diff --git a/tests/sweep_framework/sweep_utils/reduction_common.py b/tests/sweep_framework/sweep_utils/reduction_common.py
new file mode 100644
index 00000000000..e110a9ec8fc
--- /dev/null
+++ b/tests/sweep_framework/sweep_utils/reduction_common.py
@@ -0,0 +1,57 @@
+# SPDX-FileCopyrightText: © 2024 Tenstorrent Inc.
+
+# SPDX-License-Identifier: Apache-2.0
+
+from typing import Optional, Tuple
+from functools import partial
+
+import torch
+import random
+import ttnn
+from tests.sweep_framework.sweep_utils.utils import gen_shapes
+from tests.tt_eager.python_api_testing.sweep_tests.generation_funcs import gen_func_with_cast_tt
+
+from tests.ttnn.utils_for_testing import check_with_pcc, start_measuring_time, stop_measuring_time
+from models.utility_functions import torch_random
+
+# Override the default timeout in seconds for hang detection.
+TIMEOUT = 30
+
+
+def run_sum(
+    input_shape,
+    dim,
+    keepdim,
+    input_a_dtype,
+    input_a_layout,
+    input_a_memory_config,
+    output_memory_config,
+    device,
+) -> list:
+    data_seed = random.randint(0, 20000000)
+    torch.manual_seed(data_seed)
+
+    torch_input_tensor_a = gen_func_with_cast_tt(
+        partial(torch_random, low=-100, high=100, dtype=torch.float32), input_a_dtype
+    )(input_shape)
+
+    dim = dim % len(input_shape)
+
+    torch_output_tensor = torch.sum(torch_input_tensor_a, dim=dim, keepdim=keepdim)
+
+    input_tensor_a = ttnn.from_torch(
+        torch_input_tensor_a,
+        dtype=input_a_dtype,
+        layout=input_a_layout,
+        device=device,
+        memory_config=input_a_memory_config,
+    )
+
+    start_time = start_measuring_time()
+    result = ttnn.sum(input_tensor_a, dim=dim, memory_config=output_memory_config)
+    output_tensor = ttnn.to_torch(result)
+    e2e_perf = stop_measuring_time(start_time)
+
+    pcc = check_with_pcc(torch_output_tensor, output_tensor, 0.999)
+    # print(f"input_shape {input_shape} pcc {pcc}")
+    return [pcc, e2e_perf]
diff --git a/tests/sweep_framework/sweeps/reduction/argmax/argmax.py b/tests/sweep_framework/sweeps/reduction/argmax/argmax.py
index 913d4178865..573f5d5733f 100644
--- a/tests/sweep_framework/sweeps/reduction/argmax/argmax.py
+++ b/tests/sweep_framework/sweeps/reduction/argmax/argmax.py
@@ -27,10 +27,38 @@
 parameters = {
     "nightly": {
         "input_shape": gen_shapes([1, 1, 1, 1], [2, 6, 128, 128], [1, 1, 1, 1], 32)
+        + gen_shapes([1, 1, 1, 1], [2, 9, 167, 128], [1, 1, 1, 1], 32)
+        + gen_shapes([1, 1, 1, 1], [2, 6, 69, 129], [1, 1, 1, 1], 15)
         + gen_shapes([1, 1, 1], [6, 128, 128], [1, 1, 1], 32)
-        + gen_shapes([1, 1], [128, 128], [1, 1], 32),
-        "dim": [0, 1, 2, 3, None],
-        "input_a_dtype": [ttnn.bfloat16, ttnn.bfloat8_b],
+        + gen_shapes([1, 1, 1], [6, 128, 128], [1, 2, 3], 3)
+        + gen_shapes([1, 1, 1], [6, 127, 257], [1, 1, 1], 16)
+        + gen_shapes([1, 1], [128, 128], [1, 1], 32)
+        + gen_shapes([1, 1], [8, 100], [2, 3], 7)
+        + gen_shapes([1, 1], [255, 255], [1, 1], 4)
+        + gen_shapes([1], [128], [1], 32)
+        + gen_shapes([1], [128], [1], 7)
+        + gen_shapes([1], [250], [3], 4),
+        "dim": [
+            0,
+            1,
+            2,
+            3,
+            None,
+            [0, 1],
+            [0, 2],
+            [0, 3],
+            [1, 2],
+            [1, 3],
+            [2, 3],
+            [0, 1, 2],
+            [0, 1, 3],
+            [0, 1, 3],
+            [0, 2, 3],
+            [1, 2, 3],
+            [0, 1, 2, 3],
+        ],
+        "keepdim": [True, False],
+        "input_a_dtype": [ttnn.float32, ttnn.bfloat16, ttnn.bfloat8_b],
         "input_layout": [ttnn.ROW_MAJOR_LAYOUT, ttnn.TILE_LAYOUT],
         "input_a_memory_config": [ttnn.DRAM_MEMORY_CONFIG, ttnn.L1_MEMORY_CONFIG],
         "output_memory_config": [ttnn.DRAM_MEMORY_CONFIG, ttnn.L1_MEMORY_CONFIG],
@@ -53,8 +81,21 @@ def invalidate_vector(test_vector) -> Tuple[bool, Optional[str]]:
             return True, "Absolute value of dim must be less or equal than the rank of input tensor"
     if test_vector["input_layout"] == ttnn.TILE_LAYOUT:
         return True, "Tiled layout not supported"
-    if test_vector["input_layout"] == ttnn.ROW_MAJOR_LAYOUT and test_vector["input_a_dtype"] == ttnn.bfloat8_b:
-        return True, "bfloat8_b is only supported on tiled layout"
+    if test_vector["input_a_dtype"] != ttnn.bfloat16:
+        return True, "Only BFLOAT16 is supported for inputs!"
+    if test_vector["input_layout"] == ttnn.ROW_MAJOR_LAYOUT and not (
+        test_vector["input_a_dtype"] == ttnn.float32 or test_vector["input_a_dtype"] == ttnn.bfloat16
+    ):
+        return True, "Row major is only supported for fp32 & fp16"
+    if not test_vector["keepdim"]:
+        return True, "keepdim = false is not supported"
+
+    device = ttnn.open_device(device_id=0)
+    if test_vector["input_a_dtype"] == ttnn.float32 and ttnn.device.is_grayskull(device):
+        return True, "Dest Fp32 mode is not supported for arch grayskull"
+    ttnn.close_device(device)
+    del device
+
     return False, None
 
