home *** CD-ROM | disk | FTP | other *** search
- Xref: sparky comp.sys.transputer:1170 comp.os.linux:16127
- Newsgroups: comp.sys.transputer,comp.os.linux
- Path: sparky!uunet!haven.umd.edu!darwin.sura.net!spool.mu.edu!umn.edu!math.fu-berlin.de!Sirius.dfn.de!Urmel.Informatik.RWTH-Aachen.DE!gandalf!michael
- From: michael@gandalf.moria (Michael Haardt)
- Subject: Transputerlink -> Host throughput values
- Message-ID: <9211082963@gandalf.moria>
- Lines: 27
- Sender: news@Urmel.Informatik.RWTH-Aachen.DE (Newsfiles Owner)
- Nntp-Posting-Host: kaa
- Reply-To: u31b3hs@pool.informatik.rwth-aachen.de (Michael Haardt)
- Organization: An old and gray machine, somewhere in Moria.
- Date: Sun, 08 Nov 92 19:52:03 +0100
- Return-Path: <u31b3hs@pool.informatik.rwth-aachen.de>
- Lines: 27
-
- I would like to know some "throughput" values of the transputer-host
- link.
-
- When I use iserver, writing 1k blocks to the host gives me a performance
- of about 20k per second. Using smaller blocks results in very reduced
- performance.
-
- I wrote a small program to measure only transfer speed by reading data
- directly from the link. The maximum throughput was about 110k per
- second. I think the difference results in iserver being a single
- threaded server, so it can't receive new packets while it processes
- packets, which reduces performance a lot.
-
- All values were measured with Linux 0.98.pl1 and GCC 2.2.2d on a 386-20
- without processor cache. My machine has about 4500 dhrystones (at good
- days:). I expect a linear speed win with faster machines, because the
- link bandwidth is miles away from being a bottleneck. Some assembler
- code would help a lot too for sure.
-
- Please respond via e-mail with specification of your hardware, OS,
- performance with 1k blocks using iserver and by directly accessing the
- link over the OpenLink and ReadLink functions. If I get enough results,
- I will post a summary.
-
- Michael
-
- PS: A new release is being prepared, including tas, tlk and tld!!
-