diff --git a/lib.nas b/lib.nas
index 9669f67..08a1c4d 100644
--- a/lib.nas
+++ b/lib.nas
@@ -49,7 +49,7 @@ var id=func(object){
     return __id(object);
 }
 
-# int will get the integer of input number.
+# int will get the integer of input number/string.
 # but carefully use it, because int has range between -2147483648~2147483647
 var int=func(val){
     return __int(val);
diff --git a/nasal_builtin.h b/nasal_builtin.h
index ba00ea9..ad24280 100644
--- a/nasal_builtin.h
+++ b/nasal_builtin.h
@@ -193,9 +193,9 @@ nas_ref builtin_id(nas_ref* local,nasal_gc& gc)
 nas_ref builtin_int(nas_ref* local,nasal_gc& gc)
 {
     nas_ref val=local[1];
-    if(val.type!=vm_num)
+    if(val.type!=vm_num && val.type!=vm_str)
         return nil;
-    return {vm_num,f64((i32)val.num())};
+    return {vm_num,f64((i32)val.tonum())};
 }
 nas_ref builtin_floor(nas_ref* local,nasal_gc& gc)
 {
diff --git a/stl/lib.nas b/stl/lib.nas
index 9669f67..08a1c4d 100644
--- a/stl/lib.nas
+++ b/stl/lib.nas
@@ -49,7 +49,7 @@ var id=func(object){
     return __id(object);
 }
 
-# int will get the integer of input number.
+# int will get the integer of input number/string.
 # but carefully use it, because int has range between -2147483648~2147483647
 var int=func(val){
     return __int(val);
diff --git a/test/calc.nas b/test/calc.nas
index 05dfccd..061c488 100644
--- a/test/calc.nas
+++ b/test/calc.nas
@@ -57,6 +57,7 @@ var testfile=[
     "loop.nas",
     "mandel.nas",
     "mandelbrot.nas",
+    "mcpu.nas",
     "md5.nas",
     "md5compare.nas",
     "module_test.nas",
@@ -97,12 +98,6 @@ var longest=func(vec...){
 }
 var padding_length=longest(source,lib,testfile,module);
 
-var getname=func(s){
-    var (len,ch)=(size(s),' '[0]);
-    for(var i=0;i<len and s[i]!=ch;i+=1);
-    return substr(s,0,i);
-}
-
 var count=func(s,c){
     var cnt=0;
     foreach(var i;split(c,s))
@@ -119,7 +114,7 @@ var calc=func(codetype,files,path=""){
     println(codetype);
     var (bytes,ctx,line,semi,line_cnt,semi_cnt)=(0,"",0,0,0,0);
     forindex(var i;files){
-        var s=io.fin(getname(path~files[i]));
+        var s=io.exists(path~files[i])?io.fin(path~files[i]):"";
         (line_cnt,semi_cnt)=(count(s,'\n'),count(s,';'));
         println(rightpad(files[i],padding_length),'| ',
             column(line_cnt),'line | ',
diff --git a/test/mcpu.nas b/test/mcpu.nas
new file mode 100644
index 0000000..e3a3db4
--- /dev/null
+++ b/test/mcpu.nas
@@ -0,0 +1,172 @@
+var inst={
+    inst_stop:0,
+    inst_mov_reg_reg:1,
+    inst_mov_reg_imm_low:2,
+    inst_mov_reg_imm_high:3,
+    inst_add:4,
+    inst_sub:5,
+    inst_mult:6,
+    inst_div:7,
+    inst_and:8,
+    inst_or:9,
+    inst_xor:10,
+    inst_not:11,
+    inst_nand:12,
+    inst_shl:13,
+    inst_shr:14,
+    inst_jmp:15,
+    inst_jt:16,
+    inst_jf:17,
+    inst_les:18,
+    inst_grt:19,
+    inst_leq:20,
+    inst_geq:21,
+    inst_eq:22,
+    inst_in:23,
+    inst_out:24
+};
+
+var hex=func(){
+    var vec=[];
