Shuffle lane->dag data mapping

Based on suggestion from PR #13
ifdefelse · Nov 16, 2018 · e9745a4 · e9745a4
1 parent b45389b
commit e9745a4
Show file tree

Hide file tree

Showing 2 changed files with 5 additions and 3 deletions.
diff --git a/README.md b/README.md
@@ -329,10 +329,12 @@ void progPowLoop(
     // Lanes can execute in parallel and will be convergent
     for (int l = 0; l < PROGPOW_LANES; l++)
     {
-        // global load to sequential locations
+        // global load to the 256 byte DAG entry
+        // every lane can access every part of the entry
         uint32_t data_g[PROGPOW_DAG_LOADS];
+        uint32_t offset_l = offset_g * PROGPOW_LANES + (l ^ loop) % PROGPOW_LANES;
         for (int i = 0; i < PROGPOW_DAG_LOADS; i++)
-            data_g[i] = dag[(offset_g*PROGPOW_LANES + l)*PROGPOW_DAG_LOADS + i];
+            data_g[i] = dag[offset_l * PROGPOW_DAG_LOADS + i];
 
         // initialize the seed and mix destination sequence
         int mix_seq_dst[PROGPOW_REGS];

diff --git a/libprogpow/ProgPow.cpp b/libprogpow/ProgPow.cpp
@@ -134,7 +134,7 @@ std::string ProgPow::getKern(uint64_t block_number, kernel_t kern)
 		ret << "offset = share[group_id];\n";
 	}
 	ret << "offset %= PROGPOW_DAG_ELEMENTS;\n";
-	ret << "offset = offset * PROGPOW_LANES + lane_id;\n";
+	ret << "offset = offset * PROGPOW_LANES + (lane_id ^ loop) % PROGPOW_LANES;\n";
     ret << "data_dag = g_dag[offset];\n";
     ret << "// hack to prevent compiler from reordering LD and usage\n";
     if (kern == KERNEL_CUDA)