change private arrays to use new dpex api

Author: adarshyoga

dpbench/benchmarks/default/gpairs/gpairs_numba_dpex_k.py

@@ -38,9 +38,9 @@ def count_weighted_pairs_3d_intel_no_slm_ker(
 
     n_wi = 32
 
-    dsq_mat = dpex.private.array(shape=(32 * 32), dtype=dtype)
-    w0_vec = dpex.private.array(shape=(32), dtype=dtype)
-    w1_vec = dpex.private.array(shape=(32), dtype=dtype)
+    dsq_mat = kapi.PrivateArray(shape=(32 * 32), dtype=dtype)
+    w0_vec = kapi.PrivateArray(shape=(32), dtype=dtype)
+    w1_vec = kapi.PrivateArray(shape=(32), dtype=dtype)
 
     offset0 = gr0 * n_wi * lws0 + lid0
     offset1 = gr1 * n_wi * lws1 + lid1
@@ -80,7 +80,7 @@ def count_weighted_pairs_3d_intel_no_slm_ker(
 
     # update slm_hist. Use work-item private buffer of 16 tfloat elements
     for k in range(0, slm_hist_size, private_hist_size):
-        private_hist = dpex.private.array(shape=(32), dtype=dtype)
+        private_hist = kapi.PrivateArray(shape=(32), dtype=dtype)
         for p in range(private_hist_size):
             private_hist[p] = 0.0
 

dpbench/benchmarks/default/kmeans/kmeans_numba_dpex_k.py

@@ -56,7 +56,7 @@ def groupByCluster(
         for i in range(WorkPI):
             point_id = grid * WorkPI * local_size + i * local_size + lid
             if point_id < numpoints:
-                localP = dpex.private.array(dims, dtyp)
+                localP = kapi.PrivateArray(dims, dtyp)
                 for d in range(dims):
                     localP[d] = arrayP[point_id, d]
 
@@ -179,7 +179,7 @@ def updateLabels(
         for i in range(WorkPI):
             point_id = grid * WorkPI * local_size + i * local_size + lid
             if point_id < numpoints:
-                localP = dpex.private.array(dims, dtyp)
+                localP = kapi.PrivateArray(dims, dtyp)
                 for d in range(dims):
                     localP[d] = arrayP[point_id, d]
 

dpbench/benchmarks/default/knn/knn_numba_dpex_k.py

@@ -25,7 +25,7 @@ def _knn_kernel(  # noqa: C901: TODO: can we simplify logic?
     dtype = train.dtype
     i = item.get_id(0)
     # here k has to be 5 in order to match with numpy
-    queue_neighbors = dpex.private.array(shape=(5, 2), dtype=dtype)
+    queue_neighbors = kapi.PrivateArray(shape=(5, 2), dtype=dtype)
 
     for j in range(k):
         x1 = train[j]