From f4d54fe6337746fda3f9312c8360595748b3a1d0 Mon Sep 17 00:00:00 2001
From: Paul <pfultz2@yahoo.com>
Date: Wed, 13 Sep 2023 13:55:48 -0500
Subject: [PATCH 1/2] Add fp16 flag to test runenr to check models quantized to
 fp16

---
 tools/test_runner.py | 15 ++++++++++++++-
 1 file changed, 14 insertions(+), 1 deletion(-)

diff --git a/tools/test_runner.py b/tools/test_runner.py
index 8fb16fc17e8..75ed126b61e 100644
--- a/tools/test_runner.py
+++ b/tools/test_runner.py
@@ -39,6 +39,17 @@ def parse_args():
                         type=str,
                         default='gpu',
                         help='Specify where the tests execute (ref, gpu)')
+    parser.add_argument('--fp16',
+                        action='store_true',
+                        help='Quantize to fp16')
+    parser.add_argument('--atol',
+                        type=float,
+                        default=1e-3,
+                        help='The absolute tolerance parameter')
+    parser.add_argument('--rtol',
+                        type=float,
+                        default=1e-3,
+                        help='The relative tolerance parameter')
     args = parser.parse_args()
 
     return args
@@ -257,6 +268,8 @@ def main():
 
     # read and compile model
     model = migraphx.parse_onnx(model_path_name, map_input_dims=param_shapes)
+    if args.fp16:
+        migraphx.quantize_fp16(model)
     model.compile(migraphx.get_target(target))
 
     # get test cases
@@ -279,7 +292,7 @@ def main():
         output_data = run_one_case(model, input_data)
 
         # check output correctness
-        ret = check_correctness(gold_outputs, output_data)
+        ret = check_correctness(gold_outputs, output_data, atol=args.atol, rtol=args.rtol)
         if ret:
             correct_num += 1
 

From e24f190418e7803029e4e7506a73321d54739d24 Mon Sep 17 00:00:00 2001
From: Paul <pfultz2@yahoo.com>
Date: Wed, 13 Sep 2023 13:56:00 -0500
Subject: [PATCH 2/2] Format

---
 tools/test_runner.py | 9 +++++----
 1 file changed, 5 insertions(+), 4 deletions(-)

diff --git a/tools/test_runner.py b/tools/test_runner.py
index 75ed126b61e..8bcd9fbc7b5 100644
--- a/tools/test_runner.py
+++ b/tools/test_runner.py
@@ -39,9 +39,7 @@ def parse_args():
                         type=str,
                         default='gpu',
                         help='Specify where the tests execute (ref, gpu)')
-    parser.add_argument('--fp16',
-                        action='store_true',
-                        help='Quantize to fp16')
+    parser.add_argument('--fp16', action='store_true', help='Quantize to fp16')
     parser.add_argument('--atol',
                         type=float,
                         default=1e-3,
@@ -292,7 +290,10 @@ def main():
         output_data = run_one_case(model, input_data)
 
         # check output correctness
-        ret = check_correctness(gold_outputs, output_data, atol=args.atol, rtol=args.rtol)
+        ret = check_correctness(gold_outputs,
+                                output_data,
+                                atol=args.atol,
+                                rtol=args.rtol)
         if ret:
             correct_num += 1