TEAM-ADA Archives

Team Ada: Ada Programming Language Advocacy


Options: Use Forum View

Use Proportional Font
Show Text Part by Default
Show All Mail Headers

Message: [<< First] [< Prev] [Next >] [Last >>]
Topic: [<< First] [< Prev] [Next >] [Last >>]
Author: [<< First] [< Prev] [Next >] [Last >>]

Print Reply
Simon Wright <[log in to unmask]>
Reply To:
Simon Wright <[log in to unmask]>
Fri, 14 Feb 2003 20:45:13 GMT
text/plain (22 lines)
On 13/02, james hopper wrote:
| Both os x and linux.  what i was doing was passing real time radar
| images from our sim code on the server to the display app on another
| computer.  the difference was quite dramatic.  i could get 60 hz with
| AdaSockets, but only about 10hz with gnat sockets.   i was in a bit of
| a rush and never really went into what was causing the difference.

I compared transferring 1024 bytes in a round-trip using UDP over a
quiet 100BT switched LAN between two PowerPC cards running VxWorks.
The stack was out of the box, no tuning.

GNAT (3.15a1) Sockets: 1050 us
C:                     1008 us

Not surprising really, no buffering in this case and the GNAT code is
a pretty thin skin over the OS.

What was fun was working out why the C compiler wouldn't accept the
asm() statement to read the timebase register; I didn't spot the -ansi
in the Tornado command line that means you have to say
__asm__(). Foo. Much less grief with GNAT.