diff --git a/backends/arm/test/models/test_mobilenet_v2_arm.py b/backends/arm/test/models/test_mobilenet_v2_arm.py
index ac513530e04..a1f9bc0633d 100644
--- a/backends/arm/test/models/test_mobilenet_v2_arm.py
+++ b/backends/arm/test/models/test_mobilenet_v2_arm.py
@@ -32,6 +32,12 @@
 input_t = Tuple[torch.Tensor]
 
 
+quant_test_data = {
+    "per_channel_quantization=true": True,
+    "per_channel_quantization=false": False,
+}
+
+
 def test_mv2_tosa_MI():
     pipeline = TosaPipelineMI[input_t](
         mv2, model_inputs, aten_op=[], exir_op=[], use_to_edge_transform_and_lower=True
@@ -39,14 +45,15 @@ def test_mv2_tosa_MI():
     pipeline.run()
 
 
-def test_mv2_tosa_BI():
+@common.parametrize("per_channel_quantization", quant_test_data)
+def test_mv2_tosa_BI(per_channel_quantization):
     pipeline = TosaPipelineBI[input_t](
         mv2,
         model_inputs,
         aten_op=[],
         exir_op=[],
         use_to_edge_transform_and_lower=True,
-        per_channel_quantization=True,
+        per_channel_quantization=per_channel_quantization,
         atol=0.25,
         qtol=1,
     )
@@ -55,7 +62,8 @@ def test_mv2_tosa_BI():
 
 @pytest.mark.slow
 @common.XfailIfNoCorstone300
-def test_mv2_u55_BI():
+@common.parametrize("per_channel_quantization", quant_test_data)
+def test_mv2_u55_BI(per_channel_quantization):
     pipeline = EthosU55PipelineBI[input_t](
         mv2,
         model_inputs,
@@ -63,7 +71,7 @@ def test_mv2_u55_BI():
         exir_ops=[],
         run_on_fvp=True,
         use_to_edge_transform_and_lower=True,
-        per_channel_quantization=True,
+        per_channel_quantization=per_channel_quantization,
         atol=0.25,
         qtol=1,
     )
@@ -72,7 +80,8 @@ def test_mv2_u55_BI():
 
 @pytest.mark.slow
 @common.XfailIfNoCorstone320
-def test_mv2_u85_BI():
+@common.parametrize("per_channel_quantization", quant_test_data)
+def test_mv2_u85_BI(per_channel_quantization):
     pipeline = EthosU85PipelineBI[input_t](
         mv2,
         model_inputs,
@@ -80,7 +89,7 @@ def test_mv2_u85_BI():
         exir_ops=[],
         run_on_fvp=True,
         use_to_edge_transform_and_lower=True,
-        per_channel_quantization=True,
+        per_channel_quantization=per_channel_quantization,
         atol=0.25,
         qtol=1,
     )
diff --git a/backends/arm/test/ops/test_conv1d.py b/backends/arm/test/ops/test_conv1d.py
index 768da4d5c89..cc8245ba126 100644
--- a/backends/arm/test/ops/test_conv1d.py
+++ b/backends/arm/test/ops/test_conv1d.py
@@ -249,7 +249,7 @@ def forward(self, x):
     batches=1,
 )
 
-test_modules = {
+test_data_MI = {
     "2_3x2x40_nobias": lambda: conv1d_2_3x2x40_nobias,
     "3_1x3x256_st1": lambda: conv1d_3_1x3x256_st1,
     "3_1x3x12_st2_pd1": lambda: conv1d_3_1x3x12_st2_pd1,
@@ -265,53 +265,65 @@ def forward(self, x):
     "two_conv1d": lambda: two_conv1d,
 }
 
+test_data_BI = {
+    f"{k},per_channel_quant={q}": (lambda v=v, q=q: (v(), q))
+    for (k, v) in test_data_MI.items()
+    for q in [True, False]
+}
+
 
-@common.parametrize("test_module", test_modules)
-def test_convolution_1d_tosa_MI(test_module):
+@common.parametrize("test_data", test_data_MI)
+def test_convolution_1d_tosa_MI(test_data):
     pipeline = TosaPipelineMI[input_t](
-        test_module(),
-        test_module().get_inputs(),
+        test_data(),
+        test_data().get_inputs(),
         aten_op,
         exir_op,
     )
     pipeline.run()
 
 
-@common.parametrize("test_module", test_modules)
-def test_convolution_1d_tosa_BI(test_module):
+@common.parametrize("test_data", test_data_BI)
+def test_convolution_1d_tosa_BI(test_data):
+    model, per_channel_quantization = test_data()
     pipeline = TosaPipelineBI[input_t](
-        test_module(),
-        test_module().get_inputs(),
+        model,
+        model.get_inputs(),
         aten_op,
         exir_op,
+        per_channel_quantization=per_channel_quantization,
+        qtol=1,
     )
-    pipeline.change_args("run_method_and_compare_outputs", qtol=1)
     pipeline.run()
 
 
-@common.parametrize("test_module", test_modules)
+@common.parametrize("test_data", test_data_BI)
 @common.XfailIfNoCorstone300
-def test_convolution_1d_u55_BI(test_module):
+def test_convolution_1d_u55_BI(test_data):
+    model, per_channel_quantization = test_data()
     pipeline = EthosU55PipelineBI[input_t](
-        test_module(),
-        test_module().get_inputs(),
+        model,
+        model.get_inputs(),
         aten_op,
         exir_op,
         run_on_fvp=True,
+        per_channel_quantization=per_channel_quantization,
+        qtol=1,
     )
-    pipeline.change_args("run_method_and_compare_outputs", qtol=1)
     pipeline.run()
 