+    foreach(var i;['0','1','2','3','4','5','6','7','8','9','a','b','c','d','e','f'])
+        foreach(var j;['0','1','2','3','4','5','6','7','8','9','a','b','c','d','e','f'])
+            append(vec,i~j);
+    return func(n){
+        return vec[n];
+    }
+}();
+
+var hex32=func(n){
+    return hex(bits.u32_and(n/math.pow(2,24),0xff))~
+    hex(bits.u32_and(n/math.pow(2,16),0xff))~
+    hex(bits.u32_and(n/math.pow(2,8),0xff))~
+    hex(bits.u32_and(n,0xff));
+}
+
+var machine=func(disk_file){
+
+var reg=[];
+var reg_size=32;
+var pc=0;
+var ir=[0,0,0,0]; # 32 bit instruction word
+var mem=[];
+var mem_size=1024*1024*1; # memory size, byte
+var init=func(){
+    println("[",os.time(),"] init ",reg_size," registers.");
+    setsize(reg,reg_size); # 8 bit address wire
+    for(var i=0;i<reg_size;i+=1)
+        reg[i]=0;
+    println("[",os.time(),"] init memory, memory size: ",mem_size/1024/1024,"MB.");
+    setsize(mem,mem_size);
+    for(var i=0;i<mem_size;i+=1)
+        mem[i]=0;
+    println("[",os.time(),"] init completed.");
+}
+init();
+var load=func(){
+    var vec=split(" ",disk_file);
+    println("[",os.time(),"] loading boot from disk: ",size(vec)," byte.");
+    forindex(var i;vec)
+        mem[i]=int("0x"~vec[i]);
+    println("[",os.time(),"] loading complete.");
+}
+load();
+var ctx_info=func(){
+    var cnt=0;
+    println("pc    : 0x",hex32(pc));
+    println("instr : 0x",hex(ir[0]),hex(ir[1]),hex(ir[2]),hex(ir[3]));
+    for(var i=0;i<reg_size;i+=1){
+        print("reg[",hex(i),"]: 0x",hex32(reg[i])," ");
+        if(cnt==3){
+            print("\n");
+            cnt=0;
+        }else{
+            cnt+=1;
+        }
+    }
+}
+var exec=func(info=1){
+    while(1){
+        ir=[mem[pc],mem[pc+1],mem[pc+2],mem[pc+3]];
+        if(info)ctx_info();
+        var op=ir[0];
+        if(op==inst.inst_stop){
+            break;
+        }elsif(op==inst.inst_mov_reg_reg){
+            reg[ir[1]]=reg[ir[2]];
+        }elsif(op==inst.inst_mov_reg_imm_low){
+            reg[ir[1]]=bits.u32_and(reg[ir[1]],0xffff0000);
+            reg[ir[1]]=bits.u32_or(reg[ir[1]],ir[2]*math.pow(2,8)+ir[3]);
+        }elsif(op==inst.inst_mov_reg_imm_high){
+            reg[ir[1]]=bits.u32_and(reg[ir[1]],0x0000ffff);
+            reg[ir[1]]=bits.u32_or(reg[ir[1]],ir[2]*math.pow(2,24)+ir[3]*math.pow(2,16));
+        }elsif(op==inst.inst_add){
+            reg[ir[1]]=bits.u32_and(reg[ir[2]]+reg[ir[3]],0xffffffff);
+        }elsif(op==inst.inst_sub){
+            reg[ir[1]]=bits.u32_and(reg[ir[2]]-reg[ir[3]],0xffffffff);
+        }elsif(op==inst.inst_mult){
+            reg[ir[1]]=bits.u32_and(reg[ir[2]]*reg[ir[3]],0xffffffff);
+        }elsif(op==inst.inst_div){
+            reg[ir[1]]=bits.u32_and(reg[ir[2]]/reg[ir[3]],0xffffffff);
+        }elsif(op==inst.inst_and){
+            reg[ir[1]]=bits.u32_and(reg[ir[2]],reg[ir[3]]);
+        }elsif(op==inst.inst_or){
+            reg[ir[1]]=bits.u32_or(reg[ir[2]],reg[ir[3]]);
+        }elsif(op==inst.inst_xor){
+            reg[ir[1]]=bits.u32_xor(reg[ir[2]],reg[ir[3]]);
