Possibly. But
I'm not convinced. I seem to remember the biggest IBM
3090 system drew about 250kW (I'll have to look that up too!). I
imagine a typical system might draw perhaps 50kW.
That is probably for a typical installation. The actual processor is
nowhere near that value (still enough to make the utility very happy,
however). A big string of DASDs adds up very fast, especially the power
sucking IBM disks (why is that? Really big motors)?
I don't know where the figure of 250kW came from. I cannot find it
anywhere.
I have found some figures, though, for indifidual 3090 processor units
_only_, dated 1986 and 1987. While I agree with you in principle,
William, you'll be surprised at some of these.
Figures are electricity consumption in kVA - an upper bound on kW, but
not a useful indicator - and heat output in Btu/h which I have converted
to kW, which should give a lower bound on electrical power requirements.
3090 Model kVA kW Heat
150 (1986) 32 23.2 kW
400 (top of range in 1986) 83.6 61.8 kW
400 with vector coprocessor 97.8 70.6 kW
120E (1987) 29.7 21.4 kW
150E (1987, replaces 150) 30.4 22.1 kW
400E (1987, replaces 400) 79.9 59.6 kW
600E (top of range in 1987) 95.6 71.5 kW
600E with all expansion 134.6 103.6 kW
Note 1986 figures were given with and without vector facility; 1987
figures were given as minimum and maximum values.
These figures are not even for a minimum system - you have to add disk
drives and that awful 400Hz motor generator set - which can consume up
to 7kW in itself.
So, as I said, a typical system based around, say, a model 200 might
consume 50kW, but even that needn't cripple you financially.
Philip.
