pcap_sendpacket function) and measure the time. Did somebody use a UDP on any hardware using NETCONN API (zero copy, multithreaded environment) ?What was your maximum throughput ? What is the characteristic of your system ? Thanks,Stas
_______________________________________________ lwip-users mailing list lwip-users@nongnu.org https://lists.nongnu.org/mailman/listinfo/lwip-users
