#15857: Binary Forge Sweep Tests Set2 (#16087)

### Ticket
#15857 

### Problem description
Add forge sweep tests for binary ops

### What's changed
Added automation script to generate forge sweep tests for the following
ops:
- eq
- ge
- gt
- lt
- ne
- logical_and
- logical_or
- div
- remainder
- where

### Checklist
- [ ] Post commit CI passes

.github/workflows/ttnn-run-sweeps.yaml

@@ -234,24 +234,31 @@ on:
           - eltwise.binary.multiply.multiply_scalar_pytorch2
           - eltwise.binary.div.div
           - eltwise.binary.div.div_tensor_pytorch2
+          - eltwise.binary.div.div_forge
           - eltwise.binary.div_no_nan.div_no_nan
           - eltwise.binary.logical_or.logical_or_
           - eltwise.binary.logical_or.logical_or
           - eltwise.binary.logical_or.logical_or_output
+          - eltwise.binary.logical_or.logical_or_forge
           - eltwise.binary.logical_xor.logical_xor_
           - eltwise.binary.logical_xor.logical_xor
           - eltwise.binary.logical_and.logical_and_
           - eltwise.binary.logical_and.logical_and
           - eltwise.binary.logical_and.logical_and_output
+          - eltwise.binary.logical_and.logical_and_forge
           - eltwise.binary.polyval.polyval
           - eltwise.binary.remainder.remainder
+          - eltwise.binary.remainder.remainder_scalar_pytorch2
+          - eltwise.binary.remainder.remainder_forge
           - eltwise.binary.squared_difference.squared_difference
           - eltwise.binary.squared_difference_output.squared_difference_output
-          - eltwise.binary.remainder.remainder_scalar_pytorch2
           - eltwise.binary.bcast.bcast_h_sharded
           - eltwise.binary.bcast.bcast
           - eltwise.binary.eq.eq_scalar_pytorch2
+          - eltwise.binary.eq.eq_forge
+          - eltwise.binary.ge.ge_forge
           - eltwise.binary.gt.gt_scalar_pytorch2
+          - eltwise.binary.gt.gt_forge
           - eltwise.binary.le.le_tensor_pytorch2
           - eltwise.binary.fmod.fmod
           - eltwise.binary.floor_divide.floor_divide_pytorch2
@@ -260,7 +267,9 @@ on:
           - eltwise.binary.ldexp.ldexp
           - eltwise.binary.lt.lt_tensor_pytorch2
           - eltwise.binary.lt.lt_scalar_pytorch2
+          - eltwise.binary.lt.lt_forge
           - eltwise.binary.ne.ne_scalar_pytorch2
+          - eltwise.binary.ne.ne_forge
           - eltwise.binary.hypot.hypot
           - eltwise.binary.xlogy.xlogy
           - eltwise.binary_backward.ldexp_bw.ldexp_bw
@@ -293,6 +302,7 @@ on:
           - eltwise.ternary.lerp.lerp
           - eltwise.ternary.where.where
           - eltwise.ternary.where.where_pytorch2
+          - eltwise.ternary.where.where_forge
           - eltwise.ternary_backward.addcmul_bw
           - eltwise.ternary_backward.addcdiv_bw
           - embedding.embedding

