[ROCm][TunableOp] Unit test for TunableOp BLAS logging. (pytorch#148982)

Add unit test for new TunableOp BLAS logging feature.

Requires this PR to be merged in first: pytorch#148979

Pull Request resolved: pytorch#148982
Approved by: https://github.com/jeffdaily

Author: naromero77amd

test/test_linalg.py

@@ -78,12 +78,17 @@ def set_tunableop_defaults():
     torch.cuda.tunable.set_max_tuning_iterations(100)
     torch.cuda.tunable.set_rotating_buffer_size(-1)
 
-def tunableop_matmul(device, dtype):
+def tunableop_matmul(device, dtype, offline=False):
     # Helper function to test TunableOp in a subprocess
     # requires helper function since lambda function
     # not supported by multiprocessing module
     import os
     os.environ["PYTORCH_TUNABLEOP_ENABLED"] = "1"
+
+    if offline:
+        torch.cuda.tunable.tuning_enable(False)
+        torch.cuda.tunable.record_untuned_enable(True)
+
     torch.cuda.tunable.set_max_tuning_duration(1)
     A = torch.randn((17, 17), device=device, dtype=dtype)
     B = torch.randn((17, 17), device=device, dtype=dtype)
@@ -5661,6 +5666,108 @@ def test_tf32_offline_tunableop(self, device, dtype):
                 except (FileNotFoundError, PermissionError):
                     pass
 
+    @onlyCUDA
+    @skipCUDAIfNotRocm
+    @dtypes(torch.float16)
+    def test_blaslog_tunableop(self, device, dtype):
+        # Test that PYTORCH_TUNABLEOP_BLAS_LOG=1 gives
+        # an additional column of data with the BLAS
+        # parameters in offline and online tuning.
+        #
+        # We record GEMMs and then check that the
+        # BLAS_PARAMS appear in
+        # tunableop_untuned CSV file
+        # and
+        # tunableop_results CSV file
+        #
+        # NOTE: This is done in a subproceses
+        # because in the main process
+        # PYTORCH_TUNABLEOP_BLAS_LOG has already
+        # been deactivated and its value is sticky
+        import os
+        import multiprocessing as mp
+
+        set_tunableop_defaults()
+        ordinal = torch.cuda.current_device()
+
+        result_filename = f"tunableop_results{ordinal}.csv"
+        untuned_filename = f"tunableop_untuned{ordinal}.csv"
+
+        # Test in try-finally block to avoid leaking state
+        # if test is interrupted.
+        try:
+            os.putenv("PYTORCH_TUNABLEOP_BLAS_LOG", "1")
+
+            # Offline Tuning case in a subprocess
+
+            # force=True needed according to:
+            # https://docs.python.org/3/library/multiprocessing.html#multiprocessing.set_start_method
+            # This is because a different test in this process could have
+            # already set the start method
+            mp.set_start_method("spawn", force=True)
+
+            p = mp.Process(target=tunableop_matmul, args=(device, dtype, True))
+            p.start()
+            p.join()
+
+            # Make sure the results file exists and that it is not zero
+            self.assertTrue(os.path.exists(untuned_filename))
+            self.assertTrue(os.path.getsize(untuned_filename) > 0)
+
+            # Check that the BLAS PARAMS are in the CSV file
+            import csv
+            with open(untuned_filename) as file:
+                reader = csv.reader(file)
+                first_row = next(reader)
+                # Check for extra column
+                self.assertGreater(len(first_row), 3)
+                # Check for YAML entry to the right of
+                # BLAS PARAMS
+                self.assertTrue("{ function:" in first_row[2])
+
+            # Online tuning case in a subprocess
+
+            # force=True needed according to:
+            # https://docs.python.org/3/library/multiprocessing.html#multiprocessing.set_start_method
+            # This is because a different test in this process could have
+            # already set the start method
+            mp.set_start_method("spawn", force=True)
+
+            p = mp.Process(target=tunableop_matmul, args=(device, dtype, False))
+            p.start()
+            p.join()
+
+            # Make sure the results file exists and that it is not zero
+            self.assertTrue(os.path.exists(result_filename))
+            self.assertGreater(os.path.getsize(result_filename), 0)
+
+            # Check that there BLAS PARAMS are in the CSV file
+            with open(result_filename) as file:
+                reader = csv.reader(file)
+                for _ in range(5):  # Skip the first 5 lines for the validator
+                    next(reader, None)
+                # Check for extra column
+                first_row = next(reader)
+                self.assertGreater(len(first_row), 5)
+                # Check for YAML entry to the right of
+                # BLAS PARAMS
+                self.assertTrue("{ function:" in first_row[4])
+
+        finally:
+            # undo all the environment variables set
+            try:
+                del os.environ["PYTORCH_TUNABLEOP_BLAS_LOG"]
+            except KeyError:
+                pass
+
+            # clean up, remove any files that were generated
+            for filename in [untuned_filename, result_filename]:
+                try:
+                    os.remove(filename)
+                # NB: The file is locked on Windows
+                except (FileNotFoundError, PermissionError):
+                    pass
+
     @dtypes(torch.float, torch.complex64)
     def test_matmul_out_kernel_errors_with_autograd(self, device, dtype):
         a = torch.empty((256, 512), device=device, dtype=dtype, requires_grad=True).unsqueeze(0)