 
-@common.parametrize("test_module", test_modules)
+@common.parametrize("test_data", test_data_BI)
 @common.XfailIfNoCorstone320
-def test_convolution_1d_u85_BI(test_module):
+def test_convolution_1d_u85_BI(test_data):
+    model, per_channel_quantization = test_data()
     pipeline = EthosU85PipelineBI[input_t](
-        test_module(),
-        test_module().get_inputs(),
+        model,
+        model.get_inputs(),
         aten_op,
         exir_op,
         run_on_fvp=True,
+        per_channel_quantization=per_channel_quantization,
+        qtol=1,
     )
-    pipeline.change_args("run_method_and_compare_outputs", qtol=1)
     pipeline.run()
diff --git a/backends/arm/test/ops/test_conv2d.py b/backends/arm/test/ops/test_conv2d.py
index 658978d0de8..9aa479466f8 100644
--- a/backends/arm/test/ops/test_conv2d.py
+++ b/backends/arm/test/ops/test_conv2d.py
@@ -357,7 +357,7 @@ def forward(self, x):
 
 # Shenanigan to get a nicer output when test fails. With unittest it looks like:
 # FAIL: test_convolution_2d_tosa_BI_2_3x3_1x3x12x12_st2_pd1
-test_modules = {
+test_data_MI = {
     "2x2_3x2x40x40_nobias": lambda: conv2d_2x2_3x2x40x40_nobias,
     "3x3_1x3x256x256_st1": lambda: conv2d_3x3_1x3x256x256_st1,
     "3x3_1x3x12x12_st2_pd1": lambda: conv2d_3x3_1x3x12x12_st2_pd1,
@@ -380,58 +380,79 @@ def forward(self, x):
     "groups_bias": lambda: conv2d_groups_bias,
 }
 
+# Generate a new test set paired with per_channel_quant=True/False.
+test_data_BI = {
+    f"{k},per_channel_quant={q}": (lambda v=v, q=q: (v(), q))
+    for (k, v) in test_data_MI.items()
+    for q in [True, False]
+    # TODO: Invalid TOSA graph (MLETORCH-1144)
+    if (k not in ["groups", "groups_bias"]) and (q is True)
+}
+
 fvp_xfails = {
-    "2x2_3x2x40x40_nobias": "MLETORCH-520: Numerical issues on FVP.",
-    "5x5_3x2x128x128_st1": "MLETORCH-520: Numerical issues on FVP.",
+    f"{k},per_channel_quant={q}": reason
+    for k, reason in {
+        "2x2_3x2x40x40_nobias": "MLETORCH-520: Numerical issues on FVP.",
+        "5x5_3x2x128x128_st1": "MLETORCH-520: Numerical issues on FVP.",
+    }.items()
+    for q in [True, False]
 }
+
 input_t = Tuple[torch.Tensor]
 
 
-@common.parametrize("test_module", test_modules)
-def test_convolution_2d_tosa_MI(test_module):
+@common.parametrize("test_data", test_data_MI)
+def test_convolution_2d_tosa_MI(test_data):
+    model = test_data()
     pipeline = TosaPipelineMI[input_t](
-        test_module(),
-        test_module().get_inputs(),
+        model,
+        model.get_inputs(),
         aten_op,
         exir_op,
     )
     pipeline.run()
 
 
-@common.parametrize("test_module", test_modules)
-def test_convolution_2d_tosa_BI(test_module):
+@common.parametrize("test_data", test_data_BI)
+def test_convolution_2d_tosa_BI(test_data):
+    model, per_channel_quantization = test_data()
     pipeline = TosaPipelineBI[input_t](
-        test_module(),
-        test_module().get_inputs(),
+        model,
+        model.get_inputs(),
         aten_op,
         exir_op,
+        per_channel_quantization=per_channel_quantization,
+        qtol=1,
     )
-    pipeline.change_args("run_method_and_compare_outputs", qtol=1)
     pipeline.run()
 
 
-@common.parametrize("test_module", test_modules, fvp_xfails)
+@common.parametrize("test_data", test_data_BI, fvp_xfails)
 @common.XfailIfNoCorstone300
-def test_convolution_2d_u55_BI(test_module):
+def test_convolution_2d_u55_BI(test_data):
+    model, per_channel_quantization = test_data()
     pipeline = EthosU55PipelineBI[input_t](
-        test_module(),
-        test_module().get_inputs(),
+        model,
+        model.get_inputs(),
         aten_op,
         exir_op,
         run_on_fvp=True,
+        per_channel_quantization=per_channel_quantization,
     )
     pipeline.run()
 
 
-@common.parametrize("test_module", test_modules, fvp_xfails)
+@common.parametrize("test_data", test_data_BI, fvp_xfails)
 @common.XfailIfNoCorstone320
-def test_convolution_2d_u85_BI(test_module):
+def test_convolution_u85_BI(test_data):
+    model, per_channel_quantization = test_data()
     pipeline = EthosU85PipelineBI[input_t](
-        test_module(),
-        test_module().get_inputs(),
+        model,
+        model.get_inputs(),
         aten_op,
         exir_op,
         run_on_fvp=True,
+        per_channel_quantization=per_channel_quantization,
     )
     pipeline.run()
 
diff --git a/backends/arm/test/ops/test_conv3d.py b/backends/arm/test/ops/test_conv3d.py
index c7bb7c55887..1a8ea5c3dd5 100644
--- a/backends/arm/test/ops/test_conv3d.py
+++ b/backends/arm/test/ops/test_conv3d.py
@@ -304,7 +304,7 @@ def forward(self, x):
     batches=1,
 )
 
-test_modules = {
+test_data_MI = {
     "2x2_3x2x40x40_nobias": lambda: conv3d_2x2_3x2x40x40_nobias,
     "3x3_1x3x256x256_st1": lambda: conv3d_3x3_1x3x256x256_st1,
     "3x3_1x3x12x12_st2_pd1": lambda: conv3d_3x3_1x3x12x12_st2_pd1,
@@ -323,50 +323,66 @@ def forward(self, x):
     "3x3_1x3x224x224_st2_pd1": lambda: conv3d_3x3_1x3x224x224_st2_pd1,
 }
 
+# Generate a new test set paired with per_channel_quant=True/False.
+test_data_BI = {
+    f"{k},per_channel_quant={q}": (lambda v=v, q=q: (v(), q))
+    for (k, v) in test_data_MI.items()
+    for q in [True, False]
+}
+
 input_t = Tuple[torch.Tensor]
 
 
-@common.parametrize("test_module", test_modules)
+@common.parametrize("test_data", test_data_MI)
 @pytest.mark.skip  # Not implemented, skip until it is.
-def test_convolution_tosa_MI_3d(test_module):
+def test_convolution_3d_tosa_MI(test_data):
     pipeline = TosaPipelineMI[input_t](
-        test_module(), test_module().get_inputs(), aten_op, exir_op
+        test_data(), test_data().get_inputs(), aten_op, exir_op
     )
     pipeline.run()
 
 
-@common.parametrize("test_module", test_modules)
+@common.parametrize("test_data", test_data_BI)
 @pytest.mark.skip  # Not implemented, skip until it is.
-def test_convolution_tosa_BI_3d(test_module):
+def test_convolution_3d_tosa_BI(test_data):
+    model, per_channel_quantization = test_data()
     pipeline = TosaPipelineBI[input_t](
-        test_module(), test_module().get_inputs(), aten_op, exir_op
+        model,
+        model.get_inputs(),
+        aten_op,
+        exir_op,
+        per_channel_quantization=per_channel_quantization,
+        qtol=1,
     )
-    pipeline.change_args("run_method_and_compare_outputs", qtol=1)
     pipeline.run()
 
 
-@common.parametrize("test_module", test_modules)
+@common.parametrize("test_data", test_data_BI)
 @pytest.mark.skip  # Not implemented, skip until it is.
-def test_convolution_u55_BI_3d(test_module):
+def test_convolution_3d_u55_BI(test_data):
+    model, per_channel_quantization = test_data()
     pipeline = EthosU55PipelineBI[input_t](
-        test_module(),
-        test_module().get_inputs(),
+        model,
+        model.get_inputs(),
         aten_op,
         exir_op,
         run_on_fvp=True,
+        per_channel_quantization=per_channel_quantization,
     )
     pipeline.run()
 
 
-@common.parametrize("test_module", test_modules)
+@common.parametrize("test_data", test_data_BI)
 @pytest.mark.skip  # Not implemented, skip until it is.
-def test_convolution_u85_BI_3d(test_module):
+def test_convolution_3d_u85_BI(test_data):
+    model, per_channel_quantization = test_data()
     pipeline = EthosU85PipelineBI[input_t](
-        test_module(),
-        test_module().get_inputs(),
+        model,
+        model.get_inputs(),
         aten_op,
         exir_op,
         run_on_fvp=True,
+        per_channel_quantization=per_channel_quantization,
     )
     pipeline.run()
 
diff --git a/backends/arm/test/ops/test_conv_combos.py b/backends/arm/test/ops/test_conv_combos.py
index bddc30f04ab..d3218258087 100644
--- a/backends/arm/test/ops/test_conv_combos.py
+++ b/backends/arm/test/ops/test_conv_combos.py
@@ -36,6 +36,11 @@ class ComboBlockBottleneckResidual(torch.nn.Module):
         "executorch_exir_dialects_edge__ops_aten_add_Tensor",
     ]
 
+    test_data_BI = {
+        "per_channel_quant=True": True,
+        "per_channel_quant=False": False,
+    }
+
     def __init__(self):
         super().__init__()
         # (t, c, n, s) = (6, 96, 1, 1)
@@ -114,6 +119,18 @@ class ComboConvBatchnormRelu6(torch.nn.Module):
         "executorch_exir_dialects_edge__ops_aten_hardtanh_default",
     ]
 
+    test_data_MI = {
+        "affine=True": True,
+        "affine=False": False,
+    }
+
+    test_data_BI = {
+        "affine=True,per_channel_quant=True": (True, True),
+        "affine=True,per_channel_quant=False": (True, False),
+        "affine=False,per_channel_quant=True": (False, True),
+        "affine=False,per_channel_quant=False": (False, False),
+    }
+
     def __init__(self, affine: bool):
         super().__init__()
         self.conv2d = torch.nn.Conv2d(
@@ -142,7 +159,7 @@ class ComboConvRelu6(torch.nn.Module):
         "executorch_exir_dialects_edge__ops_aten_hardtanh_default",
     ]
 
-    test_data = {
+    test_data_MI = {
         "combo_conv_relu_2_x_4d": lambda: (2 * torch.randn(1, 3, 256, 256),),
         "combo_conv_relu_0_5_x_4d": lambda: (0.5 * torch.randn(1, 3, 256, 256),),
         "combo_conv_relu_4d": lambda: (torch.randn(1, 3, 256, 256),),
@@ -150,6 +167,14 @@ class ComboConvRelu6(torch.nn.Module):
         "combo_conv_relu_neg_2_x_4d": lambda: (-2 * torch.randn(1, 3, 256, 256),),
     }
 