tests/sweep_framework/sweeps/eltwise/binary/eq/eq_forge.py

@@ -0,0 +1,104 @@
+# SPDX-FileCopyrightText: © 2024 Tenstorrent Inc.
+
+# SPDX-License-Identifier: Apache-2.0
+
+from typing import Optional, Tuple
+from functools import partial
+
+import torch
+import ttnn
+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
+
+
+# Parameters provided to the test vector generator are defined here.
+# They are defined as dict-type suites that contain the arguments to the run function as keys, and lists of possible inputs as values.
+# Each suite has a key name (in this case "suite_1" and "suite_2") which will associate the test vectors to this specific suite of inputs.
+# Developers can create their own generator functions and pass them to the parameters as inputs.
+
+
+parameters = {
+    "xfail": {
+        "input_shape": [
+            {"self": [1, 1, 5, 5], "other": [1], "input_dtype": "ttnn.bfloat16"},
+            {"self": [1, 1, 7, 7], "other": [1], "input_dtype": "ttnn.bfloat16"},
+            {"self": [1, 45], "other": [1, 45], "input_dtype": "ttnn.int32"},
+            {"self": [1, 5], "other": [1, 5], "input_dtype": "ttnn.int32"},
+            {"self": [1, 6], "other": [1, 6], "input_dtype": "ttnn.int32"},
+            {"self": [1, 7], "other": [1, 7], "input_dtype": "ttnn.int32"},
+        ],
+        "input_a_layout": [ttnn.TILE_LAYOUT],
+        "input_b_layout": [ttnn.TILE_LAYOUT],
+        "input_a_memory_config": [ttnn.DRAM_MEMORY_CONFIG],
+        "input_b_memory_config": [ttnn.DRAM_MEMORY_CONFIG],
+        "output_memory_config": [ttnn.DRAM_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 or test_vector["input_b_layout"] == ttnn.ROW_MAJOR_LAYOUT:
+        return True, "Row Major layout is not supported"
+    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,
+    input_a_layout,
+    input_b_layout,
+    input_a_memory_config,
+    input_b_memory_config,
+    output_memory_config,
+    *,
+    device,
+) -> list:
+    torch.manual_seed(0)
+    if input_shape["input_dtype"] == "ttnn.bfloat16":
+        input_dtype = ttnn.bfloat16
+    elif input_shape["input_dtype"] == "ttnn.float32":
+        input_dtype = ttnn.float32
+    elif input_shape["input_dtype"] == "ttnn.int32":
+        input_dtype = ttnn.int32
+
+    torch_input_tensor_a = gen_func_with_cast_tt(
+        partial(torch_random, low=-100, high=100, dtype=torch.float32), input_dtype
+    )(input_shape["self"])
+    torch_input_tensor_b = gen_func_with_cast_tt(
+        partial(torch_random, low=-100, high=100, dtype=torch.float32), input_dtype
+    )(input_shape["other"])
+
+    golden_function = ttnn.get_golden_function(ttnn.eq)
+    torch_output_tensor = golden_function(torch_input_tensor_a, torch_input_tensor_b)
+
+    input_tensor_a = ttnn.from_torch(
+        torch_input_tensor_a,
+        dtype=input_dtype,
+        layout=input_a_layout,
+        device=device,
+        memory_config=input_a_memory_config,
+    )
+
+    input_tensor_b = ttnn.from_torch(
+        torch_input_tensor_b,
+        dtype=input_dtype,
+        layout=input_b_layout,
+        device=device,
+        memory_config=input_b_memory_config,
+    )
+
+    start_time = start_measuring_time()
+    result = ttnn.eq(input_tensor_a, input_tensor_b, memory_config=output_memory_config)
+    # ToDo: Update it once the tensor layout support with rank < 2 is supported in mid of Jan
+    output_tensor = ttnn.to_torch(result, torch_rank=len(input_shape["self"]))
+    e2e_perf = stop_measuring_time(start_time)
+
+    return [check_with_pcc(torch_output_tensor, output_tensor, 0.999), e2e_perf]

tests/sweep_framework/sweeps/eltwise/binary/ge/ge_forge.py

@@ -0,0 +1,103 @@
+# SPDX-FileCopyrightText: © 2024 Tenstorrent Inc.
+
+# SPDX-License-Identifier: Apache-2.0
+
+from typing import Optional, Tuple
+from functools import partial
+
+import torch
+import ttnn
+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
+
+
+# Parameters provided to the test vector generator are defined here.
+# They are defined as dict-type suites that contain the arguments to the run function as keys, and lists of possible inputs as values.
+# Each suite has a key name (in this case "suite_1" and "suite_2") which will associate the test vectors to this specific suite of inputs.
+# Developers can create their own generator functions and pass them to the parameters as inputs.
+
+
+parameters = {
+    "xfail": {
+        "input_shape": [[1]],
+        "input_a_dtype": [ttnn.int32],
+        "input_b_dtype": [ttnn.int32],
+        "input_a_layout": [ttnn.TILE_LAYOUT],
+        "input_b_layout": [ttnn.TILE_LAYOUT],
+        "input_a_memory_config": [ttnn.DRAM_MEMORY_CONFIG],
+        "input_b_memory_config": [ttnn.DRAM_MEMORY_CONFIG],
+        "output_memory_config": [ttnn.DRAM_MEMORY_CONFIG],
+    },
+}
+
+
+def mesh_device_fixture():
+    device = ttnn.open_device(device_id=0)
+    assert ttnn.device.is_wormhole_b0(device), "This op is available for Wormhole_B0 only"
+    yield (device, "Wormhole_B0")
+    ttnn.close_device(device)
+    del device
+
+
+# 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 or test_vector["input_b_layout"] == ttnn.ROW_MAJOR_LAYOUT:
+        return True, "Row Major layout is not supported"
+    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,
+    input_a_dtype,
+    input_b_dtype,
+    input_a_layout,
+    input_b_layout,
+    input_a_memory_config,
+    input_b_memory_config,
+    output_memory_config,
+    *,
+    device,
+) -> list:
+    torch.manual_seed(0)
+
+    torch_input_tensor_a = gen_func_with_cast_tt(
+        partial(torch_random, low=-100, high=100, dtype=torch.int32), input_a_dtype
+    )(input_shape)
+    torch_input_tensor_b = gen_func_with_cast_tt(
+        partial(torch_random, low=-100, high=100, dtype=torch.int32), input_b_dtype
+    )(input_shape)
+
+    golden_function = ttnn.get_golden_function(ttnn.ge)
+    torch_output_tensor = golden_function(torch_input_tensor_a, torch_input_tensor_b)
+
+    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,
+    )
+
+    input_tensor_b = ttnn.from_torch(
+        torch_input_tensor_b,
+        dtype=input_b_dtype,
+        layout=input_b_layout,
+        device=device,
+        memory_config=input_b_memory_config,
+    )
+
+    start_time = start_measuring_time()
+    result = ttnn.ge(input_tensor_a, input_tensor_b, memory_config=output_memory_config)
+    # ToDo: Update it once the tensor layout support with rank < 2 is supported in mid of Jan
+    output_tensor = ttnn.to_torch(result, torch_rank=len(input_shape))
+    e2e_perf = stop_measuring_time(start_time)
+
+    return [check_with_pcc(torch_output_tensor, output_tensor, 0.999), e2e_perf]

