Architecture and software support in IBM S/390 parallel enterprise sewers for IEEE floating-point arithmetic

IEEE Binary Floating-Point is an industry-standard architecture. The IBM Sy stem/360(TM) hexadecimal floating-point architecture predates the IEEE stan dard and has been carried forward through the System/370(TM) to current Sys tem/390(R) processors, The growing importance of industry standards and flo ating-point combined to produce a need for IEEE Floating-Point on System/39 0. At the same time, customer investment in IBM floating-point had to be pr eserved. This paper describes the architecture, hardware, and software effo rts that combined to produce a conforming implementation of IEEE Floating-P oint on System/390 while retaining compatibility with the original IBM arch itecture.