Hm not sure if you can see my previous thread messages, I did some little analysis:
ICS: ~23s, 377 x 8KB, 55-60ms between packets: http://pastebin.com/Uq5mphUH ICS @ 64KB: ~22s, 48 x 64KB, 450-470ms between packets: http://pastebin.com/wTat71Qe SYN: ~6s, 49 x 64KB, 110-120ms between packets: http://pastebin.com/g5svt11P Indy: ~9s, 95 x 32KB, 70-80ms between packets: http://pastebin.com/Dt2PydTR WinInet: ~6s, 190 x ~16KB, 5-55ms between packets, http://pastebin.com/jZLvua5u then tried to log timings through various stages of the post, albeit not being that helpful, I think the main issue lies on OverbyteIcsHttpProt around while FState <> httpReady do begin > {$IFDEF MSWINDOWS} > if MsgWaitForMultipleObjects(0, Pointer(nil)^, FALSE, 1000, > QS_ALLINPUT) = WAIT_OBJECT_0 then > {$ENDIF} > MessagePump; > I'll check on wireshark. On Mon, Apr 18, 2016 at 4:44 PM, RTT <[email protected]> wrote: > On 18/04/2016 15:15, brian - wrote: > >> I know ICS is made for free by volunteers, I'm not expecting/demanding a >> fix, but I think this is an issue that we could look into to try to find a >> solution/find the core problem together. >> > > To start, you could compare the characteristics of the packets being sent > (e.g. using Wireshark), and the times involved, so we can get a general > idea of what is being done differently in these HTTP components. Maybe this > general analyzes offers the clues for what code needs to be > optimized/profiled. > > -- > To unsubscribe or change your settings for TWSocket mailing list > please goto http://lists.elists.org/cgi-bin/mailman/listinfo/twsocket > Visit our website at http://www.overbyte.be > -- To unsubscribe or change your settings for TWSocket mailing list please goto http://lists.elists.org/cgi-bin/mailman/listinfo/twsocket Visit our website at http://www.overbyte.be
