This is Beagle-Ears
-> Lars
-> Professional
-> Computer History
-> 360 Models
IBM 360/370/3090/390 Model Numbers
- 360/20 - the smallest S/360 was not really a compatible member
of the family. It had only half the registers,
and the instruction set was a subset that was
not binary-compatible.
I seem to remember that it had the capacitive punched-card
microprogram storage, so that a 1401 emulation instruction set
could be field installed.
- 360/25 - what the model 20 should have been. A relative
latecomer to the family. Also had writable control store.
- 360/30 - the low-end workhorse, running DOS
(or TOS - Tape Operating System).
- 360/40 - the midrange workhorse. This may have been the most popular
machine of the series.
- 360/44 - the oddest model. It could be described as a 40 with
a hardware floating point processor and faster memory.
Had a variable precision floating point unit that could
operate on *4, *5, *6, *7, and *8 byte operands.
A rotary switch on the front panel could select between
2 different floating point formats.
It had only 1/2 word and 1 word instructions and could
therefore use a one word memory width without any speed
penalty. Due to the odd instruction set, it had its own
operating system, PS/44.
- 360/50
- 360/65
- 360/67 - the precursor to the S/370 family: The only 360
model with virtual memory support. Was also available
in a dual-processor version.
- 360/75 - the fastest of the original 360s. Had the entire
S/360 instruction set implemented in hardwired logic
(all the others had microcode). Ran at one MIPS.
- 360/85 - 1969 - the first production machine with cache memory.
Other 360 novelties in this model were extended-precision
(16-byte) floating point, relaxation of some instruction
operand alignment requirements,
and an optional I/O channel which allowed multiple disk
channel programs to run (sort of) concurrently.
These later became standard in 370's.
Writable control store?
- 360/91 - the first pipelined processor. Fully hardwired.
Most of the decimal instructions were missing
(emulated in software).
Because it had multiple
instruction execution units, it had "imprecise interrupts".
When an interrupt or an exception occurred, the program
counter might not point to the failing instruction if
the multiple execution units were all active at the
same time. For this reason, it was advisable to put
NOPs around instructions that might lead to exceptions.
Not many of these were built, and they may all have had
slightly different tweaks as each of them was successively
hand built by the same engineering team.
- 360/95 was a 91 equipped with a higher-performance thin-film
memory instead of core. Only a couple were built.
- 360/195 was a faster successor to the 91, again fully-hardwired.
It came in both 360 and (later) 370 versions.
The 370 version had the new 370 instructions plus the
370 TOD clock and control registers, but not
virtual-memory hardware.
Also had imprecise interrupts.
-
- 370/165 - 25 ns cycle time. Fastest general purpuse mainframe in early
1970s. 2.1 average cycles per instruction.
- 370/168 - same as 360/165 with dynamic address translation
(virtual memory) added. Also larger cache pushed performance to
1.6 cycles per instruction. Density: 4 circuits per chip.
168-3 was 3.5 Mips
- 3033 - last top-of-the-line before the TCM (Thermal Conduction Module)
packaging. Cycle time may have been 51 ns. Density about 40
circuits per chip, allowing 4.5-5 Mips.
303x machines were 370 architecture, with 24 bit logical address
space (but 26 bit = 64MB max physical memory).
- 4361 - small cpu, used as service (maintenance) processor for 3090
- 4381 - a larger mini
- 3081 - First TCM based machine. Cycle time around 30 ns (+-3).
308x machines were XA architecture.
3081 was a dual-processor; the first dual that could not be split
into two separate systems.
- 3083 - a single-processor version of the 3081, built for the
ACP/TCF (airline reservation systems) customer community,
since ACP did not support multi-processing. 3083 was slightly
less expensive than 3081, but gained a 15% performance
increase over running a single processor on the 3081 (because
it omitted the cache synchronization logic).
- 3084 - a twin-3081: i.e. a 4-way system that could be split
and run as two two-way machines.
- 3090 series: 12.5 ns cycle time
From Dave Glass (glass2@glass2.lexington.ibm.com)
From a copy of IBM System/370 System Summary: Processors
(GS22-7001-18) (January 1987):
[1]=The "Processor basic machine cycle time" (nanoseconds)
[2]=Storage access width (number of bytes fetched per access)
[3]=Storage cycle time, (nanoseconds)
[4]=Storage Interleaving
3115-0 3125-0 3135 3135-3 3138 3145 3145-3 3148 3158 3168
3115-2 3125-2 3158-3 3168-3
------ ------ ------ ------ ------ ------ ------ ------ ------- ------
[1] 480 480 275- 275- 275- 202.5- 180- 180- 115 80
1485 1485 1485 315 270 270
[2] 2 2 2/4 2/4 2/4 8 8 8 16 8
[3] 480 480, 770- 770- 715- 540 rd 405 rd 405 rd 1035 rd 320 A
320 935 rd 935 rd 880 rd 607.5 540 wr 540 wr 690-
935 wr 935 wr 935 wr wr 920 wr
[4] - - - - - - - - - 4-way
A=The storage cycle times given for the 3168 do not reflect the time
reductions that result from storage interleaving or, additionally,
for 3168 and 3168-3 tie reductions resulting from the use of the high-speed
buffer.
3031 3032 3033 3033 3033 3081 3083 3084 3090
MG-N MG-S
---- ---- ---- ---- ---- ---- ---- ---- ----
[1] 115 80 57 57 57 24, 24, 24, -
24.5, 24.5 24.5
26 26 26
[2] 8 8 8 8 8 8 8 8 -
[3] 345 320 342 342 342 312 312 312 -
[4] 4-way 4-way 4-way 4-way 4-way 2-way 2-way 2-way -
As for the IBM S/360 machines, here's a summary from a table in
appendix D of the book 'Programming the IBM 360' by Clarence B.
Germain (1967):
360 MODELS
# Core Size 25 30 40 50 65 75 85 91-95 44 67 20
B 4K......X...........................................X.
C 8K......X...X.......................................X.
D 16K......X...X...X...................................X.
E 32K......X...X...X...........................X.........
F 64K..........X...X...X.......................X.........
G 128K..............X...X...X...................X.........
H 256K..............X...X...X...X...............X..X......
I 512K..................X...X...X...X..............X......
J 1024K..................X...X...X...X.....X........X......
K 2048K..............................X.....X........X......
L 4096K..............................X.....X...............
2361 1024K..................X...X...X.........................
2361 2048K..................X...X...X.........................
CD 24K....................................................
DE 48K....................................................
Cycle time(uS) .9 1.5 2.5 2.0 .75 .75 1.04 .75-.125 1.0 .75 3.6
per _ bytes 2 1 2 4 8 8 16 8 4 8 1
This agrees with the information contained in my copy of 'IBM Field
Engineering Education Student Self-Study Course' SR23-3062-7
(November 1973).
$Log: model360.htm,v $
Revision 1.3 2003/01/03 16:27:56 lars
*** empty log message ***
Revision 1.2 2001/10/26 13:28:02 lars
Replaced CMC -> Beagle-Ears
Revision 1.1 2001/01/04 05:15:13 lars
Added file of S/360 Models