Use tempfile in tutorials

tutorials/auto_scheduler/tune_conv2d_layer_cuda.py

@@ -32,6 +32,8 @@
 We use a convolution layer as an example in this tutorial.
 """
 
+import tempfile
+
 import numpy as np
 import tvm
 from tvm import te, auto_scheduler, topi
@@ -82,17 +84,19 @@ def conv2d_layer(N, H, W, CO, CI, KH, KW, stride, padding):
 # * :code:`num_measure_trials` is the number of measurement trials we can use during the search.
 #   We only make 10 trials in this tutorial for a fast demonstration. In practice, 1000 is a
 #   good value for the search to converge. You can do more trials according to your time budget.
-# * In addition, we use :code:`RecordToFile` to dump measurement records into a file `conv2d.json`.
+# * In addition, we use :code:`RecordToFile` to dump measurement records into a file.
+#   Note that here we use a temporarty file for demonstraction, but in practice you should use
+#   a more maintainable file name such as `conv2d.json`.
 #   The measurement records can be used to query the history best, resume the search,
 #   and do more analyses later.
 # * see :any:`auto_scheduler.TuningOptions`,
 #   :any:`auto_scheduler.LocalRPCMeasureContext` for more parameters.
-
+logfile = tempfile.NamedTemporaryFile(prefix="conv2d", suffix=".json")
 measure_ctx = auto_scheduler.LocalRPCMeasureContext(min_repeat_ms=300)
 tune_option = auto_scheduler.TuningOptions(
     num_measure_trials=10,
     runner=measure_ctx.runner,
-    measure_callbacks=[auto_scheduler.RecordToFile("conv2d.json")],
+    measure_callbacks=[auto_scheduler.RecordToFile(logfile.name)],
 )
 
 ######################################################################
@@ -149,15 +153,15 @@ def conv2d_layer(N, H, W, CO, CI, KH, KW, stride, padding):
 # Using the record file
 # ^^^^^^^^^^^^^^^^^^^^^
 # During the search, all measuremnt records are dumpped into the record
-# file "conv2d.json". The measurement records can be used to re-apply search results,
+# file. The measurement records can be used to re-apply search results,
 # resume the search, and perform other analyses.
 
 ######################################################################
 # Here is an example where we load the best schedule from a file,
 # print the equivalent python schedule API, and build the binary again.
 
 # Load the measuremnt record for the best schedule
-inp, res = auto_scheduler.load_best("conv2d.json", task.workload_key)
+inp, res = auto_scheduler.load_best(logfile.name, task.workload_key)
 
 # Print equivalent python schedule API. This can be used for debugging and
 # learning the behavior of the auto-scheduler.
@@ -176,17 +180,16 @@ def conv2d_layer(N, H, W, CO, CI, KH, KW, stride, padding):
 # In the example below we resume the status and do more 5 trials.
 
 
-log_file = "conv2d.json"
 cost_model = auto_scheduler.XGBModel()
-cost_model.update_from_file(log_file)
+cost_model.update_from_file(logfile.name)
 search_policy = auto_scheduler.SketchPolicy(
-    task, cost_model, init_search_callbacks=[auto_scheduler.PreloadMeasuredStates(log_file)]
+    task, cost_model, init_search_callbacks=[auto_scheduler.PreloadMeasuredStates(logfile.name)]
 )
 measure_ctx = auto_scheduler.LocalRPCMeasureContext(min_repeat_ms=300)
 tune_option = auto_scheduler.TuningOptions(
     num_measure_trials=5,
     runner=measure_ctx.runner,
-    measure_callbacks=[auto_scheduler.RecordToFile(log_file)],
+    measure_callbacks=[auto_scheduler.RecordToFile(logfile.name)],
 )
 sch, args = auto_scheduler.auto_schedule(task, search_policy, tuning_options=tune_option)
 

tutorials/auto_scheduler/tune_matmul_x86.py

@@ -28,6 +28,7 @@
 
 We use matrix multiplication as an example in this tutorial.
 """
+import tempfile
 
 import numpy as np
 import tvm
@@ -75,13 +76,15 @@ def matmul_add(N, L, M, dtype):
 # * :code:`num_measure_trials` is the number of measurement trials we can use during the search.
 #   We only make 10 trials in this tutorial for a fast demonstration. In practice, 1000 is a
 #   good value for the search to converge. You can do more trials according to your time budget.
-# * In addition, we use :code:`RecordToFile` to dump measurement records into a file `matmul.json`.
+# * In addition, we use :code:`RecordToFile` to dump measurement records into a file.
+#   Note that here we use a temporarty file for demonstraction, but in practice you should use
+#   a more maintainable file name such as `matmul.json`.
 #   The measurement records can be used to query the history best, resume the search,
 #   and do more analyses later.
 # * see :any:`auto_scheduler.TuningOptions` for more parameters
-
+logfile = tempfile.NamedTemporaryFile(prefix="matmul", suffix=".json")
 tune_option = auto_scheduler.TuningOptions(
-    num_measure_trials=10, measure_callbacks=[auto_scheduler.RecordToFile("matmul.json")]
+    num_measure_trials=10, measure_callbacks=[auto_scheduler.RecordToFile(logfile.name)]
 )
 
 ######################################################################
@@ -133,15 +136,15 @@ def matmul_add(N, L, M, dtype):
 # Using the record file
 # ^^^^^^^^^^^^^^^^^^^^^
 # During the search, all measuremnt records are dumpped into the record
-# file "matmul.json". The measurement records can be used to re-apply search results,
+# file. The measurement records can be used to re-apply search results,
 # resume the search, and perform other analyses.
 
 ######################################################################
 # Here is an example where we load the best schedule from a file,
 # print the equivalent python schedule API, and build the binary again.
 
 # Load the measuremnt record for the best schedule
-inp, res = auto_scheduler.load_best("matmul.json", task.workload_key)
+inp, res = auto_scheduler.load_best(logfile.name, task.workload_key)
 
 # Print equivalent python schedule API. This can be used for debugging and
 # learning the behavior of the auto-scheduler.
@@ -160,19 +163,19 @@ def matmul_add(N, L, M, dtype):
 # In the example below we resume the status and do more 5 trials.
 
 
-def resume_search(task, log_file):
+def resume_search(task, logfile_name):
     cost_model = auto_scheduler.XGBModel()
-    cost_model.update_from_file(log_file)
+    cost_model.update_from_file(logfile_name)
     search_policy = auto_scheduler.SketchPolicy(
-        task, cost_model, init_search_callbacks=[auto_scheduler.PreloadMeasuredStates(log_file)]
+        task, cost_model, init_search_callbacks=[auto_scheduler.PreloadMeasuredStates(logfile_name)]
     )
     tune_option = auto_scheduler.TuningOptions(
-        num_measure_trials=5, measure_callbacks=[auto_scheduler.RecordToFile(log_file)]
+        num_measure_trials=5, measure_callbacks=[auto_scheduler.RecordToFile(logfile_name)]
     )
     sch, args = auto_scheduler.auto_schedule(task, search_policy, tuning_options=tune_option)
 
 
-# resume_search(task, "matmul.json")
+# resume_search(task, logfile.name)
 
 ######################################################################
 # .. note::