[ HPF Home | Versions | Compilers | Projects | Publications | Applications | Benchmarks | Events | Contact ]

---

Next: Examples of Storage Up: Storage Association Previous: Storage Association Rules

Storage Association Discussion

• []Advice to users. Under these rules, variables in a COMMON block can be mapped as long as the components of the COMMON block are the same in every scoping unit that declares the COMMON block. Rules 4 and 5 also allow variables involved in an EQUIVALENCE statement to be mapped by the mechanism of declaring a rank-one array to cover exactly the aggregate variable group and mapping that array.

  Since an HPF program is nonconforming if it specifies any mapping that would cause a scalar data object to be mapped onto more than one abstract processor, there is a constraint on the sequential variables and aggregate covers that can be mapped. In particular, programs that direct double precision or complex arrays to be mapped such that the storage units of a single array element are split because of some EQUIVALENCE statement or COMMON block layout are nonconforming.

  Correct FORTRAN 77 or Fortran 90 programs will not necessarily be correct without modification in HPF. As the examples in the next section illustrate, use of EQUIVALENCE with COMMON blocks can impact mappability of the variables in subtle ways. To allow maximum optimization for performance, the HPF default for variables is to consider them mappable. In order to get correct separate compilation for subprograms that use COMMON blocks with different aggregate variable groups in different scoping units, it will be necessary to insert the HPF SEQUENCE directive.

  As a check-list for a user to determine the status of a variable or COMMON block, the following questions can be applied, in order:

  • Does the variable appear in some explicit language context which dictates sequential (e.g. EQUIVALENCE) or nonsequential (e.g. array-valued function result variable)?
  • If not, does the variable appear in an explicit mapping directive?
  • If not, does the variable or COMMON block name appear in the list of names on a SEQENCE or NO SEQUENCE directive?
  • If not, does the scoping unit contain a nameless SEQUENCE or NO SEQUENCE?
  • If not, is the compilation affected by some special implementation-dependent environment which dictates that names default to SEQUENCE?
  • If not, then the compiler will consider the variable or COMMON block name non-sequential and is free to apply data mapping optimizations disregarding Fortran sequence and storage association.

  (End of advice to users.)

• []Advice to implementors. In order to protect the user and to facilitate portability of older codes, two implementation options are strongly recommended. First, every implementation should supply some mechanism to verify that the type and shape of every mappable array and the sizes of aggregate variable groups in COMMON blocks are the same in every scoping unit unless the COMMON blocks are declared to be sequential. This same check should also verify that identical mappings have been selected for the variables in COMMON blocks. Implementations without interprocedural information can use a link-time check. The second implementation option recommended is a mechanism to declare that variables and COMMON blocks for a given compilation should be considered sequential unless declared otherwise. The purpose of this feature is to permit compilation of large old libraries or subprograms where storage association is known to exist without requiring that the code be modified to apply the HPF SEQUENCE directive to every COMMON block. (End of advice to implementors.)

Next: