Armv7m: Add vmov cmp loop example

example.py

@@ -688,6 +688,23 @@ def core(self,slothy):
         slothy.config.variable_size=True
         slothy.config.outputs = ["r6"]
         slothy.optimize_loop("start")
+
+class Armv7mLoopVmovCmp(Example):
+    def __init__(self, var="", arch=Arch_Armv7M, target=Target_CortexM7):
+        name = "loop_vmov_cmp"
+        infile = name
+
+        if var != "":
+            name += f"_{var}"
+            infile += f"_{var}"
+        name += f"_{target_label_dict[target]}"
+
+        super().__init__(infile, name, rename=True, arch=arch, target=target)
+
+    def core(self,slothy):
+        slothy.config.variable_size=True
+        slothy.config.outputs = ["r6"]
+        slothy.optimize_loop("start")
 
 class ntt_kyber_123_4567(Example):
     def __init__(self, var="", arch=AArch64_Neon, target=Target_CortexA55, timeout=None):
@@ -1486,6 +1503,7 @@ def main():
                  LoopLe(),
                  Armv7mLoopSubs(),
                  Armv7mLoopCmp(),
+                 Armv7mLoopVmovCmp(),
 
                  CRT(),
 

examples/opt/armv7m/armv7m_simple0_opt_m7.s

@@ -1,24 +1,24 @@
 
         start:
-                                       // Instructions:    6
-                                       // Expected cycles: 5
-                                       // Expected IPC:    1.20
-                                       //
-                                       // Cycle bound:     5.0
-                                       // IPC bound:       1.20
-                                       //
-                                       // Wall time:     0.01s
-                                       // User time:     0.01s
-                                       //
-                                       // ----- cycle (expected) ------>
-                                       // 0                        25
-                                       // |------------------------|----
-        ldr r7, [r0, #4]               // *.............................
-        add r7, r2, r7                 // .*............................
-        eor.w r12, r7, r3              // ..*...........................
-        smlabt r12, r2, r2, r12        // ..*...........................
-        asrs r3, r12, #1               // ....*.........................
-        str r3, [r0, #4]               // ....*.........................
+                                      // Instructions:    6
+                                      // Expected cycles: 5
+                                      // Expected IPC:    1.20
+                                      //
+                                      // Cycle bound:     5.0
+                                      // IPC bound:       1.20
+                                      //
+                                      // Wall time:     0.02s
+                                      // User time:     0.02s
+                                      //
+                                      // ----- cycle (expected) ------>
+                                      // 0                        25
+                                      // |------------------------|----
+        ldr r6, [r0, #4]              // *.............................
+        add r6, r2, r6                // .*............................
+        eor.w r3, r6, r3              // ..*...........................
+        smlabt r12, r2, r2, r3        // ..*...........................
+        asrs r3, r12, #1              // ....*.........................
+        str r3, [r0, #4]              // ....*.........................
 
                                       // ------ cycle (expected) ------>
                                       // 0                        25

examples/opt/armv7m/loop_vmov_cmp_opt_m7.s

@@ -0,0 +1,31 @@
+/* For example, r5 represents an address where we will stop iterating and r6 is
+the actual pointer which is incremented inside the loop. */
+
+mov.w r6, #0
+add.w r5, r6, #64
+vmov s0, r5
+
+start:
+                              // Instructions:    1
+                              // Expected cycles: 1
+                              // Expected IPC:    1.00
+                              //
+                              // Cycle bound:     1.0
+                              // IPC bound:       1.00
+                              //
+                              // Wall time:     0.02s
+                              // User time:     0.02s
+                              //
+                              // ----- cycle (expected) ------>
+                              // 0                        25
+                              // |------------------------|----
+        add r6, r6, #4        // *.............................
+
+                               // ------ cycle (expected) ------>
+                               // 0                        25
+                               // |------------------------|-----
+        // add r6, r6, #4      // *..............................
+
+        vmov r5, s0
+        cmp r6, r5
+        bne start