tests/sweep_framework/sweeps/eltwise/binary/gt/gt_forge.py

@@ -0,0 +1,102 @@
+# SPDX-FileCopyrightText: © 2024 Tenstorrent Inc.
+
+# SPDX-License-Identifier: Apache-2.0
+
+from typing import Optional, Tuple
+from functools import partial
+
+import torch
+import ttnn
+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
+
+
+# Parameters provided to the test vector generator are defined here.
+# They are defined as dict-type suites that contain the arguments to the run function as keys, and lists of possible inputs as values.
+# Each suite has a key name (in this case "suite_1" and "suite_2") which will associate the test vectors to this specific suite of inputs.
+# Developers can create their own generator functions and pass them to the parameters as inputs.
+
+
+parameters = {
+    "xfail": {
+        "input_shape": [[15, 15]],
+        "input_a_dtype": [ttnn.int32],
+        "input_b_dtype": [ttnn.int32],
+        "input_a_layout": [ttnn.TILE_LAYOUT],
+        "input_b_layout": [ttnn.TILE_LAYOUT],
+        "input_a_memory_config": [ttnn.DRAM_MEMORY_CONFIG],
+        "input_b_memory_config": [ttnn.DRAM_MEMORY_CONFIG],
+        "output_memory_config": [ttnn.DRAM_MEMORY_CONFIG],
+    },
+}
+
+
+def mesh_device_fixture():
+    device = ttnn.open_device(device_id=0)
+    assert ttnn.device.is_wormhole_b0(device), "This op is available for Wormhole_B0 only"
+    yield (device, "Wormhole_B0")
+    ttnn.close_device(device)
+    del device
+
+
+# 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 or test_vector["input_b_layout"] == ttnn.ROW_MAJOR_LAYOUT:
+        return True, "Row Major layout is not supported"
+    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,
+    input_a_dtype,
+    input_b_dtype,
+    input_a_layout,
+    input_b_layout,
+    input_a_memory_config,
+    input_b_memory_config,
+    output_memory_config,
+    *,
+    device,
+) -> list:
+    torch.manual_seed(0)
+
+    torch_input_tensor_a = gen_func_with_cast_tt(
+        partial(torch_random, low=-100, high=100, dtype=torch.int32), input_a_dtype
+    )(input_shape)
+    torch_input_tensor_b = gen_func_with_cast_tt(
+        partial(torch_random, low=-100, high=100, dtype=torch.int32), input_b_dtype
+    )(input_shape)
+
+    golden_function = ttnn.get_golden_function(ttnn.gt)
+    torch_output_tensor = golden_function(torch_input_tensor_a, torch_input_tensor_b)
+
+    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,
+    )
+
+    input_tensor_b = ttnn.from_torch(
+        torch_input_tensor_b,
+        dtype=input_b_dtype,
+        layout=input_b_layout,
+        device=device,
+        memory_config=input_b_memory_config,
+    )
+
+    start_time = start_measuring_time()
+    result = ttnn.gt(input_tensor_a, input_tensor_b, memory_config=output_memory_config)
+    output_tensor = ttnn.to_torch(result)
+    e2e_perf = stop_measuring_time(start_time)
+
+    return [check_with_pcc(torch_output_tensor, output_tensor, 0.999), e2e_perf]