> It’s worth emphasising that 60-bit wasn’t just used for memory addresses. It was used throughout the system, crucially for register length and floating point arithmetic.
A bit the opposite, if I recall correctly. Memory and the "operand" registers (which held floats and/or ints) was 60 bits wide. But addresses were only 18 bits wide, and addressing was done through dedicated "address" and "index" registers which were also only 18 bits wide.
The subsequent machines (eg. 7600) retained 18 bit addressing and had to resort to a bank select scheme to access larger amounts of memory.
See "Design of a Computer, The Control Data 6600" by J E Thornton.
Hi Andrew, Thanks so much for the correction and I think that's a great point. I'm hoping to do a much longer post on the CDC 6600 in the near future so will try to go into a lot more detail to make this all clear. That Thornton book is a great resource!
NYU Courant Institute had a CDC 6600, which I used for a few years. Had an operating system called NOS, which while stood for Network Operating System, we liked to think of it as "No Operating System".
Lets just say that its OS was peculiar, though patently better than IBM mainframe nonsense of the time. Moving from NOS to Unix (Ultrix and later SunOS) was a joy.
HI Jonathan. Fantastic comment. You prompted me to do some googling and I came across this. So many operating systems. Looks like NOS was the evolution of KRONOS the Greek god of time - presumably relating to 'time sharing'
<< In 1965, Singer bought out Friden, setting it up as Singer Business Machines. It then designed a computer, originally called the Business Data Processor (BDP) and soon renamed the System Ten. In 1969, Singer Business Machines created a subsidiary, the Advanced Systems Division, in each Western European country to launch and market the Singer System Ten. Newly appointed Managers and Directors were trained in the technology and the marketing strategy, and the Singer System Ten was launched throughout Europe on April 2, 1970. >> (unsourced statements)
Great article!! BTW, both the CDC 6600 image and the Stretch document are from the permanent collection of the Computer History Museum in sunny Mountain View, California. Come on by and check out the Supercomputing gallery in our signature exhibit, "Revolution: The First 2,000 Years of Computing."
DEC 36-bit computers are mainframes not minicomputers. It's some weird myth that they are minicomputers. DEC published product literature that states those are mainframes. I guess their 32-bit VAX minicomputers were more visible due to higher volume sales so only L337 notice the difference.
> It’s worth emphasising that 60-bit wasn’t just used for memory addresses. It was used throughout the system, crucially for register length and floating point arithmetic.
A bit the opposite, if I recall correctly. Memory and the "operand" registers (which held floats and/or ints) was 60 bits wide. But addresses were only 18 bits wide, and addressing was done through dedicated "address" and "index" registers which were also only 18 bits wide.
The subsequent machines (eg. 7600) retained 18 bit addressing and had to resort to a bank select scheme to access larger amounts of memory.
See "Design of a Computer, The Control Data 6600" by J E Thornton.
Hi Andrew, Thanks so much for the correction and I think that's a great point. I'm hoping to do a much longer post on the CDC 6600 in the near future so will try to go into a lot more detail to make this all clear. That Thornton book is a great resource!
NYU Courant Institute had a CDC 6600, which I used for a few years. Had an operating system called NOS, which while stood for Network Operating System, we liked to think of it as "No Operating System".
Lets just say that its OS was peculiar, though patently better than IBM mainframe nonsense of the time. Moving from NOS to Unix (Ultrix and later SunOS) was a joy.
HI Jonathan. Fantastic comment. You prompted me to do some googling and I came across this. So many operating systems. Looks like NOS was the evolution of KRONOS the Greek god of time - presumably relating to 'time sharing'
http://bitsavers.trailing-edge.com/pdf/cdc/cyber/CDC_Operating_System_History_Mar76.pdf
I think I found another 60-bit computer. It was manufactured by Singer ("the sewing machine company") which had undergone a lot of diversification.
http://members.iinet.net.au/~daveb/S10/Sys-10.html
"The S-10 had a byte (or "character") length of 6 bits..." -- this is really BCD being described in terms of ASCII and HEX (instead of octal).
"All S-10 instructions were 10 characters long, aligned on a 10-character boundary." -- 6-bit chars times 10 chars long is 60 bits.
Thanks. That's a great find. The ISA looks really interesting. I'll definitely mention this in a future post.
It also looks like a similar architecture was adopted in the ICL System 25 when ICL took over Singer's international operations.
<< In 1965, Singer bought out Friden, setting it up as Singer Business Machines. It then designed a computer, originally called the Business Data Processor (BDP) and soon renamed the System Ten. In 1969, Singer Business Machines created a subsidiary, the Advanced Systems Division, in each Western European country to launch and market the Singer System Ten. Newly appointed Managers and Directors were trained in the technology and the marketing strategy, and the Singer System Ten was launched throughout Europe on April 2, 1970. >> (unsourced statements)
https://en.wikipedia.org/wiki/Singer_System_10
Great article!! BTW, both the CDC 6600 image and the Stretch document are from the permanent collection of the Computer History Museum in sunny Mountain View, California. Come on by and check out the Supercomputing gallery in our signature exhibit, "Revolution: The First 2,000 Years of Computing."
DEC 36-bit computers are mainframes not minicomputers. It's some weird myth that they are minicomputers. DEC published product literature that states those are mainframes. I guess their 32-bit VAX minicomputers were more visible due to higher volume sales so only L337 notice the difference.
https://www.hactrn.net/sra/alice/
Thanks. My wording was unintentionally incorrect - now fixed.