 
@@ -65,6 +106,7 @@ def invalidate_vector(test_vector) -> Tuple[bool, Optional[str]]:
 def run(
     input_shape,
     dim,
+    keepdim,
     input_a_dtype,
     input_layout,
     input_a_memory_config,
@@ -83,7 +125,7 @@ def run(
     )(input_shape)
 
     golden_function = ttnn.get_golden_function(ttnn.argmax)
-    torch_output_tensor = golden_function(torch_input_tensor_a, dim=dim)
+    torch_output_tensor = golden_function(torch_input_tensor_a, dim=dim, keepdim=keepdim)
 
     input_tensor_a = ttnn.from_torch(
         torch_input_tensor_a,
diff --git a/tests/sweep_framework/sweeps/reduction/backward/prod_bw/prod_bw.py b/tests/sweep_framework/sweeps/reduction/backward/prod_bw/prod_bw.py
index 3c60e10d148..6273d857e59 100644
--- a/tests/sweep_framework/sweeps/reduction/backward/prod_bw/prod_bw.py
+++ b/tests/sweep_framework/sweeps/reduction/backward/prod_bw/prod_bw.py
@@ -27,11 +27,41 @@
 parameters = {
     "xfail": {
         "input_shape": gen_shapes([1, 1, 32, 32], [6, 12, 256, 256], [1, 1, 32, 32], 16)
-        + gen_shapes([1, 1, 1, 1], [6, 12, 256, 256], [1, 1, 1, 1], 2),
-        "dim": [0, 1, 2, 3],
-        "grad_dtype": [ttnn.bfloat16, ttnn.bfloat8_b],
-        "input_a_dtype": [ttnn.bfloat16],
-        "input_layout": [ttnn.TILE_LAYOUT],
+        + gen_shapes([1, 1, 1, 1], [6, 12, 256, 256], [1, 1, 1, 1], 2)
+        + gen_shapes([3, 4, 5, 6], [6, 12, 256, 256], [7, 8, 9, 10], 2)
+        + gen_shapes([1, 1, 1, 1], [6, 12, 187, 188], [1, 1, 1, 1], 7)
+        + gen_shapes([1, 32, 64], [6, 48, 128], [1, 1, 1], 2)
+        + gen_shapes([1, 32, 64], [6, 77, 128], [1, 1, 1], 7)
+        + gen_shapes([1, 32, 64], [6, 10222, 1023], [1, 1, 1], 8)
+        + gen_shapes([1, 1], [6, 6], [1, 1], 2)
+        + gen_shapes([1, 1], [7, 7], [1, 2], 3)
+        + gen_shapes([1, 1], [8, 8], [1, 3], 4)
+        + gen_shapes([1], [4], [1], 2)
+        + gen_shapes([1], [14], [11], 12)
+        + gen_shapes([1], [24], [21], 22),
+        "dim": [
+            0,
+            1,
+            2,
+            3,
+            None,
+            [0, 1],
+            [0, 2],
+            [0, 3],
+            [1, 2],
+            [1, 3],
+            [2, 3],
+            [0, 1, 2],
+            [0, 1, 3],
+            [0, 1, 3],
+            [0, 2, 3],
+            [1, 2, 3],
+            [0, 1, 2, 3],
+        ],
+        "keepdim": [True, False],
+        "grad_dtype": [ttnn.float32, ttnn.bfloat16, ttnn.bfloat8_b],
+        "input_a_dtype": [ttnn.bfloat16, ttnn.bfloat8_b],
+        "input_layout": [ttnn.ROW_MAJOR_LAYOUT, ttnn.TILE_LAYOUT],
         "grad_memory_config": [ttnn.DRAM_MEMORY_CONFIG, ttnn.L1_MEMORY_CONFIG],
         "input_a_memory_config": [ttnn.DRAM_MEMORY_CONFIG, ttnn.L1_MEMORY_CONFIG],
         "output_memory_config": [ttnn.DRAM_MEMORY_CONFIG, ttnn.L1_MEMORY_CONFIG],
@@ -45,10 +75,26 @@
 def invalidate_vector(test_vector) -> Tuple[bool, Optional[str]]:
     if test_vector["input_layout"] == ttnn.ROW_MAJOR_LAYOUT:
         return True, "Unary operation requires tensor to be in Tile layout when working with non-sharded input tensor"
-    if test_vector["input_layout"] == ttnn.ROW_MAJOR_LAYOUT and (
-        test_vector["grad_dtype"] == ttnn.bfloat8_b or test_vector["input_a_dtype"] == ttnn.bfloat8_b
-    ):
-        return True, "bfloat8_b is only supported on tiled layout"
+    if test_vector["input_layout"] == ttnn.ROW_MAJOR_LAYOUT:
+        if (test_vector["input_a_dtype"] == ttnn.float32 and test_vector["grad_dtype"] == ttnn.float32) or (
+            test_vector["input_a_dtype"] == ttnn.bfloat16 or test_vector["grad_dtype"] == ttnn.bfloat16
+        ):
+            return False, None
+        else:
+            return True, "Row major is only supported for fp32 & fp16"
+    if not test_vector["keepdim"]:
+        return True, "keepdim = false is not supported"
+    if not isinstance(test_vector["dim"], int):
+        return True, "dim can only be integer value"
+
+    device = ttnn.open_device(device_id=0)
+    if (
+        test_vector["input_a_dtype"] == ttnn.float32 or test_vector["grad_dtype"] == ttnn.float32
+    ) and ttnn.device.is_grayskull(device):
+        return True, "Dest Fp32 mode is not supported for arch grayskull"
+    ttnn.close_device(device)
+    del device
+
     return False, None
 