+        }elsif(op==inst.inst_not){
+            reg[ir[1]]=bits.u32_not(reg[ir[2]]);
+        }elsif(op==inst.inst_nand){
+            reg[ir[1]]=bits.u32_nand(reg[ir[2]],reg[ir[3]]);
+        }elsif(op==inst.inst_shl){
+            reg[ir[1]]=bits.u32_and(reg[ir[2]]*math.pow(2,reg[ir[3]]),0xffffffff);
+        }elsif(op==inst.inst_shr){
+            reg[ir[1]]=bits.u32_and(reg[ir[2]]/math.pow(2,reg[ir[3]]),0xffffffff);
+        }elsif(op==inst.inst_jmp){
+            pc=reg[ir[1]];
+        }elsif(op==inst.inst_jt){
+            pc=reg[ir[1]]?reg[ir[2]]:pc;
+        }elsif(op==inst.inst_jf){
+            pc=reg[ir[1]]?pc:reg[ir[2]];
+        }elsif(op==inst.inst_les){
+            reg[ir[1]]=reg[ir[2]]<reg[ir[3]];
+        }elsif(op==inst.inst_grt){
+            reg[ir[1]]=reg[ir[2]]>reg[ir[3]];
+        }elsif(op==inst.inst_leq){
+            reg[ir[1]]=reg[ir[2]]<=reg[ir[3]];
+        }elsif(op==inst.inst_geq){
+            reg[ir[1]]=reg[ir[2]]>=reg[ir[3]];
+        }elsif(op==inst.inst_eq){
+            reg[ir[1]]=reg[ir[2]]==reg[ir[3]];
+        }elsif(op==inst.inst_in){
+            reg[0]=0; # unfinished
+        }elsif(op==inst.inst_out){
+            println("reg[",ir[1],"]: 0x",hex32(reg[ir[1]]));
+        }
+        pc+=4;
+    }
+};
+    return {exec:exec};
+}(
+hex(inst.inst_mov_reg_imm_high)~" 01 ca fe "~ # reg[1]=0xcafe0000
+hex(inst.inst_mov_reg_imm_low)~" 01 ba be "~  # reg[1]=0xcafebabe
+hex(inst.inst_out)~" 01 00 00 "~              # output reg[1]
+hex(inst.inst_mov_reg_imm_low)~" 02 00 20 "~  # reg[2]=0x10
+hex(inst.inst_jmp)~" 02 00 00 "~              # jmp *reg[2]
+hex(inst.inst_out)~" 01 00 00 "~
+hex(inst.inst_out)~" 01 00 00 "~
+hex(inst.inst_out)~" 01 00 00 "~
+hex(inst.inst_out)~" 01 00 00 "~
+hex(inst.inst_out)~" 01 00 00 "~              # should jump here
+hex(inst.inst_jf)~" 01 02 00 "~               # should not jump
+hex(inst.inst_mov_reg_imm_low)~" 03 00 04 "~  # reg[3]=4
+hex(inst.inst_mov_reg_imm_low)~" 04 00 00 "~  # reg[4]=0
+hex(inst.inst_out)~" 04 00 00 "~              # output reg[4]
+hex(inst.inst_grt)~" 00 04 03 "~              # reg[0]=reg[4]>reg[3]
+hex(inst.inst_mov_reg_imm_low)~" 05 00 30 "~  # reg[5]=0x2c
+hex(inst.inst_mov_reg_imm_low)~" 06 00 01 "~  # reg[6]=1
+hex(inst.inst_add)~" 04 04 06 "~              # reg[4]+=reg[6]
+hex(inst.inst_jf)~" 00 05 00 "                # jmp *reg[5] if reg[0]!=true
+);
+
+machine.exec(0);
\ No newline at end of file
diff --git a/test/md5compare.nas b/test/md5compare.nas
index 06d8fb8..33eaf37 100644
--- a/test/md5compare.nas
+++ b/test/md5compare.nas
@@ -51,6 +51,7 @@ var filechecksum=func(){
         "./test/leetcode1319.nas",  "./test/lexer.nas",
         "./test/life.nas",          "./test/loop.nas",
         "./test/mandel.nas",        "./test/mandelbrot.nas",
+        "./test/mcpu.nas",
         "./test/md5.nas",           "./test/md5compare.nas",
         "./test/module_test.nas",   "./test/nasal_test.nas",
         "./test/occupation.nas",    "./test/pi.nas",