[llvm-commits] [PATCH] Adding the TSVC loop benchmarks to the test suite

[Hal Finkel]

Please review the attached patch which adds the TSVC loop benchmarks to
the LLVM test suite. These are the benchmarks associated with the paper:
Maleki, S., et al. An Evaluation of Vectorizing Compilers. PACT'11.

This is the benchmark set that I have used for evaluating the
vectorizer, and I think that it will make a valuable addition to the
test suite.

The authors have kindly released the benchmark under an open-source
license for this purpose.

Compared to the original benchmark, I've made the following
enhancements:

 - I made the number of test iterations a command-line parameter (this
   makes it easy to adjust the run time)
 - I made it possible to run the tests using both float or double
 - I split out the different classes of loops into separate tests (I
   did this primarily to make performance improvements and regressions
   noticeable)
 - I removed the timing information from the output

I have adjusted each stage so that the double-precision version takes
approximately 5 seconds to run on my machine (the float version should
be faster, but is not always -- one of the reasons it will be good
have this added). With 18 classes each with double and float variants,
and assuming that the time for the native binary is comparable to the
test binary, the entire benchmark should take approximately 360 seconds
to run. Is that too long?

Do I need to do anything special with the fcmp tolerances?

Thanks again,
Hal

-- 
Hal Finkel
Postdoctoral Appointee
Leadership Computing Facility
Argonne National Laboratory
-------------- next part --------------
A non-text attachment was scrubbed...
Name: add-TSVC-20120926.patch
Type: text/x-patch
Size: 166631 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20120926/e337f358/attachment.bin>