updated the global linear kernel to its compiled version (the kernel/…

… version is outdated).
TurakhiaLab · May 27, 2024 · b7f8e22 · b7f8e22
1 parent e452caf
commit b7f8e22
Showing 1 changed file with 17 additions and 38 deletions.
diff --git a/kernels/global_linear/kernel_global_linear.cpp b/kernels/global_linear/kernel_global_linear.cpp
@@ -51,57 +51,36 @@ void GlobalLinear::PE::Compute(char_t local_query_val,
 }
 
 void GlobalLinear::UpdatePEMaximum(
-    wavefront_scores_inf_t scores,
+    const wavefront_scores_inf_t scores,
     ScorePack (&max)[PE_NUM],
-    idx_t (&ics)[PE_NUM], idx_t (&jcs)[PE_NUM],
-    idx_t (&p_col)[PE_NUM], idx_t ck_idx,
-    bool (&predicate)[PE_NUM],
-    idx_t query_len, idx_t ref_len){
-
-//     for (int i = 0; i < PE_NUM; i++)
-//     {
-// #pragma HLS unroll
-//         if (predicate[i])
-//         {
-//             if (scores[i + 1][LAYER_MAXIMIUM] > max[i].score)
-//             {
-//                 // Notice this filtering condition compared to the Local Affine kernel. 
-//                 // if ((chunk_offset + i == query_len - 1) || (pe_offset[i] == ref_len - 1))  // last row or last column
-//                 if ( (ics[i] == query_len - 1) && (jcs[i] == ref_len - 1) )
-//                 { // So we are at the last row or last column
-//                     max[i].score = scores[i + 1][LAYER_MAXIMIUM];
-//                     max[i].row = ics[i];
-//                     max[i].col = jcs[i];
-//                     max[i].p_col = p_col[i];
-//                     max[i].ck = ck_idx;
-//                     max[i].pe = i;
-//                 }
-//             }
-//         }
-//     }
+    const idx_t chunk_row_offset, const idx_t wavefront,
+    const idx_t p_cols, const idx_t ck_idx,
+    const bool (&predicate)[PE_NUM],
+    const idx_t query_len, const idx_t ref_len){
+
+
+    // PE maximum doesn't need to be updated for the global affine kernels since 
+    // we know that the traceback starts from the bottom right element of the score matrix
 }
 
-void GlobalLinear::InitializeMaxScores(ScorePack (&max)[PE_NUM], idx_t qry_len, idx_t ref_len){
+
+void GlobalLinear::InitializeMaxScores(ScorePack (&max)[PE_NUM], idx_t qry_len, idx_t ref_len)
+{
     for (int i = 0; i < PE_NUM; i++)
     {
 #pragma HLS unroll
-        max[i].score = 0; // Need a custom struct for finding the negative infinity
-        max[i].row = 0;
-        max[i].col = 0;
+        max[i].score = NINF;
         max[i].p_col = 0;
         max[i].ck = 0;
-        max[i].pe = i;
     }
     idx_t max_pe = (qry_len - 1) % PE_NUM;
-    idx_t max_ck = (qry_len - 1)/ PE_NUM;
-    max[max_pe].score = INF;  // This is dummy score by I just represent the idea it's maximum
-    max[max_pe].row = qry_len - 1;
-    max[max_pe].col = ref_len - 1;
-    max[max_pe].p_col = (max_ck + 1) * ref_len - 1; // FIXME
+    idx_t max_ck = (qry_len - 1)  / PE_NUM;
+    max[max_pe].score = INF;
+    max[max_pe].p_col = (max_ck) * (MAX_REFERENCE_LENGTH + PE_NUM - 1) + max_pe + ref_len - 1;
     max[max_pe].ck = max_ck;
-    max[max_pe].pe = max_pe;
 }
 
+
 void GlobalLinear::Traceback::StateInit(tbp_t tbp, TB_STATE &state){
     if (tbp == TB_DIAG)
     {