Peregrine,
> 2) Also, if anyone has experience with VME64 systems I'd like to know
> what the _real_ measured transfer rates are.
Under VxWorks I typically get a little over 30MByte/s average throughput
in a test in which a largish message (few Mbytes) gets ping-ponged between
2 CPUs using D64 block transfers. This includes overheads associated with
signalling between the CPUs; you could do significantly better if a slave
were just swallowing D64 blocks. Also, my application uses a device driver
to manage the transfers and this luxury degrades raw VME throughput
further. Nevertheless, if you intend to transfer rather small blocks
and/or have much else contending for the bus I wouldn't budget for _very_
much more throughput.
(These results were with MV167<-->Baja4700 transfers.)
Regards,
Nick Dillon.
------------------------------------------------------------------------
Royal Greenwich Observatory Tel: +44 1223 374740 (direct)
Madingley Road Tel: +44 1223 374000 (s/board)
Cambridge CB3 OEZ Fax: +44 1223 374700
England E-mail: [email protected]
------------------------------------------------------------------------
- References:
- PowerPC/VME64 Peregrine McGehee
- Navigate by Date:
- Prev:
PowerPC/VME64 Peregrine McGehee
- Next:
Re: tcl/tk , tcl-dp and blt which and where? Pete Jemian
- Index:
1994
1995
1996
<1997>
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
- Navigate by Thread:
- Prev:
PowerPC/VME64 Peregrine McGehee
- Next:
Re: PowerPC/VME64 Peregrine McGehee
- Index:
1994
1995
1996
<1997>
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
|