TOPIC: Implementing Glenda On the iPSC/860

By Dr. Ray Seyfarth

University Of Southern Mississippi

Summary and Comments: By Lee Emmanwori

Dr. Seyfarth intoduced the concept of Glenda which is a library of functions providing a message-passing interface for MIMD machines. Glenda uses a global tuple space model. A tuple is an ordered collection of data but for Glenda, a tuple has to be in the form ("data", node, I, row).

1. Dr. Seyfarth explained and listed some of the Glenda functions as:

1. gl_spawn(name,count);
2. gl_exit();
3. gl_out(tuple_name,...);
4. gl_in(tuple_name,...);
5. gl_rd(tuple_name,...); (this is a non-destructive function)

1. Some examples of Glenda functions are:

1. gl_out("point",i,x,y); the types for i, x and y are implicit.
2. gl_out("vector", row, size, data); the array size is determined by Glenda.
3. gl_in("point",i,?,x,?,y); a "point" tuple with equal i is fetched , x, y are assigned new values.
4. gl_in("vector",row,?,size,?,data,n);
5. gl_rd("point",i,?,x,?,y);

1. Glenda Preprocessor: Converts Glenda calls to C function calls. Reads a hash file to learn how to hash tuples. The preprocessor also converts tuple name to integers, converts each tuple component to several function parameters.