From fba54ddec18f9a57bed44d306435fe960f5e7356 Mon Sep 17 00:00:00 2001
From: carryyu <569782149@qq.com>
Date: Sun, 25 Sep 2022 16:34:37 +0800
Subject: [PATCH] Fix unit test in A10 GPU (#46450)

*  Disable TF32 to solve accuracy for test_trt_conv_pass and test_trt_deformable_conv in A10 GPU.
---
 .../fluid/tests/unittests/ir/inference/test_trt_conv_pass.py   | 2 ++
 .../tests/unittests/ir/inference/test_trt_deformable_conv.py   | 3 +++
 2 files changed, 5 insertions(+)

diff --git a/python/paddle/fluid/tests/unittests/ir/inference/test_trt_conv_pass.py b/python/paddle/fluid/tests/unittests/ir/inference/test_trt_conv_pass.py
index a934c264e4..3a16f4f7d5 100644
--- a/python/paddle/fluid/tests/unittests/ir/inference/test_trt_conv_pass.py
+++ b/python/paddle/fluid/tests/unittests/ir/inference/test_trt_conv_pass.py
@@ -22,6 +22,8 @@ import paddle.fluid.core as core
 from paddle.fluid.core import PassVersionChecker
 from paddle.fluid.core import AnalysisConfig
 
+os.environ['NVIDIA_TF32_OVERRIDE'] = '0'
+
 
 class TensorRTSubgraphPassConvTest(InferencePassTest):
 
diff --git a/python/paddle/fluid/tests/unittests/ir/inference/test_trt_deformable_conv.py b/python/paddle/fluid/tests/unittests/ir/inference/test_trt_deformable_conv.py
index 3bed89e74f..6a0e98539a 100644
--- a/python/paddle/fluid/tests/unittests/ir/inference/test_trt_deformable_conv.py
+++ b/python/paddle/fluid/tests/unittests/ir/inference/test_trt_deformable_conv.py
@@ -14,6 +14,7 @@
 
 from __future__ import print_function
 
+import os
 import unittest
 import numpy as np
 from inference_pass_test import InferencePassTest
@@ -22,6 +23,8 @@ import paddle.fluid.core as core
 from paddle.fluid.core import PassVersionChecker
 from paddle.fluid.core import AnalysisConfig
 
+os.environ['NVIDIA_TF32_OVERRIDE'] = '0'
+
 
 class TRTDeformableConvTest(InferencePassTest):
 
-- 
GitLab