Reduces LUT needed by mul extension

bench/tests/op/mulh.s

@@ -1,19 +1,54 @@
-# expected: 0000000000000000, 0000000000000000, c1b080e9, ffffffffffffffff
     .text
 _start:
-    li x2, 0xdead0
-    li x3, 0x0
-    li x4, 0x1
-    li x5, (-1)
-    li x6, 0xdead0
-    li x7, 0xdead00000000
+    li x2,  0xdead00000        # +59774074880
+    li x3,  0              # zero
+    li x4,  1              # one
+    li x5, -1              # -1
+    li x6,  0xdead00000        # duplicate of x2
+    li x7, -235555555555           # negative
+    li x8,  123456789         # positive
+    li x9, -765432000000         # negative
+    li x10, 0x7fffffffffffffff  # max positive 64b
+    li x11, 0x8000000000000000  # min negative 64b
 
-    # Mul 0
+    # --------------------------------------------------------------
+    # 1. MULH : a * 0
     mulh x31, x2, x3
+    # expected: 0000000000000000
 
-    # Basic multiplications
+    # 2. MULH : a * 1
     mulh x31, x2, x4
-    mulh x31, x7, x7
+    # expected: 0000000000000000
 
-    # Overflow
+    # 3. MULH : a * a
+    mulh x31, x2, x6
+    # expected: 00000000000000c1
+
+    # 4. MULH : -1 * a
     mulh x31, x5, x2
+    # expected: ffffffffffffffff
+
+    # 5. MULH : negative * negative
+    mulh x31, x7, x7
+    # expected: 0000000000000bbf
+
+    # 6. MULH : negative * positive
+    mulh x31, x7, x8
+    # expected: fffffffffffffffe
+
+    # 7. MULH : positive * negative
+    mulh x31, x8, x9
+    # expected: fffffffffffffffa
+
+    # 8. MULH : overflow example (max * 2)
+    li x12, 2
+    mulh x31, x10, x12
+    # expected: 0000000000000000
+
+    # 9. MULH : max * max
+    mulh x31, x10, x10
+    # expected: 3fffffffffffffff
+
+    # 10. MULH : min * min
+    mulh x31, x11, x11
+    # expected: 4000000000000000