First (not quite working) floating point benchmark.

src/instruction.cpp

@@ -159,6 +159,9 @@ void Instruction::executeOn(Core &c) {
       case SHL: reg[rdest] = reg[rsrc[0]] << reg[rsrc[1]];
                 reg[rdest].trunc(wordSz);
                 break;
+      case SHR: reg[rdest] = reg[rsrc[0]] >> reg[rsrc[1]];
+                reg[rdest].trunc(wordSz);
+                break;
       case MOD: if (reg[rsrc[1]] == 0) throw DomainException();
                 reg[rdest] = reg[rsrc[0]] % reg[rsrc[1]];
                 break;

src/obj.cpp

@@ -94,9 +94,9 @@ static uint64_t readParenExpression(const string &s, const map<string, Word> &d,
   exit(1);
 }
 
-int lexerBits;
+int lexerBytes;
 Obj *AsmReader::read(std::istream &input) {
-  lexerBits = wordSize;
+  lexerBytes = wordSize;
   FlexLexer *f = new yyFlexLexer(&input);
   Obj *o = new Obj();
   std::vector<Chunk>::reverse_iterator cur;

src/scanner.lex

@@ -16,14 +16,14 @@
 #include "include/asm-tokens.h"
 #include "include/harpfloat.h"
 
-extern int lexerBits;
+extern int lexerBytes;
 static int64_t read_number(const char *s) {
   while (!isdigit(*s) && *s != '-' && *s != '+') s++;
 
   if (strchr(s, 'f') || strchr(s, '.')) {
     double d;
     sscanf(s, "%f", &d);
-    return Harp::Word_u(Harp::Float(d, lexerBits));
+    return Harp::Word_u(Harp::Float(d, lexerBytes));
   } else {
     long long u;
     sscanf(s, "%lli", &u);

src/test/Makefile

@@ -4,11 +4,11 @@ HARPEM  = ../harptool -E
 HARPDIS = ../harptool -D
 4BARCH  = 4b16/16/2
 
-all: simple.bin sieve.bin 2thread.bin simple.4b.bin sieve.4b.bin 2thread.4b.bin bubble.bin bubble.4b.bin
+all: simple.bin sieve.bin 2thread.bin simple.4b.bin sieve.4b.bin 2thread.4b.bin bubble.bin bubble.4b.bin dotprod.bin dotprod.4b.bin
 
-run: simple.out sieve.out 2thread.out simple.4b.out sieve.4b.out 2thread.4b.out bubble.out bubble.4b.out
+run: simple.out sieve.out 2thread.out simple.4b.out sieve.4b.out 2thread.4b.out bubble.out bubble.4b.out dotprod.out dotprod.4b.out
 
-disas: simple.d sieve.d 2thread.d simple.4b.d sieve.4b.d 2thread.4b.d bubble.d bubble.4b.d
+disas: simple.d sieve.d 2thread.d simple.4b.d sieve.4b.d 2thread.4b.d bubble.d bubble.4b.d dotprod.d dotprod.4b.d
 
 %.4b.out : %.4b.bin
 	$(HARPEM) -a $(4BARCH) -c $< > $@
@@ -34,12 +34,18 @@ simple.bin : boot.HOF lib.HOF simple.HOF
 sieve.bin : boot.HOF lib.HOF sieve.HOF
 	$(HARPLD) -o $@ $^
 
+dotprod.bin : boot.HOF lib.HOF dotprod.HOF
+	$(HARPLD) -o $@ $^
+
 simple.4b.bin : boot.4b.HOF lib.4b.HOF simple.4b.HOF
 	$(HARPLD) --arch $(4BARCH) -o $@ $^
 
 sieve.4b.bin : boot.4b.HOF lib.4b.HOF sieve.4b.HOF
 	$(HARPLD) --arch $(4BARCH) -o $@ $^
 
+dotprod.4b.bin : boot.4b.HOF lib.4b.HOF sieve.4b.HOF
+	$(HARPLD) --arch $(4BARCH) -o $@ $^
+
 %.4b.bin : %.4b.HOF
 	$(HARPLD) --arch $(4BARCH) -o $@ $<
 

src/test/dotprod.s

@@ -12,7 +12,13 @@
 .perm x
 .entry
 .global
-entry: /* . . . */
+entry: ldi %r0, array_a;
+       ldi %r1, array_b;
+       ldi %r2, #1;
+       jali %r5, dotprod
+
+       ori %r7, %r0, #0;
+       jali %r5, printhex;
 
        trap;
  
@@ -20,14 +26,16 @@ entry: /* . . . */
    length in r2 */
 dotprod: ldi %r3, #0;
 dploop:  ld %r4, %r0, #0;
-         ld %r5, %r1, #0;
+         ld %r6, %r1, #0;
          subi %r2, %r2, #1;
          addi %r0, %r0, __WORD;
          addi %r1, %r1, __WORD;
          rtop @p0, %r2;
-         fadd %r4, %r4, %r5;
+         fadd %r4, %r4, %r6;
          fadd %r3, %r3, %r4;
    @p0 ? jmpi dploop;
+         ori %r0, %r3, #0;
+         jmpr %r5;
 
 .perm rw
 

src/test/lib.s

@@ -7,6 +7,24 @@
 *******************************************************************************/
 /* Library: print decimals and strings! */
 .perm x
+
+.global
+printhex:    ldi %r8, (__WORD * 8);
+             ldi %r11, #1;
+             shli %r11, %r11, (__WORD*8 - 1);
+printhex_l1: subi %r8, %r8, #4;
+             shr %r9, %r7, %r8;
+             andi %r9, %r9, #15;
+             subi %r10, %r9, #9;
+             isneg @p0, %r10;
+             notp @p1, @p0;
+       @p0 ? addi %r9, %r9, #0x30
+       @p1 ? addi %r9, %r10, #0x61
+             rtop @p0, %r8;
+             st %r9, %r11, #0;
+       @p0 ? jmpi printhex_l1;
+             jmpr %r5;
+
 .global
 printdec:    ldi %r8, #1;
              shli %r8, %r8, (__WORD*8 - 1);