Skip site navigation (1)Skip section navigation (2)
Date:      Tue, 06 Jan 1998 10:04:14 -0500
From:      Brian McGovern <bmcgover@cisco.com>
To:        questions@freebsd.org
Cc:        dtd-miele-staff@cisco.com
Subject:   PPP degrades with file type on NULL modem?
Message-ID:  <199801061504.KAA00368@bmcgover-pc.cisco.com>

next in thread | raw e-mail | index | archive | help
I was just running some tests on pppd here, between two Pentium Pro 200s 
connected via NULL-modem cable (via 16550).

I ran pppd with -bsdcomp, crtscts, at 115200 baud.

Once the ppp link was up, I FTP'ed some files across the link. The results
I found were:

File		Size		Created By		Contains	K/s

minicomp.cap	5819576		Com. Prog. Capture 	Text		10.72
tcpip.flp	1474560		dd			Disk Image	 9.71
termcap.db	1212416		FreeBSD dist.		Termcap DB	 7.66

This, to me, seems a bit odd, but I've run the test several times. Other, more
realworld tests (involving modems, different platforms - including some
test scenarios that did not involve FreeBSD at all- , etc) have yielded
results from 4.5K/s to as high as 11.03K/s.

I firmly believe things like filesize and timing have little to do with it,
as I've halfed, doubled, and tripled the sizes of all the files above (with
EXACT results), run the files from MFS as well as the local disks (again,
same results), made sure the systems were similarly busy, and made sure
nothing was swapping.

Anyone care explain to me why, when compression should not be involved, that
I can't get throughput that collates with bps rate (ie - if I can get 10.72
at 115200, why can't the other two files even come close?)? I just find
it an oddity I'd like to have an explanation for :)


	-Brian "Gotta Be in the Protocol" McGovern



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199801061504.KAA00368>