Talk:Corvus Systems

Concept and Sun-1
This isn't to beat up on either Corvus or Sun, but I thought it'd be interesting to compare the two systems. There were a lot of 68K based workstations springing up around then, so who knows? Maybe the right push at the right time could've gotten Corvus a bigger foothold.

It looks like the main differences are that the Concept was significantly less expensive, but had less expansion capabilities. Lack of UNIX at the time of its release would be another limiter, which is why I'd be very interested in knowing when a UNIX release was made available, and what that release was.

But the biggest difference is arguably graphics. The Sun-1 had over twice the display real estate and a larger monitor. The Concept's rotating display is extremely cool (the screen I'm using right now is a Samsung WSXGA+ monitor that I bought in part because it has that functionality; pity Windows doesn't handle the rotation just a bit more cleanly), but seems to have been designed for text-processing rather than raster operations. I suspect the Sun's underlying hardware raster op support would have given it a significant performance advantage as well.

--NapoliRoma (talk) 22:39, 19 December 2007 (UTC)


 * http://marc.info/?l=classiccmp&m=113190209610767&w=2 sez

""Corvus took the basic Concept hardware and added a SUN-style MMU to the memory board and tried to sell it with Unisoft Unix as a first try at a Unix computer. This machine didn't have a bitmapped display."69.125.110.223 (talk) 16:18, 26 December 2007 (UTC)
 * Interesting. So this was apparently something they OEM'ed to Valid Logic Systems, which sold it as the SCALDstation.--NapoliRoma (talk) 02:01, 28 December 2007 (UTC)
 * Um, no, the SCALDstation wasn't OEM'ed from anybody. Valid designed their own Multibus CPU cards, memory cards, and graphics cards.  I know this because I worked for Valid's founders in their hardware group.  Some things like peripheral control cards were purchased.  — Preceding unsigned comment added by 98.234.218.45 (talk) 03:16, 10 August 2013 (UTC)

The SCALDsystem IV was a Corvus machine, codenamed "Cheetah" which was also sold by Honeywell. I have the prototype. It was designed by Mike Cook and came out several years after the original Multibus SCALDstation as a cost-reduced alternative to the original Valid-designed system. 50.204.185.162 (talk) 16:14, 23 September 2014 (UTC)

64 devices, and the device address was set with a 5-bit DIP switch
how came? 2^5=32 not 64 — Preceding unsigned comment added by 85.90.116.114 (talk) 18:10, 22 April 2019 (UTC)


 * Yes, 2^5=32, but that woudl be the 'on' value of DIP switch 5. If all DIPs are set, then you have 32+16+8+2+1=63, plus the position where all DIPs are off = 0, so a total of 64 possible combinations. 2A00:23C4:9420:6501:5937:419F:F66B:561E (talk) 11:47, 6 March 2023 (UTC)

Grenada war and Corvus Networking
I was a Manager/salesperson for the Byte Shop in Seattle in 83. The Army was using Apple II computers and a Corvus network on the battlefield (as much as it was a battlefield) during the invasion of Grenada. They had a semi that had the computers all connected on the network and all they had to do was roll it out on the battlefield and they have an instant network. During the Grenada war the only reliable communications they had with Washington for a while was through this network. Some Major/Clonel told me that as he was explaining the problem they were having.

I was involved because I regularly sold them equipment including most of the Corvus equipment and they were having problems with network cards getting fried on the battlefield. I gave them some ideas as I didn't have access to their equipment. We eventually found out the generator wasn't grounded well and was causing the Corvus cards to blow.

Additional history of Corvus products and their Design Features
I’m including this information in the interest of starting a discussion about what here can be included given Wikipedia documentation requirements. Given these were among the earliest shared network storage devices and were compatible with both early Macs and PCs, we should attempt to preserve some of the knowledge of the system details and limits that were part of practical implementations and future product designs.

The Corvus twisted pair network was originally illustrated as a simple linear arrangement (bus) with fairly short leads (8-12 feet) for each PC tapped off the main twisted pair. The taps were very simple 2-part plastic housings about 3” by 1.5” that had hole on each end and contained two “v” blade penetrating connectors that were pressed onto the main twisted pair. A miniature headphone jack was used for connecting to the lead. Because of signal strength and propagation timing, the length of the main twisted pair was supposedly limited to something around 300’. However, with a little experimentation, it became apparent that this layout could be altered into a branched or star layout. For example, it was possible to add several 100’ lines, tapped from the main twisted pair near the midpoint of its length, thus still keeping any two points less that 300’ apart.

The Corvus network drives were technically disk servers rather than file servers. Disk servers are very simple but have obvious disadvantages. (I am not aware of any consumer-oriented disk servers after the Corvus product line.) With disk servers, access to the network and a particular shared drive was established when a program was run on the PC that included a login sequence. However, actual logical access was mostly based on receipt of a partition table at the time of login. Note that if a drive was being shared between multiple departments, setting up multiple partitions was the standard means of providing access control. However, this design meant that if an installation included two or more Corvus network drives, as long as the partition tables were consistent between the drives, it was possible to simply unplug the network cable supporting one drive and plugin a different cable from a different drive, and have access to all the aligned partitions. Given how small such driver were (typically around 30Mb), the ability to change between networks without a new login step was both a feature and a bug. It was thus possible to automate this physical layer switching with latching relays controlled by RS-232 output from a PC and thus create automated access and/or synchronization of files between Corvus drives, queued central print serving, etc.

These work-arounds reflect an important aspect of the early history of desktop networking in that when there was a practical need to go beyond a product’s limited functionality or intended scale, it was often possible to leverage other design limitations to improvise additional features. Zatsugaku (talk) 13:23, 23 October 2023 (UTC)