 
@@ -59,6 +105,7 @@ def invalidate_vector(test_vector) -> Tuple[bool, Optional[str]]:
 def run(
     input_shape,
     dim,
+    keepdim,
     grad_dtype,
     input_a_dtype,
     input_layout,
@@ -77,10 +124,7 @@ def run(
     torch_input_tensor_a.requires_grad = True
     torch_input_tensor_a.retain_grad()
 
-    max_dim = len(input_shape) - 1
-    dim = random.randint(-max_dim - 1, max_dim)
-
-    intermediate_result = torch.prod(torch_input_tensor_a, dim=dim, keepdim=True)
+    intermediate_result = torch.prod(torch_input_tensor_a, dim=dim, keepdim=keepdim)
     torch_grad_tensor = gen_func_with_cast_tt(
         partial(torch_random, low=-100, high=100, dtype=torch.float32), grad_dtype
     )(intermediate_result.shape)
diff --git a/tests/sweep_framework/sweeps/reduction/mean/mean.py b/tests/sweep_framework/sweeps/reduction/mean/mean.py
index b2c93e0c0d8..c5662275dbb 100644
--- a/tests/sweep_framework/sweeps/reduction/mean/mean.py
+++ b/tests/sweep_framework/sweeps/reduction/mean/mean.py
@@ -27,21 +27,74 @@
 # Developers can create their own generator functions and pass them to the parameters as inputs.
 parameters = {
     "nightly": {
-        "input_shape": gen_shapes([1, 1, 1, 1], [2, 6, 128, 128], [1, 1, 32, 32], 32),
-        "input_a_dtype": [ttnn.bfloat16, ttnn.bfloat8_b],
-        "input_layout": [ttnn.TILE_LAYOUT],
+        "input_shape": gen_shapes([1, 1, 1, 1], [2, 6, 128, 128], [1, 1, 32, 32], 32)
+        + gen_shapes([1, 1, 1, 1], [2, 6, 110, 130], [1, 1, 32, 32], 31)
+        + gen_shapes([1, 1, 1, 1], [2, 6, 120, 140], [1, 1, 32, 32], 17)
+        + gen_shapes([1, 1, 1], [2, 6, 128], [1, 1, 32], 8)
+        + gen_shapes([1, 1, 1], [2, 7, 100], [1, 1, 64], 16)
+        + gen_shapes([1, 1, 1], [2, 8, 255], [1, 1, 122], 17)
+        + gen_shapes([1, 1], [2, 26], [1, 11], 2)
+        + gen_shapes([1, 1], [2, 36], [1, 12], 3)
+        + gen_shapes([1, 1], [2, 46], [1, 13], 4)
+        + gen_shapes([1], [2, 6], [1, 1], 2)
+        + gen_shapes([1], [2, 6], [1, 1], 2)
+        + gen_shapes([1], [10, 10], [2, 3], 2),
+        "dim": [
+            0,
+            1,
+            2,
+            3,
+            None,
+            [0, 1],
+            [0, 2],
+            [0, 3],
+            [1, 2],
+            [1, 3],
+            [2, 3],
+            [0, 1, 2],
+            [0, 1, 3],
+            [0, 1, 3],
+            [0, 2, 3],
+            [1, 2, 3],
+            [0, 1, 2, 3],
+        ],
+        "keepdim": [True, False],
+        "input_a_dtype": [ttnn.float32, ttnn.bfloat16, ttnn.bfloat8_b],
+        "input_layout": [ttnn.ROW_MAJOR_LAYOUT, ttnn.TILE_LAYOUT],
         "input_a_memory_config": [ttnn.DRAM_MEMORY_CONFIG, ttnn.L1_MEMORY_CONFIG],
         "output_memory_config": [ttnn.DRAM_MEMORY_CONFIG, ttnn.L1_MEMORY_CONFIG],
     },
 }
 
