gkeyllorg · JunoRavin · Aug 5, 2026 · Jul 3, 2026 · Jul 3, 2026 · Jul 4, 2026
diff --git a/core/unit/ctest_dg_array_mask.c b/core/unit/ctest_dg_array_mask.c
@@ -106,7 +106,7 @@ void test_mask_advance_threshold(bool use_gpu)
   gkyl_array_copy(arr, arr_ho);
 
   double global_max = 0.0;
-  gkyl_array_reduce(&global_max, arr, GKYL_MAX);
+  gkyl_array_reduce(&global_max, arr_ho, GKYL_MAX);
   gkyl_dg_array_mask_advance_threshold(mask, global_max);
   gkyl_dg_array_mask_advance(mask, arr);
 
@@ -159,7 +159,7 @@ void test_mask_advance_all_below(bool use_gpu)
   gkyl_array_copy(arr, arr_ho);
 
   double global_max = 0.0;
-  gkyl_array_reduce(&global_max, arr, GKYL_MAX);
+  gkyl_array_reduce(&global_max, arr_ho, GKYL_MAX);
   gkyl_dg_array_mask_advance_threshold(mask, global_max);
   gkyl_dg_array_mask_advance(mask, arr);
 
@@ -205,7 +205,7 @@ void test_mask_advance_all_above(bool use_gpu)
   gkyl_array_copy(arr, arr_ho);
 
   double global_max = 0.0;
-  gkyl_array_reduce(&global_max, arr, GKYL_MAX);
+  gkyl_array_reduce(&global_max, arr_ho, GKYL_MAX);
   gkyl_dg_array_mask_advance_threshold(mask, global_max);
   gkyl_dg_array_mask_advance(mask, arr);
 
@@ -255,7 +255,7 @@ void test_mask_advance_negative_values(bool use_gpu)
   gkyl_array_copy(arr, arr_ho);
 
   double global_max = 0.0;
-  gkyl_array_reduce(&global_max, arr, GKYL_MAX);
+  gkyl_array_reduce(&global_max, arr_ho, GKYL_MAX);
   gkyl_dg_array_mask_advance_threshold(mask, global_max);
   gkyl_dg_array_mask_advance(mask, arr);
 
@@ -312,7 +312,7 @@ void test_mask_advance_greater_than_threshold(bool use_gpu)
   gkyl_array_copy(arr, arr_ho);
 
   double global_max = 0.0;
-  gkyl_array_reduce(&global_max, arr, GKYL_MAX);
+  gkyl_array_reduce(&global_max, arr_ho, GKYL_MAX);
   gkyl_dg_array_mask_advance_threshold(mask, global_max);
   gkyl_dg_array_mask_advance(mask, arr);
 
@@ -365,7 +365,7 @@ void test_mask_advance_greater_than_all_above(bool use_gpu)
   gkyl_array_copy(arr, arr_ho);
 
   double global_max = 0.0;
-  gkyl_array_reduce(&global_max, arr, GKYL_MAX);
+  gkyl_array_reduce(&global_max, arr_ho, GKYL_MAX);
   gkyl_dg_array_mask_advance_threshold(mask, global_max);
   gkyl_dg_array_mask_advance(mask, arr);
 
@@ -411,7 +411,7 @@ void test_mask_advance_greater_than_all_below(bool use_gpu)
   gkyl_array_copy(arr, arr_ho);
 
   double global_max = 0.0;
-  gkyl_array_reduce(&global_max, arr, GKYL_MAX);
+  gkyl_array_reduce(&global_max, arr_ho, GKYL_MAX);
   gkyl_dg_array_mask_advance_threshold(mask, global_max);
   gkyl_dg_array_mask_advance(mask, arr);
 
@@ -461,7 +461,7 @@ void test_mask_advance_greater_than_negative_values(bool use_gpu)
   gkyl_array_copy(arr, arr_ho);
 
   double global_max = 0.0;
-  gkyl_array_reduce(&global_max, arr, GKYL_MAX);
+  gkyl_array_reduce(&global_max, arr_ho, GKYL_MAX);
   gkyl_dg_array_mask_advance_threshold(mask, global_max);
   gkyl_dg_array_mask_advance(mask, arr);
 
@@ -520,7 +520,7 @@ void test_mask_eval(bool use_gpu)
   gkyl_array_copy(arr, arr_ho);
 
   double global_max = 0.0;
-  gkyl_array_reduce(&global_max, arr, GKYL_MAX);
+  gkyl_array_reduce(&global_max, arr_ho, GKYL_MAX);
   gkyl_dg_array_mask_advance_threshold(mask, global_max);
   gkyl_dg_array_mask_advance(mask, arr);
 
