Extensible compilation using composite programming for hardware

1. A method, comprising:

receiving, by a computing device, a composite computer program, the composite computer program comprising first code instructions at an object program level and second code instructions at a metaprogram level;

customizing, by the computing device, a compiler with one or more optimization instructions contained in the second code instructions;

compiling, by the computing device, the first code instructions using the customized compiler to produce an optimized set of machine code for the computing device; and

executing, by the computing device, the optimized set of machine code.

2. The method of claim 1 , wherein the first code instructions comprise at least one level of abstraction above machine code.

3. The method of claim 1 , wherein the optimized set of machine code is optimized by the customized compiler for a specified hardware configuration of the computing device based on the second code instructions.

4. The method of claim 1 , wherein the customizing further comprises:

producing, by the computing device executing the second code instructions, a plurality of alternative optimization results of a combination of optimizations for the compiler to implement; and

pruning, by the computing device executing the second code instructions, one or more of the plurality of alternative optimization results using a cost measure.

5. The method of claim 1 , further comprising:

receiving, by the computing device, a manual modification to one or more parameters specified in the second code instructions; and

repeating the customizing, compiling, and executing according to the modification.

6. The method of claim 1 , further comprising:

implementing, by the computing device executing the composite computer program, a machine learning algorithm to determine one or more parameters to modify to optimize the first code instructions for a particular hardware platform;

modifying, automatically by the computing device, the determined one or more parameters specified in the second code instructions; and

repeating the customizing, compiling, and executing according to the modification.

7. The method of claim 6 , further comprising:

receiving, by the computing device, one or more outcome goals as the second code instructions for the machine learning algorithm implementation.

8. The method of claim 1 , wherein the one or more optimization instructions contained in the second code instructions comprises at least one of constant folding, monomorphization, or list fusion.

9. A computing device, comprising:

a memory containing machine readable medium comprising machine executable code having stored thereon instructions for performing a method of implementing a composite computer program, the composite computer program comprising first code instructions at an object program level and second code instructions at a metaprogram level; and

a processor coupled to the memory, the processor configured to execute the machine executable code to cause the processor to:

receive the composite computer program;

customize a compiler of the computing device with one or more optimization instructions contained in the second code instructions;

compile only the first code instructions using the customized compiler to produce an optimized set of machine code for the computing device; and

execute the optimized set of machine code.

10. The computing device of claim 9 , wherein the first code instructions comprise at least one level of abstraction above machine code.

11. The computing device of claim 9 , wherein the optimized set of machine code is optimized by the customized compiler for a specified hardware configuration of the computing device based on the second code instructions.

12. The computing device of claim 9 , wherein the processor is further configured, as part of the customizing, to:

produce a plurality of alternative optimization results of a combination of optimizations for the compiler to implement; and

prune one or more of the plurality of alternative optimization results using a cost measure.

13. The computing device of claim 9 , wherein the processor is further configured to:

receive a manual modification to one or more parameters specified in the second code instructions; and

repeat the customization, compilation, and execution according to the modification.

14. The computing device of claim 9 , wherein the processor is further configured to:

implement a machine learning algorithm to determine one or more parameters to modify to optimize the first code instructions for a particular hardware platform;

modify, automatically by the computing device, the determined one or more parameters specified in the second code instructions; and

repeat the customization, compilation, and execution according to the modification.

15. The computing device of claim 14 , wherein the processor is further configured to:

receive one or more outcome goals as the second code instructions for the machine learning algorithm implementation.

16. A non-transitory computer-readable medium having program code recorded thereon, the program code comprising:

code for causing a computing device to receive a composite computer program, the composite computer program comprising first code instructions at an object program level and second code instructions at a metaprogram level;

code for causing the computing device to customize a compiler with one or more optimization instructions contained in the second code instructions;

code for causing the computing device to compile the first code instructions using the customized compiler to produce an optimized set of machine code for the computing device; and

code for causing the computing device to execute the optimized set of machine code.

17. The non-transitory computer-readable medium of claim 16 , wherein the optimized set of machine code is optimized by the customized compiler for a specified hardware configuration of the computing device based on the second code instructions.

18. The non-transitory computer-readable medium of claim 16 , further comprising as part of the customizing:

code for causing the computing device to produce a plurality of alternative optimization results of a combination of optimizations for the compiler to implement; and

code for causing the computing device to prune one or more of the plurality of alternative optimization results using a cost measure.

19. The non-transitory computer-readable medium of claim 16 , further comprising:

code for causing the computing device to receive a manual modification to one or more parameters specified in the second code instructions; and

code for causing the computing device to repeat the customization, compilation, and execution according to the modification.

20. The non-transitory computer-readable medium of claim 16 , further comprising:

code for causing the computing device to receive one or more outcome goals as the second code instructions for a machine learning algorithm implementation;

code for causing the computing device to implement a machine learning algorithm to determine one or more parameters to modify to optimize the first code instructions for a particular hardware platform, based on the received one or more outcome goals;

code for causing the computing device to automatically modify the determined one or more parameters specified in the second code instructions; and

code for causing the computing device to repeat the customization, compilation, and execution according to the modification.