 
+# Invalidate vector is called during the generation phase where each vector will be passed in.
+# If invalidated, the vector will still be stored but will be skipped.
+# Returns False, None if the vector is valid, and True, str with a reason for invalidation if it is invalid.
+def invalidate_vector(test_vector) -> Tuple[bool, Optional[str]]:
+    if test_vector["input_layout"] == ttnn.ROW_MAJOR_LAYOUT and not (
+        test_vector["input_a_dtype"] == ttnn.float32 or test_vector["input_a_dtype"] == ttnn.bfloat16
+    ):
+        return True, "Row major is only supported for fp32 & fp16"
+    if not test_vector["keepdim"]:
+        return True, "keepdim = false is not supported"
+
+    device = ttnn.open_device(device_id=0)
+    if test_vector["input_a_dtype"] == ttnn.float32 and ttnn.device.is_grayskull(device):
+        return True, "Dest Fp32 mode is not supported for arch grayskull"
+    ttnn.close_device(device)
+    del device
+
+    return False, None
+
+
 # This is the run instructions for the test, defined by the developer.
 # The run function must take the above-defined parameters as inputs.
 # The runner will call this run function with each test vector, and the returned results from this function will be stored.
 # If you defined a mesh_device_fixture above, the object you yielded will be passed into this function as 'device'. Otherwise, it will be the default ttnn device opened by the infra.
 def run(
     input_shape,
+    dim,
+    keepdim,
     input_a_dtype,
     input_layout,
     input_a_memory_config,
@@ -60,7 +113,7 @@ def run(
     )(input_shape)
 
     golden_function = ttnn.get_golden_function(ttnn.mean)
-    torch_output_tensor = golden_function(torch_input_tensor_a, dim=-1, keepdim=True)
+    torch_output_tensor = golden_function(torch_input_tensor_a, dim=dim, keepdim=keepdim)
 
     input_tensor_a = ttnn.from_torch(
         torch_input_tensor_a,
@@ -71,7 +124,7 @@ def run(
     )
 
     start_time = start_measuring_time()
-    output_tensor = ttnn.mean(input_tensor_a, dim=-1, memory_config=output_memory_config)
+    output_tensor = ttnn.mean(input_tensor_a, dim=dim, memory_config=output_memory_config)
     output_tensor = ttnn.to_torch(output_tensor)
 
     e2e_perf = stop_measuring_time(start_time)
diff --git a/tests/sweep_framework/sweeps/reduction/prod.py b/tests/sweep_framework/sweeps/reduction/prod.py
index e5a098dcffa..39add1f91fd 100644
--- a/tests/sweep_framework/sweeps/reduction/prod.py
+++ b/tests/sweep_framework/sweeps/reduction/prod.py
@@ -30,16 +30,59 @@
         + gen_shapes([32, 32], [256, 256], [32, 32], 2)
         + gen_shapes([1, 1, 1, 1], [6, 12, 256, 256], [1, 1, 1, 1], 2)
         + gen_shapes([1, 1, 1], [12, 256, 256], [1, 1, 1], 2)
-        + gen_shapes([1, 1], [256, 256], [1, 1], 2),
-        "dim": [0, 1, 2, 3],
-        "input_a_dtype": [ttnn.bfloat16, ttnn.bfloat8_b],
-        "input_a_layout": [ttnn.TILE_LAYOUT],
+        + gen_shapes([1, 1], [256, 256], [1, 1], 2)
+        + gen_shapes([1], [256], [1], 8)
+        + gen_shapes([1, 1, 1, 1], [6, 12, 200, 255], [1, 1, 1, 1], 5)
+        + gen_shapes([1, 1, 1], [12, 555, 128], [1, 1, 1], 4)
+        + gen_shapes([1, 1], [32, 32], [1, 1], 32),
+        "dim": [
+            0,
+            1,
+            2,
+            3,
+            None,
+            [0, 1],
+            [0, 2],
+            [0, 3],
+            [1, 2],
+            [1, 3],
+            [2, 3],
+            [0, 1, 2],
+            [0, 1, 3],
+            [0, 1, 3],
+            [0, 2, 3],
+            [1, 2, 3],
+            [0, 1, 2, 3],
+        ],
+        "keepdim": [True, False],
+        "input_a_dtype": [ttnn.float32, ttnn.bfloat16, ttnn.bfloat8_b],
+        "input_a_layout": [ttnn.ROW_MAJOR_LAYOUT, ttnn.TILE_LAYOUT],
         "input_a_memory_config": [ttnn.DRAM_MEMORY_CONFIG, ttnn.L1_MEMORY_CONFIG],
         "output_memory_config": [ttnn.DRAM_MEMORY_CONFIG, ttnn.L1_MEMORY_CONFIG],
     },
 }
 
 
+# Invalidate vector is called during the generation phase where each vector will be passed in.
+# If invalidated, the vector will still be stored but will be skipped.
+# Returns False, None if the vector is valid, and True, str with a reason for invalidation if it is invalid.
+def invalidate_vector(test_vector) -> Tuple[bool, Optional[str]]:
+    if test_vector["input_a_layout"] == ttnn.ROW_MAJOR_LAYOUT and not (
+        test_vector["input_a_dtype"] == ttnn.float32 or test_vector["input_a_dtype"] == ttnn.bfloat16
+    ):
+        return True, "Row major is only supported for fp32 & fp16"
+    if not test_vector["keepdim"]:
+        return True, "keepdim = false is not supported"
+
+    device = ttnn.open_device(device_id=0)
+    if test_vector["input_a_dtype"] == ttnn.float32 and ttnn.device.is_grayskull(device):
+        return True, "Dest Fp32 mode is not supported for arch grayskull"
+    ttnn.close_device(device)
+    del device
+
+    return False, None
+
+
 # This is the run instructions for the test, defined by the developer.
 # The run function must take the above-defined parameters as inputs.
 # The runner will call this run function with each test vector, and the returned results from this function will be stored.
@@ -47,6 +90,7 @@
 def run(
     input_shape,
     dim,
+    keepdim,
     input_a_dtype,
     input_a_layout,
     input_a_memory_config,
@@ -63,7 +107,7 @@ def run(
 
     dim = dim % len(input_shape)
 
