Speed up truffle with 256b TBL instructions

256b wide SVE vectors allow some simplification of truffle.
Up to 40% speedup on graviton3. Going from 12500 MB/s to 17000 MB/s
onhe microbenchmark.
SVE2 also offer this capability for 128b vector with a speedup around
25% compared to normal SVE

Add unit tests and benchmark for this wide variant

Signed-off-by: Yoan Picchi <yoan.picchi@arm.com>

src/util/arch/arm/arm.h

@@ -53,5 +53,11 @@
 #define HAVE_SVE2_BITPERM
 #endif
 
+#if defined(HAVE_SVE2)
+#define CAN_USE_WIDE_TRUFFLE 1
+#elif defined(HAVE_SVE)
+#define CAN_USE_WIDE_TRUFFLE (svcntb() >= 32)
+#endif
+
 #endif // UTIL_ARCH_ARM_H_
 

src/util/arch/arm/simd_types.h

@@ -34,5 +34,9 @@
 typedef int32x4_t m128;
 #endif
 
+#if !defined(m256) && defined(m128)
+typedef struct {m128 lo; m128 hi;} m256;
+#endif
+
 #endif /* SIMD_TYPES_ARM_H */
 

src/util/supervector/arch/arm/types.hpp

@@ -31,3 +31,6 @@
 typedef int32x4_t m128;
 #endif
 
+#if !defined(m256) && defined(m128)
+typedef struct {m128 lo; m128 hi;} m256;
+#endif