mwang@watmath.UUCP (mwang) (08/07/84)
_D_E_P_A_R_T_M_E_N_T _O_F _C_O_M_P_U_T_E_R _S_C_I_E_N_C_E _U_N_I_V_E_R_S_I_T_Y _O_F _W_A_T_E_R_L_O_O _S_E_M_I_N_A_R _A_C_T_I_V_I_T_I_E_S _S_Y_S_T_E_M_S _S_E_M_I_N_A_R - Friday, August 17, 1984. J.R. Jagannathan, a graduate student of this depart- ment, will speak on ``Eazyflow: A Hybrid Model for Parallel Processing.'' TIME: 3:00 PM (Please note) ROOM: MC 5158 ABSTRACT A hybrid dataflow model is proposed that combines the data-driven and demand-driven models in such a way as to reap their benefits while avoiding their drawbacks. The model assumes that programs to be implemented this way are expressed in Lucid. The meanings of Lucid pro- grams are given by its mathematical semantics, which does not explicitly refer to operational concepts. For Lucid, it is possible to give different operational se- mantics which are equally correct. In fact, it is pos- sible to devise a classification scheme that partitions Lucid programs into data-driveable and demand-driveable parts. The scheme is justified by showing that no part that is classified as demand-driveable can be data- driven, whereas all parts classified as data-driveable really can be data-driven. The requirements of an abstract engine embodying the hybrid model are outlined, the interpretations of the various operators, by such an engine, are outlined, and the data-driven/demand-driven interface is discussed. August 7, 1984