Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP Path: utzoo!watmath!clyde!burl!ulysses!allegra!mit-eddie!think!harvard!seismo!mcvax!enea!ttds!duvan!ahi From: ahi@duvan.UUCP (Anders Hillbo) Newsgroups: net.lan Subject: Problems with IP/UDP speed on Vax750/Unix4.2 Message-ID: <776@duvan.UUCP> Date: Fri, 7-Feb-86 23:03:54 EST Article-I.D.: duvan.776 Posted: Fri Feb 7 23:03:54 1986 Date-Received: Tue, 11-Feb-86 06:08:41 EST Reply-To: ahi@duvan.UUCP (Anders Hillbo) Organization: The Royal Inst. of Techn., Stockholm Lines: 35 We want to connect a VAX 750, running UNIX BSD 4.2, with an Interlan controller and a custom made display station (Intel iSBC 186/51 & 82586 based) to transfer image data. We decided to use Ethernet and an UDP/IP implementation on the displaystation to be able to run in user mode on the Vax. The problem is SPEED (on the 750). A typical image consists of something between 64 kbyte and 1 Mbyte of data and the time to transfer the picture is how long you wait for the picture to show on the screen. The microprocessor based display station is just now capable of sending data in up to 300 kbyte/s , while the VAX computer only send data in 33 kbyte/s (UDP/IP, 1024 byte buffers, one user on the system). If raw IP is used (super user alone on the system), the speed increases to 182 kbyte/s. Also when sending TO the Vax the interface or something seems to drop packets if 4 of them arrive immediatly after each other. (I thought Interlan could handle back-to-back packets). We know that the packets leave the sender correctly by spying on the ether from a another displaystation which of course receives all packets with out any problem. The goal is a transfer rate at about 100 kbyte/s. The questions: . Why cant the UDP-interface send packets in a faster speed? Maybe if you could use bigger packets? . Is the limitation in the Vax750 or the interface? . Is this a well known problem? Anyone else with similar problems? . What are the real rules for trailer encapsulation. 4.2 only seems to do it when the package is 512 or 1024 bytes. Anders Hillbo CS Dept, Royal Inst of Tech. Stockholm, Sweden