Start of multithreaded matrix multiply.

src/test/matmul-mt.s

@@ -0,0 +1,124 @@
+/*******************************************************************************
+ Harptools by Chad D. Kersey, Summer 2011
+********************************************************************************
+
+ Sample HARP assmebly program.
+
+*******************************************************************************/
+/* Matrix multiply: find matrix product */
+.def THREADS 2
+
+.align 4096
+.perm x
+.entry
+.global
+entry: ldi %r0, matrix_a;
+       ldi %r1, #3;
+       jali %r5, matgen;
+
+       ldi %r0, matrix_b;
+       ldi %r1, #3;
+       jali %r5, matgen;
+
+       ldi %r0, matrix_a;
+       ldi %r1, matrix_b;
+       ldi %r2, matrix_r;
+       ldi %r3, #3;
+       jali %r5, matmul;
+
+       ldi %r0, #64;
+       ldi %r1, matrix_r;
+ploop: ld %r7, %r1, #0;
+       jali %r5, printfloat;
+       subi %r0, %r0, #1;
+       addi %r1, %r1, __WORD;
+       rtop @p0, %r0;
+ @p0 ? jmpi ploop;
+
+       trap;
+
+/* Generate a random 2^Nx2^N matrix at %r0, with N in %r1 */
+matgen: ldi %r2, #0;
+        st %r5, %r2, retaddr;
+        ldi %r2, #1;
+        shli %r2, %r2, (`__WORD + 1); /* Multiply r0 by 2*__WORD */
+        shl %r2, %r2, %r1;
+        ori %r3, %r0, #0;
+
+mgloop: jali %r5, randf;
+        st %r0, %r3, #0;
+        addi %r3, %r3, __WORD;
+        subi %r2, %r2, #1;
+        rtop @p0, %r2;
+  @p0 ? jmpi mgloop;
+        
+        ldi %r2, #0;
+        ld %r5, %r2, retaddr;
+        jmpr %r5;        
+
+/* Write the matrix product of square matrix at (%r0) and (%r1) to (%r2). The
+   size of these matrices is 2^Nx2^N, where N = %r3 */
+
+matmul: ldi %r4, #1;
+        ldi %r10, (`__WORD); /* ` is the log base 2 operator */
+        shl %r4, %r4, %r3;
+        add %r10, %r10, %r3;
+        ldi %r14, #1;
+        shl %r14, %r14, %r10;
+
+        divi %r17, %r14, THREADS; /* Spawn threads */
+sloop:  
+
+        jmpr %r5;
+
+/* One thread of matrix multiplication. Expected register values at start:
+ *   %r0 - matrix a pointer (plus offset)
+ *   %r1 - matrix b pointer
+ *   %r2 - destination matrix pointer (plus offset)
+ *   %r17 - row count
+ */
+matmulthd: ldi %r9, #0; /* result row: %r9 */
+rloop:     ldi %r6, #0; /* result col: %r6 */
+cloop:     shli %r16, %r6, (`__WORD);
+           shl %r15, %r9, %r10;
+
+           add %r11, %r15, %r0;
+           add %r12, %r16, %r1;
+           ldi %r13, #0;
+
+           ldi %r8, #0 /* dot prod position: %r8 */
+iloop:     ld %r7, %r11, #0;
+           ld %r17, %r12, #0;
+           fmul %r7, %r7, %r17
+           fadd %r13, %r13, %r7;
+
+           addi %r8, %r8, #1;
+           addi %r11, %r11, __WORD;
+           add %r12, %r12, %r14;
+           sub %r7, %r8, %r4;
+           rtop @p0, %r7;
+     @p0 ? jmpi iloop;
+
+           add %r15, %r15, %r16;
+           add %r15, %r15, %r2;
+           st %r13, %r15, #0;
+
+           addi %r6, %r6, #1;
+           sub %r7, %r6, %r4;
+           rtop @p0, %r7;
+     @p0 ? jmpi cloop;
+
+           addi %r9, %r9, #1;
+           sub %r7, %r9, %r17;
+           rtop @p0, %r7;
+     @p0 ? jmpi rloop;
+
+           jmprt %r5;
+
+.align 4096
+.perm rw
+matrix_a: .space 64;
+matrix_b: .space 64;
+matrix_r: .space 64
+
+retaddr: .word 0