+    # Generate a new test set paired with per_channel_quant=True/False.
+    test_data_BI = {
+        # test_name: (input, per_channel_quant)
+        f"{k},per_channel_quant={q}": (lambda v=v, q=q: (v(), q))
+        for (k, v) in test_data_MI.items()
+        for q in [True, False]
+    }
+
     def __init__(self):
         super().__init__()
         self.conv2d = torch.nn.Conv2d(
@@ -169,13 +194,21 @@ class ComboConvAvgPool2d(torch.nn.Module):
         "executorch_exir_dialects_edge__ops_aten_avg_pool2d_default",
     ]
 
-    test_data = {
+    test_data_MI = {
         "combo_conv_avgpool_20_x_4d": lambda: (20 * torch.randn(1, 3, 64, 32),),
         "combo_conv_avgpool_4d": lambda: (torch.randn(1, 3, 100, 200),),
         "combo_conv_avgpool_5_x_4d_randn": lambda: (5 * torch.randn(1, 3, 256, 256),),
         "combo_conv_avgpool_2_x_4d": lambda: (torch.rand(1, 3, 512, 128),),
     }
 
+    # Generate a new test set paired with per_channel_quant=True/False.
+    test_data_BI = {
+        # test_name: (input, per_channel_quant)
+        f"{k},per_channel_quant={q}": (lambda v=v, q=q: (v(), q))
+        for (k, v) in test_data_MI.items()
+        for q in [True, False]
+    }
+
     def __init__(self):
         super().__init__()
         self.conv2d = torch.nn.Conv2d(
@@ -196,7 +229,6 @@ def forward(self, x):
 
 def test_convolution_2d_tosa_MI_meandim():
     model = ComboConv2dMeandim()
-
     pipeline = TosaPipelineMI[input_t1](
         model,
         model.get_inputs(),
@@ -246,11 +278,11 @@ def test_convolution_2d_u85_BI_meandim():
 ##############################
 ## Conv + batch norm + relu ##
 ##############################
-affine_params = {"affine": True, "_no_affine": False}
 
 
-@common.parametrize("affine", affine_params)
-def test_convolution_2d_tosa_MI_batchnorm_relu6(affine):
+@common.parametrize("test_data", ComboConvBatchnormRelu6.test_data_MI)
+def test_convolution_2d_tosa_MI_batchnorm_relu6(test_data):
+    affine = test_data
     model = ComboConvBatchnormRelu6(affine)
     pipeline = TosaPipelineMI[input_t1](
         model,
@@ -262,21 +294,25 @@ def test_convolution_2d_tosa_MI_batchnorm_relu6(affine):
 
 
 @pytest.mark.flaky(reruns=5)  # TODO: Investigate flakyness (MLTORCH-307)
-@common.parametrize("affine", affine_params)
-def test_convolution_2d_tosa_BI_batchnorm_relu6(affine):
+@common.parametrize("test_data", ComboConvBatchnormRelu6.test_data_BI)
+def test_convolution_2d_tosa_BI_batchnorm_relu6(test_data):
+    affine, per_channel_quantization = test_data
     model = ComboConvBatchnormRelu6(affine)
     pipeline = TosaPipelineBI[input_t1](
         model,
         model.get_inputs(),
         aten_op=[],
         exir_op=ComboConvBatchnormRelu6.edge_op_list,
+        per_channel_quantization=per_channel_quantization,
+        qtol=1,
     )
     pipeline.run()
 
 
-@common.parametrize("affine", affine_params)
+@common.parametrize("test_data", ComboConvBatchnormRelu6.test_data_BI)
 @common.XfailIfNoCorstone300
-def test_convolution_2d_u55_BI_batchnorm_relu6(affine):
+def test_convolution_2d_u55_BI_batchnorm_relu6(test_data):
+    affine, per_channel_quantization = test_data
     model = ComboConvBatchnormRelu6(affine)
     pipeline = EthosU55PipelineBI[input_t1](
         model,
@@ -284,13 +320,15 @@ def test_convolution_2d_u55_BI_batchnorm_relu6(affine):
         aten_ops=[],
         exir_ops=[],
         run_on_fvp=True,
+        per_channel_quantization=per_channel_quantization,
     )
     pipeline.run()
 
 
-@common.parametrize("affine", affine_params)
+@common.parametrize("test_data", ComboConvBatchnormRelu6.test_data_BI)
 @common.XfailIfNoCorstone320
-def test_convolution_2d_u85_BI_batchnorm_relu6(affine):
+def test_convolution_2d_u85_BI_batchnorm_relu6(test_data):
+    affine, per_channel_quantization = test_data
     model = ComboConvBatchnormRelu6(affine)
     pipeline = EthosU85PipelineBI[input_t1](
         model,
@@ -298,6 +336,7 @@ def test_convolution_2d_u85_BI_batchnorm_relu6(affine):
         aten_ops=[],
         exir_ops=[],
         run_on_fvp=True,
+        per_channel_quantization=per_channel_quantization,
     )
     pipeline.run()
 
@@ -307,8 +346,8 @@ def test_convolution_2d_u85_BI_batchnorm_relu6(affine):
 ##################
 
 
-@common.parametrize("test_data", ComboConvRelu6.test_data)
-def test_convolution_2d_tosa_MI_relu6(test_data: torch.Tensor):
+@common.parametrize("test_data", ComboConvRelu6.test_data_MI)
+def test_convolution_2d_tosa_MI_relu6(test_data):
     model = ComboConvRelu6()
     pipeline = TosaPipelineMI[input_t1](
         model,
@@ -320,42 +359,48 @@ def test_convolution_2d_tosa_MI_relu6(test_data: torch.Tensor):
 