-    torch_output_tensor = torch.prod(torch_input_tensor_a, dim=dim, keepdim=True)
+    torch_output_tensor = torch.prod(torch_input_tensor_a, dim=dim, keepdim=keepdim)
 
     input_tensor_a = ttnn.from_torch(
         torch_input_tensor_a,
diff --git a/tests/sweep_framework/sweeps/reduction/std/std.py b/tests/sweep_framework/sweeps/reduction/std/std.py
index a39657d49e0..4f95723031b 100644
--- a/tests/sweep_framework/sweeps/reduction/std/std.py
+++ b/tests/sweep_framework/sweeps/reduction/std/std.py
@@ -27,21 +27,74 @@
 # Developers can create their own generator functions and pass them to the parameters as inputs.
 parameters = {
     "xfail": {
-        "input_shape": gen_shapes([1, 1, 1, 1], [2, 6, 128, 128], [1, 1, 32, 32], 32),
-        "input_a_dtype": [ttnn.bfloat16, ttnn.bfloat8_b],
-        "input_layout": [ttnn.TILE_LAYOUT],
+        "input_shape": gen_shapes([1, 1, 1, 1], [2, 6, 128, 128], [1, 1, 32, 32], 32)
+        + gen_shapes([1, 1, 1, 1], [2, 6, 254, 129], [1, 1, 20, 33], 33)
+        + gen_shapes([1, 1, 1, 1], [2, 7, 255, 130], [1, 1, 21, 34], 34)
+        + gen_shapes([1, 1, 1], [2, 6, 254], [1, 1, 32], 8)
+        + gen_shapes([1, 1, 1], [4, 12, 255], [1, 1, 32], 16)
+        + gen_shapes([1, 1, 1], [8, 18, 256], [1, 1, 32], 32)
+        + gen_shapes([1, 1], [2, 6], [1, 1], 2)
+        + gen_shapes([1, 1], [3, 7], [1, 1], 2)
+        + gen_shapes([1, 1], [4, 8], [1, 1], 2)
+        + gen_shapes([1], [32], [1], 4)
+        + gen_shapes([1], [33], [1], 5)
+        + gen_shapes([1], [34], [1], 6),
+        "dim": [
+            0,
+            1,
+            2,
+            3,
+            None,
+            [0, 1],
+            [0, 2],
+            [0, 3],
+            [1, 2],
+            [1, 3],
+            [2, 3],
+            [0, 1, 2],
+            [0, 1, 3],
+            [0, 1, 3],
+            [0, 2, 3],
+            [1, 2, 3],
+            [0, 1, 2, 3],
+        ],
+        "keepdim": [True, False],
+        "input_a_dtype": [ttnn.float32, ttnn.bfloat16, ttnn.bfloat8_b],
+        "input_layout": [ttnn.ROW_MAJOR_LAYOUT, ttnn.TILE_LAYOUT],
         "input_a_memory_config": [ttnn.DRAM_MEMORY_CONFIG, ttnn.L1_MEMORY_CONFIG],
         "output_memory_config": [ttnn.DRAM_MEMORY_CONFIG, ttnn.L1_MEMORY_CONFIG],
     },
 }
 
 
+# Invalidate vector is called during the generation phase where each vector will be passed in.
+# If invalidated, the vector will still be stored but will be skipped.
+# Returns False, None if the vector is valid, and True, str with a reason for invalidation if it is invalid.
+def invalidate_vector(test_vector) -> Tuple[bool, Optional[str]]:
+    if test_vector["input_layout"] == ttnn.ROW_MAJOR_LAYOUT and not (
+        test_vector["input_a_dtype"] == ttnn.float32 or test_vector["input_a_dtype"] == ttnn.bfloat16
+    ):
+        return True, "Row major is only supported for fp32 & fp16"
+    if not test_vector["keepdim"]:
+        return True, "keepdim = false is not supported"
+
+    device = ttnn.open_device(device_id=0)
+    if test_vector["input_a_dtype"] == ttnn.float32 and ttnn.device.is_grayskull(device):
+        return True, "Dest Fp32 mode is not supported for arch grayskull"
+    ttnn.close_device(device)
+    del device
+
+    return False, None
+
+
 # This is the run instructions for the test, defined by the developer.
 # The run function must take the above-defined parameters as inputs.
 # The runner will call this run function with each test vector, and the returned results from this function will be stored.
 # If you defined a mesh_device_fixture above, the object you yielded will be passed into this function as 'device'. Otherwise, it will be the default ttnn device opened by the infra.
 def run(
     input_shape,
+    dim,
+    keepdim,
     input_a_dtype,
     input_layout,
     input_a_memory_config,
@@ -60,7 +113,7 @@ def run(
     )(input_shape)
 
     golden_function = ttnn.get_golden_function(ttnn.std)
-    torch_output_tensor = golden_function(torch_input_tensor_a, dim=-1, keepdim=True)
+    torch_output_tensor = golden_function(torch_input_tensor_a, dim=dim, keepdim=keepdim)
 
     input_tensor_a = ttnn.from_torch(
         torch_input_tensor_a,
@@ -71,7 +124,7 @@ def run(
     )
 
     start_time = start_measuring_time()
-    output_tensor = ttnn.std(input_tensor_a, dim=-1, memory_config=output_memory_config)
+    output_tensor = ttnn.std(input_tensor_a, dim=dim, memory_config=output_memory_config)
     output_tensor = ttnn.to_torch(output_tensor)
 
     e2e_perf = stop_measuring_time(start_time)
diff --git a/tests/sweep_framework/sweeps/reduction/sum.py b/tests/sweep_framework/sweeps/reduction/sum.py
index cddaa82be78..7e89c43be05 100644
--- a/tests/sweep_framework/sweeps/reduction/sum.py
+++ b/tests/sweep_framework/sweeps/reduction/sum.py
@@ -13,6 +13,7 @@
 
 from tests.ttnn.utils_for_testing import check_with_pcc, start_measuring_time, stop_measuring_time
 from models.utility_functions import torch_random
+from tests.sweep_framework.sweep_utils.reduction_common import run_sum
 
