| Title: | The Digital way of working |
| Moderator: | QUARK::LIONEL ON |
| Created: | Fri Feb 14 1986 |
| Last Modified: | Fri Jun 06 1997 |
| Last Successful Update: | Fri Jun 06 1997 |
| Number of topics: | 5321 |
| Total number of notes: | 139771 |
Seeking help from someone to identify the computer horsepower needed
for complex multi. interactive aircraft simulation project that is
proposed. Hope to use Alpha based Multi Processor Systems in a Series
or Parallel mode.
| T.R | Title | User | Personal Name | Date | Lines |
|---|---|---|---|---|---|
| 4156.1 | A rough cut | ODIXIE::GARAVANO | Thu Sep 28 1995 15:27 | 27 | |
there are several way to apporach this - but they all require the
answer to the following:
Software: Are you looking for COTS? if so there is some simulation
software that helps start the process - but it needs to be virtual reality
not just sim or the coordinates will not be in realtime - Ian Ellery @ORO
can help you with the VR software availablle on Alpha
Hardware: Is rack mounting a requirement? depending on the need to keep
the simulation results and any and all manipulation etc - the following
would be my cut:
At least two 2100 5/250, 8200 or 8400 5/300 systems with 512 to 1
gigabyte of memory to start. Make it an ASE cluster - Use dual HSJ40
with KZPSA-BB (FWD) (thats the fastest available I/O option other than
IPI) - use rz29's in either a sw800 or sw500 - for Display use the
Evans Sutherland boards PCI based with VRC21-PA for the monitor. The
only other option would be a VR headset -
If reception of incoming realtime data a part of the equation - use the
upcomning VME 64 PCI interface from CSS - that will provide the
bandwidth to handle large data streams
Hope this helps
| |||||
| 4156.2 | RLTIME::COOK | Thu Sep 28 1995 18:29 | 31 | ||
> If reception of incoming realtime data a part of the equation - use the
> upcomning VME 64 PCI interface from CSS - that will provide the
> bandwidth to handle large data streams
Are you assuming this is a legacy simulation? I didn't read that into .0.
If there is VME in this system, you may want to consider the AXPvme 230 with
and FDDI PMC I/O card. This could give you quite a bit of front end number
crunching with good band width back to the main simulation systems. Or, for
a more traditional approach, racks of AXPvme 230s with VMIC reflective memory
PMC cards...just don't call me to debug the simulation. :-)
Lots of ways to go, without a lot of detail in .0. As in, I'm assuming this
is a training simulator and not "hardware in the loop" systems simluation. Do
you need freeze data and roll back? Reflective memory helps here since
simulation data can be kept in reflected pages and frame by frame data capture
can be done.
Anyway, every simulation heavy company I've worked with Mickey D, Boing,
Lockhead, they all had their own set of simulation benchmarks to determine CPU
suitability. If your working with one of the majors, you may want to talk to
them about this.
Good luck,
Al Cook
| |||||
| 4156.3 | Engage CSS directly! | NCMAIL::SCHOLZ | Mon Oct 02 1995 10:54 | 6 | |
Engage CSS directly. They have this expertise. Launch a call into
Mark Woodbury at 264-0850. They also have complementary technologies
that handle continuous I/O into a rack of Alpha's.
Regards,
Steve
| |||||