@@ -623,7 +623,7 @@ void test_mask_scale_by_cell(bool use_gpu)
   gkyl_array_copy(arr, arr_ho);
 
   double global_max = 0.0;
-  gkyl_array_reduce(&global_max, arr, GKYL_MAX);
+  gkyl_array_reduce(&global_max, arr_ho, GKYL_MAX);
   gkyl_dg_array_mask_advance_threshold(mask, global_max);
   gkyl_dg_array_mask_advance(mask, arr);
 
@@ -781,7 +781,7 @@ void test_mask_advance_frac_threshold(bool use_gpu)
   }
   gkyl_array_copy(arr, arr_ho);
   double global_max = 0.0;
-  gkyl_array_reduce(&global_max, arr, GKYL_MAX);
+  gkyl_array_reduce(&global_max, arr_ho, GKYL_MAX);
   gkyl_dg_array_mask_advance_threshold(mask, global_max);
   gkyl_dg_array_mask_advance(mask, arr);
 
@@ -839,7 +839,7 @@ void test_mask_advance_frac_threshold_greater(bool use_gpu)
   gkyl_array_copy(arr, arr_ho);
 
   double global_max = 0.0;
-  gkyl_array_reduce(&global_max, arr, GKYL_MAX);
+  gkyl_array_reduce(&global_max, arr_ho, GKYL_MAX);
   gkyl_dg_array_mask_advance_threshold(mask, global_max);
   gkyl_dg_array_mask_advance(mask, arr);
 
@@ -929,7 +929,7 @@ void test_mask_advance_frac_threshold_spatial(bool use_gpu)
   gkyl_array_copy(arr, arr_ho);
 
   double global_max = 0.0;
-  gkyl_array_reduce(&global_max, arr, GKYL_MAX);
+  gkyl_array_reduce(&global_max, arr_ho, GKYL_MAX);
   gkyl_dg_array_mask_advance_threshold(mask, global_max);
   gkyl_dg_array_mask_advance(mask, arr);
 
@@ -1019,7 +1019,7 @@ void test_mask_advance_frac_threshold_spatial_greater(bool use_gpu)
   gkyl_array_copy(arr, arr_ho);
 
   double global_max = 0.0;
-  gkyl_array_reduce(&global_max, arr, GKYL_MAX);
+  gkyl_array_reduce(&global_max, arr_ho, GKYL_MAX);
   gkyl_dg_array_mask_advance_threshold(mask, global_max);
   gkyl_dg_array_mask_advance(mask, arr);
 
@@ -1105,7 +1105,7 @@ void test_mask_advance_threshold_ext_range(bool use_gpu, int ncell, int nghost_c
   gkyl_array_copy(arr, arr_ho);
 
   double global_max = 0.0;
-  gkyl_array_reduce(&global_max, arr, GKYL_MAX);
+  gkyl_array_reduce(&global_max, arr_ho, GKYL_MAX);
   gkyl_dg_array_mask_advance_threshold(mask, global_max);
   gkyl_dg_array_mask_advance(mask, arr);
 
@@ -1209,7 +1209,7 @@ void test_mask_advance_frac_threshold_ext_range(bool use_gpu)
   gkyl_array_copy(arr, arr_ho);
 
   double global_max = 0.0;
-  gkyl_array_reduce(&global_max, arr, GKYL_MAX);
+  gkyl_array_reduce(&global_max, arr_ho, GKYL_MAX);
   gkyl_dg_array_mask_advance_threshold(mask, global_max);
   gkyl_dg_array_mask_advance(mask, arr);
 
@@ -1353,7 +1353,7 @@ void test_mask_advance_frac_threshold_spatial_ext_range(bool use_gpu)
   gkyl_array_copy(arr, arr_ho);
 
   double global_max = 0.0;
-  gkyl_array_reduce(&global_max, arr, GKYL_MAX);
+  gkyl_array_reduce(&global_max, arr_ho, GKYL_MAX);
   gkyl_dg_array_mask_advance_threshold(mask, global_max);
   gkyl_dg_array_mask_advance(mask, arr);
 

diff --git a/core/zero/range.c b/core/zero/range.c
@@ -290,6 +290,9 @@ gkyl_range_deflate(struct gkyl_range* srng,
   SET_SUB_RANGE(srng->flags);
 
