temporary benchmark, <vvaddint32> working

rvvector/benchmark_temp/vx_vec_memcpy.s

@@ -0,0 +1,17 @@
+.type vx_vec_memcpy, @function
+.global vx_vec_memcpy
+# void *memcpy(void* dest, const void* src, size_t n)
+# a0=dest, a1=src, a2=n
+#
+vx_vec_memcpy:
+#  memcpy
+  mv a3, a0 # Copy destination
+  vsetvli t0, a2, e8,m8  # Vectors of 8b
+loop:
+  vlb.v v0, (a1)                # Load bytes
+    add a1, a1, t0              # Bump pointer
+    sub a2, a2, t0              # Decrement count
+  vsb.v v0, (a3)                # Store bytes
+    add a3, a3, t0              # Bump pointer
+    bnez a2, loop               # Any more?
+    ret                         # Return