support mask_rcnn for kunlun (#1890)

* support mask_rcnn for kunlun

* minor

ppdet/utils/check.py

@@ -28,12 +28,32 @@ logger = logging.getLogger(__name__)
 
 __all__ = [
     'check_gpu',
+    'check_xpu',
     'check_version',
     'check_config',
     'check_py_func',
 ]
 
 
+def check_xpu(use_xpu):
+    """
+    Log error and exit when set use_xpu=true in paddlepaddle
+    cpu/gpu version.
+    """
+    err = "Config use_xpu cannot be set as true while you are " \
+          "using paddlepaddle cpu/gpu version ! \nPlease try: \n" \
+          "\t1. Install paddlepaddle-xpu to run model on XPU \n" \
+          "\t2. Set use_xpu as false in config file to run " \
+          "model on CPU/GPU"
+
+    try:
+        if use_xpu and not fluid.is_compiled_with_xpu():
+            logger.error(err)
+            sys.exit(1)
+    except Exception as e:
+        pass
+
+
 def check_gpu(use_gpu):
     """
     Log error and exit when set use_gpu=true in paddlepaddle

tools/eval.py

@@ -27,7 +27,7 @@ import paddle.fluid as fluid
 
 from ppdet.utils.eval_utils import parse_fetches, eval_run, eval_results, json_eval_results
 import ppdet.utils.checkpoint as checkpoint
-from ppdet.utils.check import check_gpu, check_version, check_config, enable_static_mode
+from ppdet.utils.check import check_gpu, check_xpu, check_version, check_config, enable_static_mode
 
 from ppdet.data.reader import create_reader
 
@@ -49,15 +49,27 @@ def main():
     check_config(cfg)
     # check if set use_gpu=True in paddlepaddle cpu version
     check_gpu(cfg.use_gpu)
+    use_xpu = False
+    if hasattr(cfg, 'use_xpu'):
+        check_xpu(cfg.use_xpu)
+        use_xpu = cfg.use_xpu
     # check if paddlepaddle version is satisfied
     check_version()
 
+    assert not (use_xpu and cfg.use_gpu), \
+            'Can not run on both XPU and GPU'
+
     main_arch = cfg.architecture
 
     multi_scale_test = getattr(cfg, 'MultiScaleTEST', None)
 
     # define executor
-    place = fluid.CUDAPlace(0) if cfg.use_gpu else fluid.CPUPlace()
+    if cfg.use_gpu:
+        place = fluid.CUDAPlace(0)
+    elif use_xpu:
+        place = fluid.XPUPlace(0)
+    else:
+        place = CPUPlace()
     exe = fluid.Executor(place)
 
     # build program
@@ -91,6 +103,8 @@ def main():
         return
 
     compile_program = fluid.CompiledProgram(eval_prog).with_data_parallel()
+    if use_xpu:
+        compile_program = eval_prog
 
     assert cfg.metric != 'OID', "eval process of OID dataset \
                           is not supported."

tools/train.py

@@ -43,7 +43,7 @@ from ppdet.utils import dist_utils
 from ppdet.utils.eval_utils import parse_fetches, eval_run, eval_results
 from ppdet.utils.stats import TrainingStats
 from ppdet.utils.cli import ArgsParser
-from ppdet.utils.check import check_gpu, check_version, check_config, enable_static_mode
+from ppdet.utils.check import check_gpu, check_xpu, check_version, check_config, enable_static_mode
 import ppdet.utils.checkpoint as checkpoint
 
 import logging
@@ -73,9 +73,16 @@ def main():
     check_config(cfg)
     # check if set use_gpu=True in paddlepaddle cpu version
     check_gpu(cfg.use_gpu)
+    use_xpu = False
+    if hasattr(cfg, 'use_xpu'):
+        check_xpu(cfg.use_xpu)
+        use_xpu = cfg.use_xpu
     # check if paddlepaddle version is satisfied
     check_version()
 
+    assert not (use_xpu and cfg.use_gpu), \
+            'Can not run on both XPU and GPU'
+
     save_only = getattr(cfg, 'save_prediction_only', False)
     if save_only:
         raise NotImplementedError('The config file only support prediction,'
@@ -84,14 +91,25 @@ def main():
 
     if cfg.use_gpu:
         devices_num = fluid.core.get_cuda_device_count()
+    elif use_xpu:
+        # ToDo(qingshu): XPU only support single card now
+        devices_num = 1
     else:
         devices_num = int(os.environ.get('CPU_NUM', 1))
 
-    if 'FLAGS_selected_gpus' in env:
+    if cfg.use_gpu and 'FLAGS_selected_gpus' in env:
         device_id = int(env['FLAGS_selected_gpus'])
+    elif use_xpu and 'FLAGS_selected_xpus' in env:
+        device_id = int(env['FLAGS_selected_xpus'])
     else:
         device_id = 0
-    place = fluid.CUDAPlace(device_id) if cfg.use_gpu else fluid.CPUPlace()
+
+    if cfg.use_gpu:
+        place = fluid.CUDAPlace(device_id)
+    elif use_xpu:
+        place = fluid.XPUPlace(device_id)
+    else:
+        place = fluid.CPUPlace()
     exe = fluid.Executor(place)
 
     lr_builder = create('LearningRate')
@@ -184,9 +202,13 @@ def main():
         loss_name=loss.name,
         build_strategy=build_strategy,
         exec_strategy=exec_strategy)
+    if use_xpu:
+        compiled_train_prog = train_prog
 
     if FLAGS.eval:
         compiled_eval_prog = fluid.CompiledProgram(eval_prog)
+        if use_xpu:
+            compiled_eval_prog = eval_prog
 
     fuse_bn = getattr(model.backbone, 'norm_type', None) == 'affine_channel'