Using a concurrent partial inspector loop with speculative parallelism
A method for executing a loop in an application that includes executing iterations in a first segment of the loop by a base thread, logging memory transactions that occur during execution of iterations in the first segment by a co-inspector thread to obtain a co-inspector log, executing iterations in a second segment of the loop by a co-thread to obtain temporary results, logging memory transactions that occur during execution of iterations in the second segment to obtain a co-thread log, and comparing the co-inspector log and the co-thread log to determine whether a thread interdependency exists.
1. A method for executing a loop in an application comprising:
executing iterations in a first segment of the loop by a base thread;
logging memory transactions that occur during execution of iterations in the first segment by a co-inspector thread to obtain a co-inspector log;
executing iterations in a second segment of the loop by a co-thread to obtain temporary results;
logging memory transactions that occur during execution of iterations in the second segment to obtain a co-thread log;
identifying a thread interdependency by comparing the co-inspector log and the co-thread log;
determining, after executing iterations in the first segment, a location of the thread interdependency in the second segment;
committing temporary results up to the location; and
re-starting execution of the second segment at the location by the base thread.
2. The method of claim 1 , wherein the thread interdependency is associated with an indeterminable variable.
3. The method of claim 1 , wherein logging the memory transactions by the co-thread comprises:
executing a subset of instructions associated with an indeterminable variable in the first segment; and
logging memory transactions associated with the indeterminable variable.
4. The method of claim 1 , wherein the loop comprises a plurality of instrumented code instructions.
5. The method of claim 4 , wherein the plurality of instrumented code instructions enable logging of indeterminable variables in the loop.
6. The method of claim 1 , wherein the base thread, the co-inspector thread, and the co-thread execute in parallel.
7. The method of claim 1 , wherein comparing the co-inspector log and the co-thread log occurs after the co-inspector thread completes logging a portion of the first segment.
8. The method of claim 1 , wherein comparing the co-inspector log and the co-thread log occurs after the base thread completes logging a portion of the first segment.
9. A computer system for executing a loop comprising:
a processor; and
a memory operatively connected to the processor and storing a plurality of instructions comprising functionality to:
execute iterations in a first segment of the loop by a base thread;
log memory transactions that occur during execution of iterations in the first segment by a co-inspector thread to obtain a co-inspector log;
execute iterations in a second segment of the loop by a co-thread to obtain temporary results;
log memory transactions that occur during execution of iterations in the second segment to obtain a co-thread log;
identify a thread interdependency by comparing the co-inspector log and the co-thread log;
determine, after executing iterations in the first segment, a location of the thread interdependency in the second segment;
commit temporary results up to the location; and
re-start execution of the second segment at the location by the base thread.
10. The computer system of claim 9 , wherein the plurality of instructions further comprise functionality to:
commit results of the execution of the base thread.
11. The computer system of claim 9 , wherein the base thread, the co-inspector thread, and the co-thread execute in parallel.
12. A distributed computer system for executing a loop in an application comprising:
a first processor;
a second processor; and
a memory operatively connected to the first processor and the second processor and comprising software instructions for causing the distributed computer system to perform a method of:
executing, on the first processor, iterations in a first segment of the loop by a base thread;
logging memory transactions that occur during execution of iterations in the first segment by a co-inspector thread to obtain a co-inspector log;
executing, on the second processor, iterations in a second segment of the loop by a co-thread to obtain temporary results;
logging memory transactions that occur during execution of iterations in the second segment to obtain a co-thread log;
identifying a thread interdependency by comparing the co-inspector log and the co-thread log;
determining, after executing iterations in the first segment, a location of the thread interdependency in the second segment;
committing temporary results up to the location; and
re-starting execution of the second segment at the location by the base thread.