🚀 go-pugleaf

RetroBBS NetNews Server

Inspired by RockSolid Light RIP Retro Guy

Thread View: comp.arch.bus.vmebus
3 messages
3 total messages Started by Harry Langenbacher Sat, 21 Nov 1992 00:18
vme bus i/o speed question
#26
Author: Harry Langenbacher
Date: Sat, 21 Nov 1992 00:18
23 lines
1477 bytes
I have to build a "neural-net co-processor" board that will
plug into a sun 3/160 VME back-plane. My consultant says that
the VME bus i/o rates will be slower than on an 8mhz ISA (pc-at) bus
(which is ~8 Mb/sec), because the sun 3/160 won't support burst mode
VME transfers.

How fast can I transfer data w/o burst mode on the VME bus ?
(Obviously I'm looking for ballpark figures.)

What computers (esp. sun) can do VME i/o faster (burst mode?) ?

I built a VME card before, and it worked fine, but I was never
concerned about the data throughput. I switched to ISA bus cards
because it seemed simpler, and because PC-clones are so cheap and
everyone has one, and proto-cards are so easy to get, but now my
sponsor wants VME cards, so ...

	Thanks for reading,
			Harry
-- 
Harry Langenbacher 818-354-9513 harry%neuron6@jpl-mil.jpl.nasa.gov
Concurrent Processing Devices Group, FAX 818-393-4540
JPL,  M/S 302-231, 4800 Oak Grove Dr, Pasadena CA 91109        USA
Re: vme bus i/o speed question
#52
Author: Adams
Date: Mon, 23 Nov 1992 02:35
18 lines
1455 bytes
In article <1992Nov21.001825.18644@jato.jpl.nasa.gov> harry@neuron6.jpl.nasa.gov (Harry Langenbacher) writes:

>   I have to build a "neural-net co-processor" board that will
>   plug into a sun 3/160 VME back-plane. My consultant says that
>   the VME bus i/o rates will be slower than on an 8mhz ISA (pc-at) bus
>   (which is ~8 Mb/sec), because the sun 3/160 won't support burst mode
>   VME transfers.

SUN3/xx are perfect capable of running 32/32 cycles. 8Mbytes/s would
require only 500ns per cycle. This is not too fast at all. 

>   How fast can I transfer data w/o burst mode on the VME bus ?
>   (Obviously I'm looking for ballpark figures.)

Depends very much on your peripherals, some disk controllers are
running at 20Mbytes/s, using burst mode.

best, adams
Re: vme bus i/o speed question
#63
Author: HD Associates
Date: Mon, 23 Nov 1992 16:01
21 lines
1294 bytes
In article <1992Nov21.001825.18644@jato.jpl.nasa.gov> harry@neuron6.jpl.nasa.gov
 (Harry Langenbacher) writes:

>   I have to build a "neural-net co-processor" board that will
>   plug into a sun 3/160 VME back-plne. My consultant says that
>   the VME bus i/o rates will be slower than on an 8mhz ISA (pc-at) bus
>  (which is ~8 Mb/sec), because the sun 3/160 won't support burst mode
>   VME transfers.


Burst mode isn't the major factor.

The Alliant FX/2800 can maintain an average 25 MB/s into memory*
without burst mode.  There is no advantage to burst mode on that
particular system.  I've been told by the designer that you can decode
the addresses quickly enough with new PALS to eliminate the burst mode
advantage.

*With this incantation:
Ironics Full Speed Data Transporter as data source; large buffers; 32
bit transfers; consecutive transfers.
Thread Navigation

This is a paginated view of messages in the thread with full content displayed inline.

Messages are displayed in chronological order, with the original post highlighted in green.

Use pagination controls to navigate through all messages in large threads.

Back to All Threads