Initial integration with QDP-JIT

 - packers are fully threaded and use raw QDPJIT pointer access.
 - packers work only when QDP-JIT is run with -layout ocsri

Todo:
 - recheck integration has not broken non QDP-JIT build
 - bulletproof (with assertions) to ensure layout is ocsri, or adapta also to oscri
 - Use as 'native' dslash: only possible in ocsri mode since that is what is wired into QDPJIT
 - lots of testing.
        - test full chroma QDP-JIT build (e.g. Leapfrog)

NEWS

@@ -1,3 +1,4 @@
+6/24/2015 (Balint Joo/Thorsten Kurth) - Added lots of cleanup for KNL. Added packers for QDP-JIT (OCSRI layout for now)
 1/12/2016 (Balint Joo/Aaron Walden) - Added AVX512 output from code generator. Tested on Intel SDE. Single and Double precision work, 16-bit precision is currently broken.
 
 4/9/2015 (Balint Joo): Added Basic Multi-Shift solvers -- but can still add more optimizations. Need to test multi-node etc.

TODO

@@ -1,10 +1,13 @@
+To do with QDPJIT:
+  - Thorough testing for SOA combinations (initially done with 4)
+  - Bullet proofing assertions that data is in OCSRI layout or make the code generically figure out the right layout.
+  - Write native dslash using raw pointers: need OSCRI layout assertion and InnerLength assertion (easy)
+    - but issue w.r.t currently onle inner=8 from QDPJIT apparently working in my tests
+
 TO Do:
 - Add cache blocking in X -- hooks are already present in the kernels, but need to adapt loops.
-- Add other solvers specially multi-shift
 - Solve N-systems at once (ongoing PhD project at Old Dominion University Computer Science Department: ODU-JLab collaboration)
 - Clean code for non ICC compilers (work ongoing by Diptorup Deb, Renaissance Computing Institute, University of North Carolina, Chapel Hill
-- Add other processor targets: SSE, AVX2, AVX512 - codegen supports these already but need to be added
-- Add other fermions: e.g. Twisted Mass, HiSQ, Domain Wall?
 - Bullet proof a little (A LOT!!!)
 - Better tuning (implement 1/2/3/4 threads per core instead of just all 4)
 - reduce verbosite and use master printf to better effect (log levels?)

include/qphix/Makefile.am

@@ -39,7 +39,7 @@ qphix_include_HEADERS += \
 #
 
 if QPHIX_BUILD_WITH_QDP
-qphix_include_HEADERS += ./qdp_packer.h
+qphix_include_HEADERS += ./qdp_packer.h ./qdp_packer_parscalar.h qdp_packer_qdpjit.h
 endif