Update cpuinfo (#2379)

* update cpuinfo

* update

* lint

* lint

* lint

* install numactl

* update num workers

* skip if gpu

---------

Co-authored-by: Ankith Gunapal <agunapal@ischool.Berkeley.edu>
Co-authored-by: Mark Saroufim <marksaroufim@fb.com>

test/pytest/test_example_intel_extension_for_pytorch.py

@@ -5,6 +5,7 @@
 import pytest
 import requests
 import test_utils
+import torch
 from test_handler import run_inference_using_url_with_data
 
 REPO_ROOT = os.path.join(os.path.dirname(os.path.abspath(__file__)), "../../")
@@ -83,7 +84,8 @@ def scale_workers_with_core_pinning(scaled_num_workers):
 
 
 @pytest.mark.skipif(
-    not ipex_xeon_run_cpu_available,
+    not ipex_xeon_run_cpu_available
+    or ((torch.cuda.device_count() > 0) and torch.cuda.is_available()),
     reason="Make sure intel-extension-for-pytorch is installed and torch.backends.xeon.run_cpu is available",
 )
 def test_single_worker_affinity():
@@ -102,16 +104,16 @@ def test_single_worker_affinity():
     ), "single-worker inference with core pinning failed"
 
     affinity = get_worker_affinity(num_workers, worker_idx)
-    print("affinity: ", affinity)
     assert affinity in open(TS_LOG).read(), "workers are not correctly pinned to cores"
 
 
 @pytest.mark.skipif(
-    not ipex_xeon_run_cpu_available,
+    not ipex_xeon_run_cpu_available
+    or ((torch.cuda.device_count() > 0) and torch.cuda.is_available()),
     reason="Make sure intel-extension-for-pytorch is installed and torch.backends.xeon.run_cpu is available",
 )
 def test_multi_worker_affinity():
-    num_workers = 4
+    num_workers = 2
     setup_torchserve()
     requests.post(
         "http://localhost:8081/models?initial_workers={}&synchronous=true&url=resnet-18.mar".format(
@@ -132,19 +134,20 @@ def test_multi_worker_affinity():
 
 
 @pytest.mark.skipif(
-    not ipex_xeon_run_cpu_available,
+    not ipex_xeon_run_cpu_available
+    or ((torch.cuda.device_count() > 0) and torch.cuda.is_available()),
     reason="Make sure intel-extension-for-pytorch is installed and torch.backends.xeon.run_cpu is available",
 )
 def test_worker_scale_up_affinity():
-    initial_num_workers = 2
+    initial_num_workers = 1
     setup_torchserve()
     requests.post(
         "http://localhost:8081/models?initial_workers={}&synchronous=true&url=resnet-18.mar".format(
             initial_num_workers
         )
     )
 
-    scaled_up_num_workers = 4
+    scaled_up_num_workers = 2
     response = scale_workers_with_core_pinning(scaled_up_num_workers)
     resnet18_list = json.loads(response.content)
     assert (
@@ -164,19 +167,20 @@ def test_worker_scale_up_affinity():
 
 
 @pytest.mark.skipif(
-    not ipex_xeon_run_cpu_available,
+    not ipex_xeon_run_cpu_available
+    or ((torch.cuda.device_count() > 0) and torch.cuda.is_available()),
     reason="Make sure intel-extension-for-pytorch is installed and torch.backends.xeon.run_cpu is available",
 )
 def test_worker_scale_down_affinity():
-    initial_num_workers = 4
+    initial_num_workers = 2
     setup_torchserve()
     requests.post(
         "http://localhost:8081/models?initial_workers={}&synchronous=true&url=resnet-18.mar".format(
             initial_num_workers
         )
     )
 
-    scaled_down_num_workers = 2
+    scaled_down_num_workers = 1
     response = scale_workers_with_core_pinning(scaled_down_num_workers)
     resnet18_list = json.loads(response.content)
     assert (