 
 @pytest.mark.flaky(reruns=5)  # TODO: Investigate flakyness (MLTORCH-307)
-@common.parametrize("test_data", ComboConvRelu6.test_data)
-def test_convolution_2d_tosa_BI_relu6(test_data: torch.Tensor):
+@common.parametrize("test_data", ComboConvRelu6.test_data_BI)
+def test_convolution_2d_tosa_BI_relu6(test_data):
+    input, per_channel_quantization = test_data()
     model = ComboConvRelu6()
     pipeline = TosaPipelineBI[input_t1](
         model,
-        test_data(),
+        input,
         aten_op=[],
         exir_op=ComboConvRelu6.edge_op_list,
+        per_channel_quantization=per_channel_quantization,
     )
     pipeline.run()
 
 
-@common.parametrize("test_data", ComboConvRelu6.test_data)
+@common.parametrize("test_data", ComboConvRelu6.test_data_BI)
 @common.XfailIfNoCorstone300
-def test_convolution_2d_u55_BI_relu6(test_data: torch.Tensor):
+def test_convolution_2d_u55_BI_relu6(test_data):
+    input, per_channel_quantization = test_data()
     model = ComboConvRelu6()
     pipeline = EthosU55PipelineBI[input_t1](
         model,
-        test_data(),
+        input,
         aten_ops=[],
         exir_ops=ComboConvRelu6.edge_op_list,
         run_on_fvp=True,
+        per_channel_quantization=per_channel_quantization,
     )
     pipeline.run()
 
 
-@common.parametrize("test_data", ComboConvRelu6.test_data)
+@common.parametrize("test_data", ComboConvRelu6.test_data_BI)
 @common.XfailIfNoCorstone320
-def test_convolution_2d_u85_BI_relu6(test_data: torch.Tensor):
+def test_convolution_2d_u85_BI_relu6(test_data):
+    input, per_channel_quantization = test_data()
     model = ComboConvRelu6()
     pipeline = EthosU85PipelineBI[input_t1](
         model,
-        test_data(),
+        input,
         aten_ops=[],
         exir_ops=ComboConvRelu6.edge_op_list,
         run_on_fvp=True,
+        per_channel_quantization=per_channel_quantization,
     )
     pipeline.run()
 
@@ -374,21 +419,26 @@ def test_convolution_2d_tosa_MI_block_bottleneck():
     pipeline.run()
 
 
+@common.parametrize("test_data", ComboBlockBottleneckResidual.test_data_BI)
 @pytest.mark.flaky(reruns=5)  # TODO: Investigate flakyness (MLTORCH-307)
-def test_convolution_2d_tosa_BI_block_bottleneck():
+def test_convolution_2d_tosa_BI_block_bottleneck(test_data):
+    per_channel_quantization = test_data
     model = ComboBlockBottleneckResidual()
     pipeline = TosaPipelineBI[input_t1](
         model,
         model.get_inputs(),
         aten_op=[],
         exir_op=ComboBlockBottleneckResidual.edge_op_list,
+        per_channel_quantization=per_channel_quantization,
     )
     pipeline.change_args("run_method_and_compare_outputs", model.get_inputs(), qtol=1)
     pipeline.run()
 
 
+@common.parametrize("test_data", ComboBlockBottleneckResidual.test_data_BI)
 @common.XfailIfNoCorstone300
-def test_convolution_2d_u55_BI_block_bottleneck():
+def test_convolution_2d_u55_BI_block_bottleneck(test_data):
+    per_channel_quantization = test_data
     model = ComboBlockBottleneckResidual()
     pipeline = EthosU55PipelineBI[input_t1](
         model,
@@ -396,12 +446,15 @@ def test_convolution_2d_u55_BI_block_bottleneck():
         aten_ops=[],
         exir_ops=[],
         run_on_fvp=True,
+        per_channel_quantization=per_channel_quantization,
     )
     pipeline.run()
 
 
+@common.parametrize("test_data", ComboBlockBottleneckResidual.test_data_BI)
 @common.XfailIfNoCorstone320
-def test_convolution_2d_u85_BI_block_bottleneck():
+def test_convolution_2d_u85_BI_block_bottleneck(test_data):
+    per_channel_quantization = test_data
     model = ComboBlockBottleneckResidual()
     pipeline = EthosU85PipelineBI[input_t1](
         model,
@@ -409,6 +462,7 @@ def test_convolution_2d_u85_BI_block_bottleneck():
         aten_ops=[],
         exir_ops=[],
         run_on_fvp=True,
+        per_channel_quantization=per_channel_quantization,
     )
     pipeline.run()
 
@@ -418,8 +472,8 @@ def test_convolution_2d_u85_BI_block_bottleneck():
 ######################
 
 
-@common.parametrize("test_data", ComboConvAvgPool2d.test_data)
-def test_convolution_2d_tosa_MI_avgpool2d(test_data: torch.Tensor):
+@common.parametrize("test_data", ComboConvAvgPool2d.test_data_MI)
+def test_convolution_2d_tosa_MI_avgpool2d(test_data):
     model = ComboConvAvgPool2d()
     pipeline = TosaPipelineMI[input_t1](
         model,
@@ -431,41 +485,47 @@ def test_convolution_2d_tosa_MI_avgpool2d(test_data: torch.Tensor):
 