 # Override the default timeout in seconds for hang detection.
 TIMEOUT = 30
@@ -30,16 +31,59 @@
         + gen_shapes([32, 32], [256, 256], [32, 32], 2)
         + gen_shapes([1, 1, 1, 1], [6, 12, 256, 256], [1, 1, 1, 1], 2)
         + gen_shapes([1, 1, 1], [12, 256, 256], [1, 1, 1], 2)
-        + gen_shapes([1, 1], [256, 256], [1, 1], 2),
-        "dim": [0, 1, 2, 3],
-        "input_a_dtype": [ttnn.bfloat16, ttnn.bfloat8_b],
-        "input_a_layout": [ttnn.TILE_LAYOUT],
+        + gen_shapes([1, 1], [256, 256], [1, 1], 2)
+        + gen_shapes([1, 1, 1, 1], [6, 12, 100, 100], [1, 1, 1, 1], 2)
+        + gen_shapes([1, 1, 1], [12, 128, 128], [1, 1, 1], 7)
+        + gen_shapes([1, 1], [11, 11], [1, 1], 2)
+        + gen_shapes([1], [256], [16], 2),
+        "dim": [
+            0,
+            1,
+            2,
+            3,
+            None,
+            [0, 1],
+            [0, 2],
+            [0, 3],
+            [1, 2],
+            [1, 3],
+            [2, 3],
+            [0, 1, 2],
+            [0, 1, 3],
+            [0, 1, 3],
+            [0, 2, 3],
+            [1, 2, 3],
+            [0, 1, 2, 3],
+        ],
+        "keepdim": [True, False],
+        "input_a_dtype": [ttnn.float32, ttnn.bfloat16, ttnn.bfloat8_b],
+        "input_a_layout": [ttnn.ROW_MAJOR_LAYOUT, ttnn.TILE_LAYOUT],
         "input_a_memory_config": [ttnn.DRAM_MEMORY_CONFIG, ttnn.L1_MEMORY_CONFIG],
         "output_memory_config": [ttnn.DRAM_MEMORY_CONFIG, ttnn.L1_MEMORY_CONFIG],
     },
 }
 
 
+# Invalidate vector is called during the generation phase where each vector will be passed in.
+# If invalidated, the vector will still be stored but will be skipped.
+# Returns False, None if the vector is valid, and True, str with a reason for invalidation if it is invalid.
+def invalidate_vector(test_vector) -> Tuple[bool, Optional[str]]:
+    if test_vector["input_a_layout"] == ttnn.ROW_MAJOR_LAYOUT and not (
+        test_vector["input_a_dtype"] == ttnn.float32 or test_vector["input_a_dtype"] == ttnn.bfloat16
+    ):
+        return True, "Row major is only supported for fp32 & fp16"
+    if not test_vector["keepdim"]:
+        return True, "keepdim = false is not supported"
+
+    device = ttnn.open_device(device_id=0)
+    if test_vector["input_a_dtype"] == ttnn.float32 and ttnn.device.is_grayskull(device):
+        return True, "Dest Fp32 mode is not supported for arch grayskull"
+    ttnn.close_device(device)
+    del device
+
+    return False, None
+
+
 # This is the run instructions for the test, defined by the developer.
 # The run function must take the above-defined parameters as inputs.
 # The runner will call this run function with each test vector, and the returned results from this function will be stored.
@@ -47,6 +91,7 @@
 def run(
     input_shape,
     dim,
+    keepdim,
     input_a_dtype,
     input_a_layout,
     input_a_memory_config,
@@ -54,31 +99,23 @@ def run(
     *,
     device,
 ) -> list:
-    data_seed = random.randint(0, 20000000)
-    torch.manual_seed(data_seed)
+    return run_sum(
+        input_shape, dim, keepdim, input_a_dtype, input_a_layout, input_a_memory_config, output_memory_config, device
+    )
 
-    torch_input_tensor_a = gen_func_with_cast_tt(
-        partial(torch_random, low=-100, high=100, dtype=torch.float32), input_a_dtype
-    )(input_shape)
 
-    dim = dim % len(input_shape)
-    # print(f"dim {dim} input_shape {input_shape} input_a_dtype {input_a_dtype}")
+import pytest
 
-    torch_output_tensor = torch.sum(torch_input_tensor_a, dim=dim, keepdim=True)
 
