Altera Nios II C2H Compiler - Manual (Page 70)

3–30 9.1 Altera Corporation
Nios II C2H Compiler User Guide November 2009
Scheduling
Consider the simple struct declaration shown in Example 3–25 .
Example 3–25. Structure Pointer Declaration
struct s_ptr
{
int element_a;
int element_b;
int element_c;
} * my_struct;
In this example, the expression (my_struct->element_c) translates
to the following:
*((int *)((void *)mystruct + 2*sizeof(int)))
Scheduling This section describes how the C2H Compiler schedules operations. The
C2H Compiler is similar to a traditional C compiler in many respects: It
parses code, creates a graph of the dependencies, performs some
optimizations, schedules the sequence to execute each operation, and
outputs an object file in the form of a hardware accelerator. However,
fundamental differences exist between scheduling for a microprocessor
and scheduling for a hardware accelerator.
Scheduling Concepts for Hardware Accelerators
A microprocessor has limited computational resources, defined by its
arithmetic logic unit (ALU), and limited I/O resources, defined by its
data bus architecture. In contrast, a hardware accelerator can have
arbitrary computational and I/O resources, limited only by the practical
bounds of resource utilization and the ability to achieve frequency
performance. These resources can operate in parallel, stalling only to wait
for data dependencies to resolve.
The C2H Compiler uses the following fundamental rule for scheduling:
Perform computation operations and I/O operations as soon as data
dependencies are resolved.
State Machines
Sections “One-to-One C-to-Hardware Mapping” on page 3–1 , “Variable
Declarations” on page 3–13 , and “Memory Accesses” on page 3–15
described how the C2H Compiler translates individual operations,
assignments, and memory accesses to atomic functional units in