 
 @pytest.mark.flaky(reruns=5)  # TODO: Investigate flakyness (MLTORCH-307)
-@common.parametrize("test_data", ComboConvAvgPool2d.test_data)
-def test_convolution_2d_tosa_BI_avgpool2d(test_data: torch.Tensor):
+@common.parametrize("test_data", ComboConvAvgPool2d.test_data_BI)
+def test_convolution_2d_tosa_BI_avgpool2d(test_data):
+    input, per_channel_quantization = test_data()
     model = ComboConvAvgPool2d()
     pipeline = TosaPipelineBI[input_t1](
         model,
-        test_data(),
+        input,
         aten_op=[],
         exir_op=ComboConvAvgPool2d.edge_op_list,
+        per_channel_quantization=per_channel_quantization,
     )
     pipeline.run()
 
 
-@common.parametrize("test_data", ComboConvAvgPool2d.test_data)
+@common.parametrize("test_data", ComboConvAvgPool2d.test_data_BI)
 @common.XfailIfNoCorstone300
-def test_convolution_2d_u55_BI_avgpool2d(test_data: torch.Tensor):
+def test_convolution_2d_u55_BI_avgpool2d(test_data):
+    input, per_channel_quantization = test_data()
     model = ComboConvAvgPool2d()
     pipeline = EthosU55PipelineBI[input_t1](
         model,
-        test_data(),
+        input,
         aten_ops=[],
         exir_ops=[],
         run_on_fvp=True,
+        per_channel_quantization=per_channel_quantization,
     )
     pipeline.run()
 
 
-@common.parametrize("test_data", ComboConvAvgPool2d.test_data)
+@common.parametrize("test_data", ComboConvAvgPool2d.test_data_BI)
 @common.XfailIfNoCorstone320
-def test_convolution_2d_u85_BI_avgpool2d(test_data: torch.Tensor):
+def test_convolution_2d_u85_BI_avgpool2d(test_data):
+    input, per_channel_quantization = test_data()
     model = ComboConvAvgPool2d()
     pipeline = EthosU85PipelineBI[input_t1](
         model,
-        test_data(),
+        input,
         aten_ops=[],
         exir_ops=[],
         run_on_fvp=True,
+        per_channel_quantization=per_channel_quantization,
     )
     pipeline.run()
diff --git a/backends/arm/test/ops/test_depthwise_conv.py b/backends/arm/test/ops/test_depthwise_conv.py
index 1213a04426b..4a6150317b5 100644
--- a/backends/arm/test/ops/test_depthwise_conv.py
+++ b/backends/arm/test/ops/test_depthwise_conv.py
@@ -154,7 +154,7 @@
 )
 
 # Shenanigan to get a nicer output when test fails.
-testsuite_conv2d = {
+test_data_conv2d_MI = {
     "2x2_1x6x4x4_gp6_st1": lambda: dw_conv2d_2x2_1x6x4x4_gp6_st1,
     "3x3_1x3x256x256_gp3_st1": lambda: dw_conv2d_3x3_1x3x256x256_gp3_st1,
     "3x3_1x4x256x256_gp4_nobias": lambda: dw_conv2d_3x3_1x4x256x256_gp4_nobias,
@@ -163,26 +163,45 @@
     "two_dw_conv2d": lambda: two_dw_conv2d,
 }
 
-testsuite_conv2d_u85 = {
-    "2x2_1x6x4x4_gp6_st1": lambda: dw_conv2d_2x2_1x6x4x4_gp6_st1,
-    "3x3_1x3x256x256_gp3_st1": lambda: dw_conv2d_3x3_1x3x256x256_gp3_st1,
-    "3x3_1x4x256x256_gp4_st1": lambda: dw_conv2d_3x3_1x4x256x256_gp4_st1,
-    "3x3_1x4x256x256_gp4_nobias": lambda: dw_conv2d_3x3_1x4x256x256_gp4_nobias,
+# Generate a new test set paired with per_channel_quant=True/False.
+test_data_conv2d_BI = {
+    f"{k},per_channel_quant={q}": (lambda v=v, q=q: (v(), q))
+    for (k, v) in test_data_conv2d_MI.items()
+    for q in [True, False]
 }
 
-testsuite_conv1d = {
+# Generate a new test set paired with per_channel_quant=True/False.
+test_data_conv2d_u85 = {
+    f"{k},per_channel_quant={q}": (lambda v=v, q=q: (v(), q))
+    for (k, v) in {
+        "2x2_1x6x4x4_gp6_st1": lambda: dw_conv2d_2x2_1x6x4x4_gp6_st1,
+        "3x3_1x3x256x256_gp3_st1": lambda: dw_conv2d_3x3_1x3x256x256_gp3_st1,
+        "3x3_1x4x256x256_gp4_st1": lambda: dw_conv2d_3x3_1x4x256x256_gp4_st1,
+        "3x3_1x4x256x256_gp4_nobias": lambda: dw_conv2d_3x3_1x4x256x256_gp4_nobias,
+    }.items()
+    for q in [True, False]
+}
+
+test_data_conv1d_MI = {
     "2_1x6x4_gp6_st1": lambda: dw_conv1d_2_1x6x4_gp6_st1,
     "two_dw_conv1d": lambda: two_dw_conv1d,
     "3_1x3x256_gp3_st1": lambda: dw_conv1d_3_1x3x256_gp3_st1,
     "3_1x3x14_gp3_st1": lambda: dw_conv1d_3_1x3x14_gp3_st1,
 }
 
+# Generate a new test set paired with per_channel_quant=True/False.
+test_data_conv1d_BI = {
+    f"{k},per_channel_quant={q}": (lambda v=v, q=q: (v(), q))
+    for (k, v) in test_data_conv1d_MI.items()
+    for q in [True, False]
+}
+
 