-    input_tensor_a = ttnn.from_torch(
-        torch_input_tensor_a,
-        dtype=input_a_dtype,
-        layout=input_a_layout,
-        device=device,
-        memory_config=input_a_memory_config,
+@pytest.mark.parametrize(
+    "input_shape, dim, input_a_dtype, input_a_layout, input_a_memory_config, output_memory_config",
+    [
+        ([7, 32, 4, 96], 3, True, ttnn.float32, ttnn.TILE_LAYOUT, ttnn.L1_MEMORY_CONFIG, ttnn.L1_MEMORY_CONFIG),
+    ],
+)
+def test_reduction_sum_localrun_fail_only(
+    device, input_shape, dim, keepdim, input_a_dtype, input_a_layout, input_a_memory_config, output_memory_config
+):
+    run_sum(
+        input_shape, dim, keepdim, input_a_dtype, input_a_layout, input_a_memory_config, output_memory_config, device
     )
-
-    start_time = start_measuring_time()
-    result = ttnn.sum(input_tensor_a, dim=dim, memory_config=output_memory_config)
-    output_tensor = ttnn.to_torch(result)
-    e2e_perf = stop_measuring_time(start_time)
-
-    pcc = check_with_pcc(torch_output_tensor, output_tensor, 0.999)
-    # print(f"input_shape {input_shape} pcc {pcc}")
-    return [pcc, e2e_perf]
diff --git a/tests/sweep_framework/sweeps/reduction/topk/topk.py b/tests/sweep_framework/sweeps/reduction/topk/topk.py
index e38a9134e5c..d65b9e498d9 100644
--- a/tests/sweep_framework/sweeps/reduction/topk/topk.py
+++ b/tests/sweep_framework/sweeps/reduction/topk/topk.py
@@ -28,12 +28,36 @@
 parameters = {
     "nightly": {
         "input_shape": gen_shapes([1, 1, 32, 64], [2, 6, 128, 128], [1, 1, 32, 64], 64)
-        + gen_shapes([1, 32, 64], [12, 256, 1024], [1, 32, 64], 8)
-        + gen_shapes([32, 64], [256, 1024], [32, 64], 8),
-        "dim": [-1, -2, -3, -4],
-        "largest": [True],
+        + gen_shapes([1, 1, 33, 65], [2, 6, 127, 129], [1, 1, 33, 63], 128)
+        + gen_shapes([1, 1, 31, 63], [2, 6, 128, 128], [1, 1, 32, 64], 7)
+        + gen_shapes([1, 32, 64], [12, 200, 1025], [1, 32, 64], 8)
+        + gen_shapes([1, 32, 64], [12, 256, 1023], [1, 32, 164], 9)
+        + gen_shapes([1, 7, 20], [12, 300, 1024], [1, 32, 64], 10)
+        + gen_shapes([32, 64], [256, 1024], [32, 64], 8)
+        + gen_shapes([32, 6], [256, 404], [32, 264], 18)
+        + gen_shapes([32, 17], [256, 124], [32, 624], 28),
+        "dim": [
+            0,
+            1,
+            2,
+            3,
+            None,
+            [0, 1],
+            [0, 2],
+            [0, 3],
+            [1, 2],
+            [1, 3],
+            [2, 3],
+            [0, 1, 2],
+            [0, 1, 3],
+            [0, 1, 3],
+            [0, 2, 3],
+            [1, 2, 3],
+            [0, 1, 2, 3],
+        ],
+        "largest": [True, False],
         "k": [32],  # only k = 32 is supported for now
-        "input_a_dtype": [ttnn.bfloat16, ttnn.bfloat8_b],
+        "input_a_dtype": [ttnn.float32, ttnn.bfloat16, ttnn.bfloat8_b],
         "input_layout": [ttnn.TILE_LAYOUT, ttnn.ROW_MAJOR_LAYOUT],
         "input_a_memory_config": [ttnn.DRAM_MEMORY_CONFIG, ttnn.L1_MEMORY_CONFIG],
         "output_memory_config": [ttnn.DRAM_MEMORY_CONFIG, ttnn.L1_MEMORY_CONFIG],
@@ -42,10 +66,28 @@
         "input_shape": gen_shapes([1, 1, 32, 64], [6, 12, 256, 1024], [1, 1, 32, 64], 64)
         + gen_shapes([1, 32, 64], [12, 256, 1024], [1, 32, 64], 8)
         + gen_shapes([32, 64], [256, 1024], [32, 64], 8),
-        "dim": [-1, -2, -3, -4],
+        "dim": [
+            0,
+            1,
+            2,
+            3,
+            None,
+            [0, 1],
+            [0, 2],
+            [0, 3],
+            [1, 2],
+            [1, 3],
+            [2, 3],
+            [0, 1, 2],
+            [0, 1, 3],
+            [0, 1, 3],
+            [0, 2, 3],
+            [1, 2, 3],
+            [0, 1, 2, 3],
+        ],
         "largest": [True, False],
         "k": [32],
-        "input_a_dtype": [ttnn.bfloat16, ttnn.bfloat8_b],
+        "input_a_dtype": [ttnn.float32, ttnn.bfloat16, ttnn.bfloat8_b],
         "input_layout": [ttnn.TILE_LAYOUT, ttnn.ROW_MAJOR_LAYOUT],
         "input_a_memory_config": [ttnn.DRAM_MEMORY_CONFIG, ttnn.L1_MEMORY_CONFIG],
         "output_memory_config": [ttnn.DRAM_MEMORY_CONFIG, ttnn.L1_MEMORY_CONFIG],
@@ -65,8 +107,17 @@ def invalidate_vector(test_vector) -> Tuple[bool, Optional[str]]:
         return True, "Absolute value of dim must be less or equal than the rank of input tensor"
     if test_vector["input_layout"] == ttnn.ROW_MAJOR_LAYOUT:
         return True, "Unary operation requires tensor to be in Tile layout when working with non-sharded input tensor"
-    if test_vector["input_layout"] == ttnn.ROW_MAJOR_LAYOUT and test_vector["input_a_dtype"] == ttnn.bfloat8_b:
-        return True, "bfloat8_b is only supported on tiled layout"
+    if test_vector["input_layout"] == ttnn.ROW_MAJOR_LAYOUT and not (
+        test_vector["input_a_dtype"] == ttnn.float32 or test_vector["input_a_dtype"] == ttnn.bfloat16
+    ):
+        return True, "Row major is only supported for fp32 & fp16"
+
+    device = ttnn.open_device(device_id=0)
+    if test_vector["input_a_dtype"] == ttnn.float32 and ttnn.device.is_grayskull(device):
+        return True, "Dest Fp32 mode is not supported for arch grayskull"
+    ttnn.close_device(device)
+    del device
+
     return False, None
 
