[Genabel-commits] r1764 - pkg/OmicABELnoMM/src

[noreply at r-forge.r-project.org]

Author: afrank
Date: 2014-07-15 10:57:10 +0200 (Tue, 15 Jul 2014)
New Revision: 1764

Modified:
   pkg/OmicABELnoMM/src/AIOwrapper.h
   pkg/OmicABELnoMM/src/Algorithm.cpp
Log:
Code Cleanup with one performance improvement.

Modified: pkg/OmicABELnoMM/src/AIOwrapper.h
===================================================================
--- pkg/OmicABELnoMM/src/AIOwrapper.h	2014-07-15 08:17:29 UTC (rev 1763)
+++ pkg/OmicABELnoMM/src/AIOwrapper.h	2014-07-15 08:57:10 UTC (rev 1764)
@@ -173,7 +173,7 @@
 
         void * fgls_malloc_impl( const char* file, long line, size_t size );
 
-            public: type_fileh FHandler;
+        type_fileh FHandler;
         type_fileh* Fhandler;
 
         FILE* fp_Ar;

Modified: pkg/OmicABELnoMM/src/Algorithm.cpp
===================================================================
--- pkg/OmicABELnoMM/src/Algorithm.cpp	2014-07-15 08:17:29 UTC (rev 1763)
+++ pkg/OmicABELnoMM/src/Algorithm.cpp	2014-07-15 08:57:10 UTC (rev 1764)
@@ -479,8 +479,6 @@
             get_ticks(end_tick);
             out.acc_gemm += ticks2sec(end_tick,start_tick2);
 
-            int aL_idx =  0*l * n;
-            int aR_idx = 0* r * n * a_block_size;
 
             get_ticks(start_tick3);
             for (int jj = 0; jj < y_block_size; jj++)
@@ -491,9 +489,9 @@
 
                 get_ticks(start_tick2);
 
-                copy_vec(backupAL, &AL[aL_idx], n * l);//try to remove!
+                copy_vec(backupAL, AL, n * l);//try to remove!
 
-                replace_with_zeros(&y_nan_idxs[jj], &AL[aL_idx], n, l, 1);
+                replace_with_zeros(&y_nan_idxs[jj], AL, n, l, 1);
 
 
                 get_ticks(end_tick);//2%
@@ -503,30 +501,30 @@
 
                 //! Generate Stl
                 cblas_ssyrk(CblasColMajor, CblasUpper, CblasTrans,
-                            l, n, 1.0, &AL[aL_idx], lda, 0.0, Stl, l);
+                            l, n, 1.0, AL, lda, 0.0, Stl, l);
 
                 get_ticks(end_tick);
                 out.acc_stl += ticks2sec(end_tick,start_tick2);
 
 
-                get_ticks(start_tick2);
+//                get_ticks(start_tick2);
 
-                copy_vec(backupAR,&AR[aR_idx], n*r*a_block_size);//!10%//try to remove!
+                //copy_vec(backupAR,AR, n*r*a_block_size);//!10%//try to remove!
 
 
-                replace_with_zeros(&y_nan_idxs[jj], backupAR,  n, r, a_block_size);
+                //replace_with_zeros(&y_nan_idxs[jj], AR,  n, r, a_block_size);
 
 
 
-                get_ticks(end_tick);
-                out.acc_other += ticks2sec(end_tick,start_tick2);
+//                get_ticks(end_tick);
+//                out.acc_other += ticks2sec(end_tick,start_tick2);
 
                 get_ticks(start_tick2);
 
                 //! Generate Str
                 cblas_sgemm(CblasColMajor, CblasTrans, CblasNoTrans,
-                            l, r * a_block_size, n, 1.0, &AL[aL_idx],
-                            n, &AR[aR_idx], n, 0.0, Str, l);//!45
+                            l, r * a_block_size, n, 1.0, AL,
+                            n, AR, n, 0.0, Str, l);//!45
 
                 get_ticks(end_tick);
                 out.acc_str += ticks2sec(end_tick,start_tick2);
@@ -546,14 +544,22 @@
                 for (int ii= 0; ii < a_block_size; ii++)
                 {
                     //cout << omp_get_thread_num() << endl << flush;
+
+                    get_ticks(start_tick2);
+
+                    copy_vec(&backupAR[ii*r*n],&AR[ii*r*n], n*r);//!10%//try to remove!
+                    replace_with_zeros(&y_nan_idxs[jj], &AR[ii*r*n],  n, r, 1);
+
+                    get_ticks(end_tick);
+                    out.acc_other += ticks2sec(end_tick,start_tick2);
 
 
                     get_ticks(start_tick2);
 
                     //! Generate Sbr
                     cblas_sgemm(CblasColMajor, CblasTrans, CblasNoTrans,
-                                r, r, n, 1.0, &AR[aR_idx+ii*r*n], n,
-                                &AR[aR_idx + ii * r * n], n, 0.0,
+                                r, r, n, 1.0, &AR[ii*r*n], n,
+                                &AR[ii * r * n], n, 0.0,
                                 &Sbr[ii * r * r], r);
 
 
@@ -595,7 +601,7 @@
                     {
                         #pragma omp critical
                         {
-                            check_result(AL, &AR[aR_idx+ii*r*n], n, p,
+                            check_result(AL, &AR[ii*r*n], n, p,
                                          1, r, &Y[jj*n], &Ay[ii*p]);
                         }
                     }