re PR libfortran/78379 (Processor-specific versions for matmul)

2017-05-25  Thomas Koenig  <tkoenig@gcc.gnu.org>

	PR libfortran/78379
	* Makefile.am: Add generated/matmulavx128_*.c files.
	Handle them for compiling and setting the right flags.
	* acinclude.m4: Add tests for FMA3, FMA4 and AVX128.
	* configure.ac: Call them.
	* Makefile.in: Regenerated.
	* config.h.in: Regenerated.
	* configure: Regenerated.
	* m4/matmul.m4:  Handle AMD chips by calling 128-bit AVX
	versions which use FMA3 or FMA4.
	* m4/matmulavx128.m4: New file.
        * generated/matmul_c10.c: Regenerated.
        * generated/matmul_c16.c: Regenerated.
        * generated/matmul_c4.c: Regenerated.
        * generated/matmul_c8.c: Regenerated.
        * generated/matmul_i1.c: Regenerated.
        * generated/matmul_i16.c: Regenerated.
        * generated/matmul_i2.c: Regenerated.
        * generated/matmul_i4.c: Regenerated.
        * generated/matmul_i8.c: Regenerated.
        * generated/matmul_r10.c: Regenerated.
        * generated/matmul_r16.c: Regenerated.
        * generated/matmul_r4.c: Regenerated.
        * generated/matmul_r8.c: Regenerated.
        * generated/matmulavx128_c10.c: New file.
        * generated/matmulavx128_c16.c: New file.
        * generated/matmulavx128_c4.c: New file.
        * generated/matmulavx128_c8.c: New file.
        * generated/matmulavx128_i1.c: New file.
        * generated/matmulavx128_i16.c: New file.
        * generated/matmulavx128_i2.c: New file.
        * generated/matmulavx128_i4.c: New file.
        * generated/matmulavx128_i8.c: New file.
        * generated/matmulavx128_r10.c: New file.
        * generated/matmulavx128_r16.c: New file.
        * generated/matmulavx128_r4.c: New file.
        * generated/matmulavx128_r8.c: New file.

From-SVN: r248472

1 files changed, 38 insertions, 0 deletions

diff --git a/libgfortran/generated/matmul_i4.c b/libgfortran/generated/matmul_i4.c
index 475b214ae79..f8f0cdb96cc 100644
--- a/libgfortran/generated/matmul_i4.c
+++ b/libgfortran/generated/matmul_i4.c
@@ -1734,6 +1734,24 @@ matmul_i4_avx512f (gfc_array_i4 * const restrict retarray,
 
 #endif  /* HAVE_AVX512F */
 
+/* AMD-specifix funtions with AVX128 and FMA3/FMA4.  */
+
+#if defined(HAVE_AVX) && defined(HAVE_FMA3) && defined(HAVE_AVX128)
+void
+matmul_i4_avx128_fma3 (gfc_array_i4 * const restrict retarray, 
+	gfc_array_i4 * const restrict a, gfc_array_i4 * const restrict b, int try_blas,
+	int blas_limit, blas_call gemm) __attribute__((__target__("avx,fma")));
+internal_proto(matmul_i4_avx128_fma3);
+#endif
+
+#if defined(HAVE_AVX) && defined(HAVE_FMA4) && defined(HAVE_AVX128)
+void
+matmul_i4_avx128_fma4 (gfc_array_i4 * const restrict retarray, 
+	gfc_array_i4 * const restrict a, gfc_array_i4 * const restrict b, int try_blas,
+	int blas_limit, blas_call gemm) __attribute__((__target__("avx,fma4")));
+internal_proto(matmul_i4_avx128_fma4);
+#endif
+
 /* Function to fall back to if there is no special processor-specific version.  */
 static void
 matmul_i4_vanilla (gfc_array_i4 * const restrict retarray, 
@@ -2332,6 +2350,26 @@ void matmul_i4 (gfc_array_i4 * const restrict retarray,
 	    }
 #endif  /* HAVE_AVX */
         }
+    else if (__cpu_model.__cpu_vendor == VENDOR_AMD)
+      {
+#if defined(HAVE_AVX) && defined(HAVE_FMA3) && defined(HAVE_AVX128)
+        if ((__cpu_model.__cpu_features[0] & (1 << FEATURE_AVX))
+	    && (__cpu_model.__cpu_features[0] & (1 << FEATURE_FMA)))
+	  {
+            matmul_fn = matmul_i4_avx128_fma3;
+	    goto store;
+	  }
+#endif
+#if defined(HAVE_AVX) && defined(HAVE_FMA4) && defined(HAVE_AVX128)
+        if ((__cpu_model.__cpu_features[0] & (1 << FEATURE_AVX))
+	     && (__cpu_model.__cpu_features[0] & (1 << FEATURE_FMA4)))
+	  {
+            matmul_fn = matmul_i4_avx128_fma4;
+	    goto store;
+	  }
+#endif
+
+      }
    store:
       __atomic_store_n (&matmul_p, matmul_fn, __ATOMIC_RELAXED);
    }