 
diff --git a/tests/sweep_framework/sweeps/reduction/var/var.py b/tests/sweep_framework/sweeps/reduction/var/var.py
index 27f5dc70157..ee158d18869 100644
--- a/tests/sweep_framework/sweeps/reduction/var/var.py
+++ b/tests/sweep_framework/sweeps/reduction/var/var.py
@@ -27,21 +27,77 @@
 # Developers can create their own generator functions and pass them to the parameters as inputs.
 parameters = {
     "xfail": {
-        "input_shape": gen_shapes([1, 1, 1, 1], [2, 6, 128, 128], [1, 1, 32, 32], 32),
-        "input_a_dtype": [ttnn.bfloat16, ttnn.bfloat8_b],
-        "input_layout": [ttnn.TILE_LAYOUT],
+        "input_shape": gen_shapes([1, 1, 1, 1], [2, 6, 128, 128], [1, 1, 32, 32], 32)
+        + gen_shapes([1, 1, 1, 1], [2, 6, 128, 255], [1, 1, 132, 132], 42)
+        + gen_shapes([1, 1, 1, 1], [2, 6, 128, 255], [1, 1, 132, 132], 52)
+        + gen_shapes([1, 1, 1], [2, 6, 128], [1, 1, 32], 32)
+        + gen_shapes([1, 1, 1], [2, 16, 129], [1, 1, 64], 16)
+        + gen_shapes([1, 1, 1], [2, 26, 130], [1, 1, 128], 8)
+        + gen_shapes([1, 1], [2, 6], [1, 1], 4)
+        + gen_shapes([1, 1], [2, 16], [1, 1], 5)
+        + gen_shapes([1, 1], [2, 26], [1, 1], 6)
+        + gen_shapes([1], [8], [1], 2)
+        + gen_shapes([1], [18], [1], 2)
+        + gen_shapes([1], [28], [1], 2),
+        "dim": [
+            0,
+            1,
+            2,
+            3,
+            None,
+            [0, 1],
+            [0, 2],
+            [0, 3],
+            [1, 2],
+            [1, 3],
+            [2, 3],
+            [0, 1, 2],
+            [0, 1, 3],
+            [0, 1, 3],
+            [0, 2, 3],
+            [1, 2, 3],
+            [0, 1, 2, 3],
+        ],
+        "keepdim": [True, False],
+        "input_a_dtype": [ttnn.float32, ttnn.bfloat16, ttnn.bfloat8_b],
+        "input_layout": [ttnn.ROW_MAJOR_LAYOUT, ttnn.TILE_LAYOUT],
         "input_a_memory_config": [ttnn.DRAM_MEMORY_CONFIG, ttnn.L1_MEMORY_CONFIG],
         "output_memory_config": [ttnn.DRAM_MEMORY_CONFIG, ttnn.L1_MEMORY_CONFIG],
     },
 }
 
 
+# Invalidate vector is called during the generation phase where each vector will be passed in.
+# If invalidated, the vector will still be stored but will be skipped.
+# Returns False, None if the vector is valid, and True, str with a reason for invalidation if it is invalid.
+def invalidate_vector(test_vector) -> Tuple[bool, Optional[str]]:
+    if test_vector["input_layout"] == ttnn.ROW_MAJOR_LAYOUT and not (
+        test_vector["input_a_dtype"] == ttnn.float32 or test_vector["input_a_dtype"] == ttnn.bfloat16
+    ):
+        return True, "Row major is only supported for fp32 & fp16"
+    if not test_vector["keepdim"]:
+        return True, "keepdim = false is not supported"
+
+    if len(test_vector["input_shape"]) < 2:
+        return True, "For var with scalar(1-D) input, degrees of freedom will be <= 0."
+
+    device = ttnn.open_device(device_id=0)
+    if test_vector["input_a_dtype"] == ttnn.float32 and ttnn.device.is_grayskull(device):
+        return True, "Dest Fp32 mode is not supported for arch grayskull"
+    ttnn.close_device(device)
+    del device
+
+    return False, None
+
+
 # This is the run instructions for the test, defined by the developer.
 # The run function must take the above-defined parameters as inputs.
 # The runner will call this run function with each test vector, and the returned results from this function will be stored.
 # If you defined a mesh_device_fixture above, the object you yielded will be passed into this function as 'device'. Otherwise, it will be the default ttnn device opened by the infra.
 def run(
     input_shape,
+    dim,
+    keepdim,
     input_a_dtype,
     input_layout,
     input_a_memory_config,
@@ -60,7 +116,7 @@ def run(
     )(input_shape)
 
     golden_function = ttnn.get_golden_function(ttnn.var)
-    torch_output_tensor = golden_function(torch_input_tensor_a, dim=-1, keepdim=True)
+    torch_output_tensor = golden_function(torch_input_tensor_a, dim=dim, keepdim=keepdim)
 
     input_tensor_a = ttnn.from_torch(
         torch_input_tensor_a,
@@ -71,7 +127,7 @@ def run(
     )
 
     start_time = start_measuring_time()
-    output_tensor = ttnn.var(input_tensor_a, dim=-1, memory_config=output_memory_config)
+    output_tensor = ttnn.var(input_tensor_a, dim=dim, memory_config=output_memory_config)
     output_tensor = ttnn.to_torch(output_tensor)
 
     e2e_perf = stop_measuring_time(start_time)