-@common.parametrize("test_module", testsuite_conv1d | testsuite_conv2d)
-def test_convolution_2d_tosa_MI_depth_wise(test_module: torch.nn.Module):
+@common.parametrize("test_data", test_data_conv1d_MI | test_data_conv2d_MI)
+def test_depthwise_convolution_2d_tosa_MI(test_data: torch.nn.Module):
     pipeline = TosaPipelineMI[input_t](
-        test_module(),
-        test_module().get_inputs(),
+        test_data(),
+        test_data().get_inputs(),
         aten_op=[],
         exir_op=exir_op,
     )
@@ -190,70 +209,84 @@ def test_convolution_2d_tosa_MI_depth_wise(test_module: torch.nn.Module):
 
 
 @pytest.mark.flaky(reruns=5)  # TODO: Investigate flakyness (MLTORCH-307)
-@common.parametrize("test_module", testsuite_conv1d | testsuite_conv2d)
-def test_convolution_2d_tosa_BI_depth_wise(test_module: torch.nn.Module):
+@common.parametrize("test_data", test_data_conv1d_BI | test_data_conv2d_BI)
+def test_depthwise_convolution_2d_tosa_BI(test_data):
+    model, per_channel_quantization = test_data()
     pipeline = TosaPipelineBI[input_t](
-        test_module(),
-        test_module().get_inputs(),
+        model,
+        model.get_inputs(),
         aten_op=[],
         exir_op=exir_op,
+        per_channel_quantization=per_channel_quantization,
     )
     pipeline.run()
 
 
 x_fails = {
-    "3x3_2x8x198x198_gp8_st3": "MLETORCH-517: Operators fail with batches > 1",
-    "two_dw_conv2d": "MLETORCH-517: Operators fail with batches > 1",
+    f"{k},per_channel_quant={q}": reason
+    for k, reason in {
+        "3x3_2x8x198x198_gp8_st3": "MLETORCH-517: Operators fail with batches > 1",
+        "two_dw_conv2d": "MLETORCH-517: Operators fail with batches > 1",
+    }.items()
+    for q in [True, False]
 }
 
 
 @common.XfailIfNoCorstone300  # TODO: MLETORCH-516
-@common.parametrize("test_module", testsuite_conv2d, x_fails)
-def test_convolution_2d_u55_BI_depth_wise(test_module: torch.nn.Module):
+@common.parametrize("test_data", test_data_conv2d_BI, x_fails)
+def test_depthwise_convolution_2d_u55_BI(test_data):
+    model, per_channel_quantization = test_data()
     pipeline = EthosU55PipelineBI[input_t](
-        test_module(),
-        test_module().get_inputs(),
+        model,
+        model.get_inputs(),
         aten_ops=[],
         exir_ops=exir_op,
         run_on_fvp=True,
+        per_channel_quantization=per_channel_quantization,
     )
     pipeline.run()
 
 
 @common.XfailIfNoCorstone300  # TODO: MLETORCH-516
-@common.parametrize("test_module", testsuite_conv1d)
-def test_convolution_1d_u55_BI_depth_wise(test_module: torch.nn.Module):
+@common.parametrize("test_data", test_data_conv1d_BI)
+def test_depthwise_convolution_1d_u55_BI(test_data):
+    model, per_channel_quantization = test_data()
     pipeline = EthosU55PipelineBI[input_t](
-        test_module(),
-        test_module().get_inputs(),
+        model,
+        model.get_inputs(),
         aten_ops=[],
         exir_ops=exir_op,
         run_on_fvp=True,
+        per_channel_quantization=per_channel_quantization,
     )
     pipeline.run()
 
 
 @common.XfailIfNoCorstone320  # TODO: MLETORCH-516
-@common.parametrize("test_module", testsuite_conv2d, x_fails)
-def test_convolution_2d_u85_BI_depth_wise(test_module: torch.nn.Module):
+@common.parametrize("test_data", test_data_conv2d_BI, x_fails)
+def test_depthwise_convolution_2d_u85_BI(test_data):
+    model, per_channel_quantization = test_data()
     pipeline = EthosU85PipelineBI[input_t](
-        test_module(),
-        test_module().get_inputs(),
+        model,
+        model.get_inputs(),
         aten_ops=[],
         exir_ops=exir_op,
         run_on_fvp=True,
+        per_channel_quantization=per_channel_quantization,
     )
     pipeline.run()
 
 
 @common.XfailIfNoCorstone320  # TODO: MLETORCH-516
-@common.parametrize("test_module", testsuite_conv1d, x_fails)
-def test_convolution_1d_u85_BI_depth_wise(test_module: torch.nn.Module):
+@common.parametrize("test_data", test_data_conv1d_BI, x_fails)
+def test_depthwise_convolution_1d_u85_BI(test_data):
+    model, per_channel_quantization = test_data()
     pipeline = EthosU85PipelineBI[input_t](
-        test_module(),
-        test_module().get_inputs(),
+        model,
+        model.get_inputs(),
         aten_ops=[],
         exir_ops=exir_op,
         run_on_fvp=True,
+        per_channel_quantization=per_channel_quantization,
     )
     pipeline.run()
diff --git a/backends/arm/test/ops/test_linear.py b/backends/arm/test/ops/test_linear.py
index 56d33097999..14f65a07192 100644
--- a/backends/arm/test/ops/test_linear.py
+++ b/backends/arm/test/ops/test_linear.py
@@ -24,8 +24,8 @@
 
 input_t1 = Tuple[torch.Tensor]
 
