avf: use write combining store for queues tail update

Performance improvement: on supported platforms, currently only Intel Tremont,
use a write combining store to update the tail pointers.
Also, Tremont node variant is added for all.

Type: improvement

Signed-off-by: Radu Nicolau <radu.nicolau@intel.com>
Change-Id: Ie9606e403b7d9655184f778e3ffee3027c8c9edd
diff --git a/src/cmake/cpu.cmake b/src/cmake/cpu.cmake
index 37bc24c..b3f9f0e 100644
--- a/src/cmake/cpu.cmake
+++ b/src/cmake/cpu.cmake
@@ -80,6 +80,10 @@
   if(compiler_flag_march_haswell)
     list(APPEND MARCH_VARIANTS "hsw\;-march=haswell -mtune=haswell")
   endif()
+  check_c_compiler_flag("-march=tremont" compiler_flag_march_tremont)
+  if(compiler_flag_march_tremont)
+    list(APPEND MARCH_VARIANTS "trm\;-march=tremont -mtune=tremont")
+  endif()
   if (GNU_ASSEMBLER_AVX512_BUG)
      message(WARNING "AVX-512 multiarch variant(s) disabled due to GNU Assembler bug")
   else()