clang format

lattice · Nov 30, 2023 · fc3ea42 · fc3ea42
1 parent 13075c7
commit fc3ea42
Show file tree

Hide file tree

Showing 3 changed files with 7 additions and 8 deletions.
diff --git a/include/tunable_block_reduction.h b/include/tunable_block_reduction.h
@@ -169,8 +169,7 @@ namespace quda
         // we can advance spin/block-color since this is valid
         if (param.block.z < vector_length_z && param.block.z < device::max_threads_per_block_dim(2)
             && param.block.x * param.block.y * (param.block.z + step_z) <= device::max_threads_per_block()
-            && ((param.block.z + step_z) <= max_block_z)
-            && shared_bytes <= this->maxSharedBytesPerBlock()) {
+            && ((param.block.z + step_z) <= max_block_z) && shared_bytes <= this->maxSharedBytesPerBlock()) {
           param.block.z += step_z;
           param.grid.z = (vector_length_z + param.block.z - 1) / param.block.z;
           param.shared_bytes = shared_bytes;

diff --git a/include/tunable_reduction.h b/include/tunable_reduction.h
@@ -332,8 +332,7 @@ namespace quda
 
         if (param.block.z < n_batch && param.block.z < device::max_threads_per_block_dim(2)
             && param.block.x * param.block.y * (param.block.z + 1) <= device::max_threads_per_block()
-            && param.block.z < n_batch_block_max
-            && shared_bytes <= this->maxSharedBytesPerBlock()) {
+            && param.block.z < n_batch_block_max && shared_bytes <= this->maxSharedBytesPerBlock()) {
           param.block.z++;
           param.grid.z = (n_batch + param.block.z - 1) / param.block.z;
           param.shared_bytes = shared_bytes;

diff --git a/include/tune_quda.h b/include/tune_quda.h
@@ -342,16 +342,17 @@ namespace quda {
      * @brief self-consistency check that the shared memory is set
      * correctly (e.g., check that block size has been correctly
      * factored in when set setting shared_bytes)
-    */
-    void checkSharedBytes(const TuneParam &tp) const {
+     */
+    void checkSharedBytes(const TuneParam &tp) const
+    {
       auto tp2 = TuneParam(tp);
       auto expected = setSharedBytes(tp2);
       if (tp.shared_bytes < expected)
         errorQuda("Shared bytes %u insufficient (expected %u)", tp.shared_bytes, expected);
 
       if (sharedBytesPerThread() && sharedBytesPerBlock(tp))
-        errorQuda("Not supported: non-zero shared bytes per thread (%u) and per block (%u)",
-                  sharedBytesPerThread(), sharedBytesPerBlock(tp));
+        errorQuda("Not supported: non-zero shared bytes per thread (%u) and per block (%u)", sharedBytesPerThread(),
+                  sharedBytesPerBlock(tp));
     }
 
     /**