   // for CUDA ops
+  struct gkyl_range sub_range;
+  gkyl_range_init(&sub_range, srng->ndim, srng->lower, srng->upper);
+  gkyl_copy_long_arr(GKYL_MAX_DIM+1, sub_range.ac, srng->iac);
   srng->nthreads = GKYL_DEFAULT_NUM_THREADS;
   srng->nblocks = srng->volume/srng->nthreads + 1;
 }

diff --git a/core/zero/tensor_field.c b/core/zero/tensor_field.c
@@ -109,9 +109,9 @@ gkyl_tensor_field_cu_dev_new(size_t rank, size_t ndim, size_t size, const enum g
   // so that the whole tfld->on_dev struct can be passed to a device kernel
   tfld->on_dev = gkyl_cu_malloc(sizeof(struct gkyl_tensor_field));
   gkyl_cu_memcpy(tfld->on_dev, tfld, sizeof(struct gkyl_tensor_field), GKYL_CU_MEMCPY_H2D);
-  // set device-side data pointer in tfld->on_dev to tfld->data 
+  // set device-side data pointer in tfld->on_dev to tfld->data->on-dev
   // (which is the host-side pointer to the device data)
-  gkyl_cu_memcpy(&((tfld->on_dev)->tdata), &tfld->tdata, sizeof(void*), GKYL_CU_MEMCPY_H2D);
+  gkyl_cu_memcpy(&((tfld->on_dev)->tdata), &tfld->tdata->on_dev, sizeof(void*), GKYL_CU_MEMCPY_H2D);
 
   return tfld;
 }

diff --git a/gyrokinetic/apps/gyrokinetic.c b/gyrokinetic/apps/gyrokinetic.c
@@ -155,7 +155,6 @@ gyrokinetic_deflate_delta_ts(struct gkyl_gyrokinetic_app* app, struct gkyl_array
   gkyl_translate_dim_release(transd_2d_1d);
   gkyl_array_release(buffer_perp);
   gkyl_array_copy(app->delta_ts_x_up, delta_ts_x_dev);
-  gkyl_array_release(delta_ts_x_dev);
 }
 
 gkyl_gyrokinetic_app*

diff --git a/gyrokinetic/ker/inflate_surf/gkyl_inflate_surf_kernels.h b/gyrokinetic/ker/inflate_surf/gkyl_inflate_surf_kernels.h
diff --git a/gyrokinetic/ker/inflate_surf/inflate_surf_1x1v_ser_p1.c b/gyrokinetic/ker/inflate_surf/inflate_surf_1x1v_ser_p1.c
diff --git a/gyrokinetic/ker/inflate_surf/inflate_surf_1x1v_ser_p2.c b/gyrokinetic/ker/inflate_surf/inflate_surf_1x1v_ser_p2.c
diff --git a/gyrokinetic/ker/inflate_surf/inflate_surf_1x2v_ser_p1.c b/gyrokinetic/ker/inflate_surf/inflate_surf_1x2v_ser_p1.c
diff --git a/gyrokinetic/ker/inflate_surf/inflate_surf_1x2v_ser_p2.c b/gyrokinetic/ker/inflate_surf/inflate_surf_1x2v_ser_p2.c
diff --git a/gyrokinetic/ker/inflate_surf/inflate_surf_1x_ser_p1.c b/gyrokinetic/ker/inflate_surf/inflate_surf_1x_ser_p1.c
diff --git a/gyrokinetic/ker/inflate_surf/inflate_surf_1x_ser_p2.c b/gyrokinetic/ker/inflate_surf/inflate_surf_1x_ser_p2.c
diff --git a/gyrokinetic/ker/inflate_surf/inflate_surf_2x2v_ser_p1.c b/gyrokinetic/ker/inflate_surf/inflate_surf_2x2v_ser_p1.c
diff --git a/gyrokinetic/ker/inflate_surf/inflate_surf_2x2v_ser_p2.c b/gyrokinetic/ker/inflate_surf/inflate_surf_2x2v_ser_p2.c
diff --git a/gyrokinetic/ker/inflate_surf/inflate_surf_2x_ser_p1.c b/gyrokinetic/ker/inflate_surf/inflate_surf_2x_ser_p1.c
diff --git a/gyrokinetic/ker/inflate_surf/inflate_surf_2x_ser_p2.c b/gyrokinetic/ker/inflate_surf/inflate_surf_2x_ser_p2.c