cet1@cl.cam.ac.uk (C.E. Thompson) (12/15/90)
In article <548@cadlab.sublink.ORG> staff@cadlab.sublink.ORG (Alex Martelli) writes: > ..., and on others such as IBM RS6000 one would hardly notice >the single vs double precision difference; ... If you want full IEEE single-precision conformance on the RS/6000 (I am not familiar with the compiler switches) it will cost you, as it requires an LRSP (load and round to single precision) order after every arithmetic operation. The floating-point unit works entirely in IEEE double precision internally. Chris Thompson JANET: cet1@uk.ac.cam.phx Internet: cet1%phx.cam.ac.uk@nsfnet-relay.ac.uk