diff --git a/tgt-vvp/eval_vec4.c b/tgt-vvp/eval_vec4.c
index ebad86799..ae9524d37 100644
--- a/tgt-vvp/eval_vec4.c
+++ b/tgt-vvp/eval_vec4.c
@@ -1191,8 +1191,7 @@ static void draw_unary_inc_dec(ivl_expr_t sub, bool incr, bool pre)
       if (pre) {
 	      /* prefix means we add the result first, and store the
 		 result, as well as leaving a copy on the stack. */
-	    fprintf(vvp_out, "    %%pushi/vec4 1, 0, %u;\n", wid);
-	    fprintf(vvp_out, "    %s;\n", cmd);
+	    fprintf(vvp_out, "    %si 1, 0, %u;\n", cmd, wid);
 	    fprintf(vvp_out, "    %%dup/vec4;\n");
 	    fprintf(vvp_out, "    %%store/vec4 v%p_0, 0, %u;\n", sig, wid);
 
@@ -1200,8 +1199,7 @@ static void draw_unary_inc_dec(ivl_expr_t sub, bool incr, bool pre)
 	      /* The post-fix decrement returns the non-decremented
 		 version, so there is a slight re-arrange. */
 	    fprintf(vvp_out, "    %%dup/vec4;\n");
-	    fprintf(vvp_out, "    %%pushi/vec4 1, 0, %u;\n", wid);
-	    fprintf(vvp_out, "    %s;\n", cmd);
+	    fprintf(vvp_out, "    %si 1, 0, %u;\n", cmd, wid);
 	    fprintf(vvp_out, "    %%store/vec4 v%p_0, 0, %u;\n", sig, wid);
       }
 }
@@ -1245,8 +1243,7 @@ static void draw_unary_vec4(ivl_expr_t expr)
 	  case '-':
 	    draw_eval_vec4(sub);
 	    fprintf(vvp_out, "    %%inv;\n");
-	    fprintf(vvp_out, "    %%pushi/vec4 1, 0, %u;\n", ivl_expr_width(sub));
-	    fprintf(vvp_out, "    %%add;\n");
+	    fprintf(vvp_out, "    %%addi 1, 0, %u;\n", ivl_expr_width(sub));
 	    break;
 
 	  case 'A': /* nand (~&) */
@@ -1287,13 +1284,11 @@ static void draw_unary_vec4(ivl_expr_t expr)
 
 	      /* Test if (m) < 0 */
 	    fprintf(vvp_out, "    %%dup/vec4;\n");
-	    fprintf(vvp_out, "    %%pushi/vec4 0, 0, %u;\n", ivl_expr_width(sub));
-	    fprintf(vvp_out, "    %%cmp/s;\n");
+	    fprintf(vvp_out, "    %%cmpi/s 0, 0, %u;\n", ivl_expr_width(sub));
 	    fprintf(vvp_out, "    %%jmp/0xz T_%u.%u, 5;\n", thread_count, local_count);
 	      /* If so, calculate -(m) */
 	    fprintf(vvp_out, "    %%inv;\n");
-	    fprintf(vvp_out, "    %%pushi/vec4 1, 0, %u;\n", ivl_expr_width(sub));
-	    fprintf(vvp_out, "    %%add;\n");
+	    fprintf(vvp_out, "    %%addi 1, 0, %u;\n", ivl_expr_width(sub));
 	    fprintf(vvp_out, "T_%u.%u ;\n", thread_count, local_count);
 	    local_count += 1;
 	    break;
diff --git a/tgt-vvp/stmt_assign.c b/tgt-vvp/stmt_assign.c
index 83bcd7df3..a21f77a53 100644
--- a/tgt-vvp/stmt_assign.c
+++ b/tgt-vvp/stmt_assign.c
@@ -134,8 +134,7 @@ static void get_vec_from_lval_slice(ivl_lval_t lval, struct vec_slice_info*slice
 	    } else {
 		  fprintf(vvp_out, "    %%load/vec4 v%p_%lu;\n", sig, use_word);
 	    }
-	    fprintf(vvp_out, "    %%pushi/vec4 %lu, 0, 32;\n", part_off);
-	    fprintf(vvp_out, "    %%part/u %u;\n", wid);
+	    fprintf(vvp_out, "    %%parti/u %u, %lu, 32;\n", wid, part_off);
 
       } else if (ivl_signal_dimensions(sig)==0 && part_off_ex!=0 && word_ix==0) {
 
diff --git a/tgt-vvp/vvp_proc_loops.c b/tgt-vvp/vvp_proc_loops.c
index 7c03b0d9d..043540756 100644
--- a/tgt-vvp/vvp_proc_loops.c
+++ b/tgt-vvp/vvp_proc_loops.c
@@ -164,14 +164,12 @@ int show_stmt_repeat(ivl_statement_t net, ivl_scope_t sscope)
 	   unsigned, then we only need to try to escape if expr==0 as
 	   it will never be <0. */
       fprintf(vvp_out, "T_%u.%u %%dup/vec4;\n", thread_count, lab_top);
-      fprintf(vvp_out, "    %%pushi/vec4 0, 0, %u;\n", ivl_expr_width(expr));
-      fprintf(vvp_out, "    %%cmp/%s;\n", sign);
+      fprintf(vvp_out, "    %%cmpi/%s 0, 0, %u;\n", sign, ivl_expr_width(expr));
       if (ivl_expr_signed(expr))
 	    fprintf(vvp_out, "    %%jmp/1xz T_%u.%u, 5;\n", thread_count, lab_out);
       fprintf(vvp_out, "    %%jmp/1 T_%u.%u, 4;\n", thread_count, lab_out);
 	/* This adds -1 (all ones in 2's complement) to the count. */
-      fprintf(vvp_out, "    %%pushi/vec4 1, 0, %u;\n", ivl_expr_width(expr));
-      fprintf(vvp_out, "    %%sub;\n");
+      fprintf(vvp_out, "    %%subi 1, 0, %u;\n",  ivl_expr_width(expr));
 
       rc += show_statement(ivl_stmt_sub_stmt(net), sscope);