-test_data_suite_rank1 = {
-    # (test_name, test_data, out_features, has_bias)
+test_data_rank1_MI = {
+    # test_name: (test_data, out_features, has_bias)
     "model_linear_rank1_zeros": lambda: (
         torch.zeros(10),
         15,
@@ -58,8 +58,8 @@
     ),
 }
 
-test_data_suite_rank4 = {
-    # (test_name, test_data, out_features, has_bias)
+test_data_rank4_MI = {
+    # test_name: (test_data, out_features, has_bias)
     "model_linear_rank4_zeros": lambda: (
         torch.zeros(5, 10, 25, 20),
         30,
@@ -92,6 +92,20 @@
     ),
 }
 
+# Generate a new test set paired with per_channel_quant=True/False.
+test_data_rank1_BI = {
+    f"{k},per_channel_quant={q}": (lambda v=v, q=q: (*v(), q))
+    for (k, v) in test_data_rank1_MI.items()
+    for q in [True, False]
+}
+
+# Generate a new test set paired with per_channel_quant=True/False.
+test_data_rank4_BI = {
+    f"{k},per_channel_quant={q}": (lambda v=v, q=q: (*v(), q))
+    for (k, v) in test_data_rank4_MI.items()
+    for q in [True, False]
+}
+
 
 class Linear(torch.nn.Module):
     def __init__(
@@ -111,7 +125,7 @@ def forward(self, x):
         return self.fc(x)
 
 
-@common.parametrize("test_data", test_data_suite_rank1 | test_data_suite_rank4)
+@common.parametrize("test_data", test_data_rank1_MI | test_data_rank4_MI)
 def test_linear_tosa_MI(test_data: torch.Tensor):
     test_data, out_features, has_bias = test_data()
     in_features = test_data.shape[-1]
@@ -129,9 +143,9 @@ def test_linear_tosa_MI(test_data: torch.Tensor):
 
 
 @pytest.mark.flaky(reruns=5)  # TODO: Investigate flakyness.
-@common.parametrize("test_data", test_data_suite_rank1 | test_data_suite_rank4)
+@common.parametrize("test_data", test_data_rank1_BI | test_data_rank4_BI)
 def test_linear_tosa_BI(test_data: torch.Tensor):
-    test_data, out_features, has_bias = test_data()
+    test_data, out_features, has_bias, per_channel_quantization = test_data()
     in_features = test_data.shape[-1]
     pipeline = TosaPipelineBI[input_t1](
         Linear(
@@ -142,15 +156,16 @@ def test_linear_tosa_BI(test_data: torch.Tensor):
         (test_data,),
         aten_op,
         exir_op=[],
+        per_channel_quantization=per_channel_quantization,
         use_to_edge_transform_and_lower=True,
     )
     pipeline.run()
 
 
-@common.parametrize("test_data", test_data_suite_rank1)
+@common.parametrize("test_data", test_data_rank1_BI)
 @common.XfailIfNoCorstone300
 def test_linear_u55_BI(test_data: torch.Tensor):
-    test_data, out_features, has_bias = test_data()
+    test_data, out_features, has_bias, per_channel_quantization = test_data()
     in_features = test_data.shape[-1]
     EthosU55PipelineBI[input_t1](
         Linear(
@@ -162,28 +177,33 @@ def test_linear_u55_BI(test_data: torch.Tensor):
         aten_op,
         exir_ops=[],
         run_on_fvp=True,
+        per_channel_quantization=per_channel_quantization,
         use_to_edge_transform_and_lower=True,
     ).run()
 
 
 x_fail = {
-    "model_linear_rank4_zeros": "AssertionError: Output 0 does not match reference output.",
-    "model_linear_rank4_ones": "AssertionError: Output 0 does not match reference output.",
-    "model_linear_rank4_negative_ones": "AssertionError: Output 0 does not match reference output.",
-    "model_linear_rank4_rand": "AssertionError: Output 0 does not match reference output.",
-    "model_linear_rank4_negative_large_rand": "AssertionError: Output 0 does not match reference output.",
-    "model_linear_rank4_large_randn": "AssertionError: Output 0 does not match reference output.",
+    f"{k},per_channel_quant={q}": reason
+    for k, reason in {
+        "model_linear_rank4_zeros": "AssertionError: Output 0 does not match reference output.",
+        "model_linear_rank4_ones": "AssertionError: Output 0 does not match reference output.",
+        "model_linear_rank4_negative_ones": "AssertionError: Output 0 does not match reference output.",
+        "model_linear_rank4_rand": "AssertionError: Output 0 does not match reference output.",
+        "model_linear_rank4_negative_large_rand": "AssertionError: Output 0 does not match reference output.",
+        "model_linear_rank4_large_randn": "AssertionError: Output 0 does not match reference output.",
+    }.items()
+    for q in [True, False]
 }
 
 
 @common.parametrize(
     "test_data",
-    test_data_suite_rank1 | test_data_suite_rank4,
+    test_data_rank1_BI | test_data_rank4_BI,
     x_fail,
 )
 @common.XfailIfNoCorstone320
 def test_linear_u85_BI(test_data: torch.Tensor):
-    test_data, out_features, has_bias = test_data()
+    test_data, out_features, has_bias, per_channel_quantization = test_data()
     in_features = test_data.shape[-1]
     EthosU85PipelineBI[input_t1](
         Linear(
@@ -195,5 +215,6 @@ def test_linear_u85_BI(test_data: torch.Tensor):
         aten_op,
         exir_ops=[],
         run_on_fvp=True,
+        per_channel_quantization=per_channel_quantization,
         use_to_edge_transform_and_lower=True,
     ).run()