Add bitwise ops sweeps, add gen_rand_bitwise_left_shift function (#13366

)

* #11512: Add bitwise ops, add gen_rand_bitwise_left_shift function in sweep_framework/utils.py

* #11512: Minor fix

* #11512: Change bitwise ops names

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

@@ -39,8 +39,14 @@ on:
           - eltwise.unary.log1p.log1p
           - eltwise.unary.log2.log2
           - eltwise.unary.log10.log10
+          - eltwise.unary.bitwise.bitwise_and
+          - eltwise.unary.bitwise.bitwise_left_shift
+          - eltwise.unary.bitwise.bitwise_not
+          - eltwise.unary.bitwise.bitwise_or
+          - eltwise.unary.bitwise.bitwise_right_shift
+          - eltwise.unary.bitwise.bitwise_xor
           - eltwise.binary.subtract.subtract
-          - eltwise.binary.multiply.multipl
+          - eltwise.binary.multiply.multiply
           - eltwise.binary.div.div
           - eltwise.binary.div_no_nan.div_no_nan
           - eltwise.binary.logical_or_.logical_or_

tests/sweep_framework/sweeps/eltwise/unary/bitwise/bitwise_and.py

@@ -0,0 +1,84 @@
+# 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.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
+
+random.seed(0)
+
+# 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") 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 = {
+    "nightly": {
+        "input_shape": gen_shapes([1, 1, 32, 32], [6, 12, 128, 128], [1, 1, 32, 32], 32)
+        + gen_shapes([1, 32, 32], [12, 256, 256], [1, 32, 32], 32)
+        + gen_shapes([32, 32], [256, 256], [32, 32], 32),
+        "input_a_dtype": [ttnn.int32],
+        "input_a_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],
+    },
+}
+
+
+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
+
+
+# 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 device_mesh_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_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=0, high=2147483648, dtype=torch.int64), input_a_dtype
+    )(input_shape)
+
+    scalar = torch.randint(0, 2147483648, (1,), dtype=torch.int32).item()
+
+    torch_output_tensor = torch.bitwise_and(torch_input_tensor_a, scalar)
+
+    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.bitwise_and(input_tensor_a, value=scalar, 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]

tests/sweep_framework/sweeps/eltwise/unary/bitwise/bitwise_left_shift.py

@@ -0,0 +1,105 @@
+# 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.utils import gen_shapes, gen_rand_bitwise_left_shift
+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
+
+random.seed(0)
+
+# 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") 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 = {
+    "nightly": {
+        "input_shape": gen_shapes([1, 1, 32, 32], [6, 12, 128, 128], [1, 1, 32, 32], 4)
+        + gen_shapes([1, 32, 32], [12, 256, 256], [1, 32, 32], 4)
+        + gen_shapes([32, 32], [256, 256], [32, 32], 4),
+        "shift_bits": list(range(1, 31)),
+        "use_safe_nums": [True],
+        "input_a_dtype": [ttnn.int32],
+        "input_a_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],
+    },
+    "xfail": {
+        "input_shape": gen_shapes([1, 1, 32, 32], [6, 12, 128, 128], [1, 1, 32, 32], 4)
+        + gen_shapes([1, 32, 32], [12, 256, 256], [1, 32, 32], 4)
+        + gen_shapes([32, 32], [256, 256], [32, 32], 4),
+        "shift_bits": list(range(1, 31)),
+        "use_safe_nums": [False],
+        "input_a_dtype": [ttnn.int32],
+        "input_a_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],
+    },
+}
+
+
+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
+
+
+# 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 device_mesh_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,
+    shift_bits,
+    use_safe_nums,
+    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)
+
+    # In ttnn.bitwise_left_shift, only bits from positions 0 to 30 are included during shifting (sign bit remains the same).
+    # That is not the case with torch.bitwise_left_shift, all of the bits are included during shifting.
+    # use_safe_nums argument makes sure that those two bits are the same, so the results between the two versions are the same.
+    if use_safe_nums is True:
+        torch_input_tensor_a = gen_func_with_cast_tt(
+            partial(gen_rand_bitwise_left_shift, shift_bits=shift_bits, low=-2147483647, high=2147483648), input_a_dtype
+        )(input_shape)
+    else:
+        torch_input_tensor_a = gen_func_with_cast_tt(
+            partial(torch_random, low=-2147483647, high=2147483648, dtype=torch.int64), input_a_dtype
+        )(input_shape)
+
+    torch_output_tensor = torch.bitwise_left_shift(torch_input_tensor_a, shift_bits).to(torch.int32)
+
+    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.bitwise_left_shift(input_tensor_a, shift_bits=shift_bits, memory_config=output_memory_config)
+    output_tensor = ttnn.to_torch(result).to(torch.int32)
+    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/unary/bitwise/bitwise_not.py

@@ -0,0 +1,86 @@
+# 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.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
+
+random.seed(0)
+
+# 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") 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 = {
+    "nightly": {
+        "input_shape": gen_shapes([1, 1, 32, 32], [6, 12, 512, 512], [1, 1, 32, 32], 16)
+        + gen_shapes([1, 32, 32], [12, 1024, 1024], [1, 32, 32], 16)
+        + gen_shapes([32, 32], [1024, 1024], [32, 32], 16),
+        "input_a_dtype": [ttnn.int32],
+        "input_a_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],
+    },
+}
+
+
+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
+
+
+# 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 device_mesh_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_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=-2147483647, high=2147483648, dtype=torch.int64), input_a_dtype
+    )(input_shape)
+
+    torch_input_tensor_a = torch.full(size=input_shape, fill_value=-2147483647).to(torch.int32)
+
+    scalar = torch.randint(-100, 101, (1,)).item()
+
+    torch_output_tensor = torch.bitwise_not(torch_input_tensor_a)
+
+    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.bitwise_not(input_tensor_a, value=scalar, 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]

tests/sweep_framework/sweeps/eltwise/unary/bitwise/bitwise_or.py

@@ -0,0 +1,84 @@
+# 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.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
+
+random.seed(0)
+
+# 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") 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 = {
+    "nightly": {
+        "input_shape": gen_shapes([1, 1, 32, 32], [6, 12, 128, 128], [1, 1, 32, 32], 8)
+        + gen_shapes([1, 32, 32], [12, 256, 256], [1, 32, 32], 8)
+        + gen_shapes([32, 32], [256, 256], [32, 32], 8),
+        "input_a_dtype": [ttnn.int32],
+        "input_a_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],
+    },
+}
+
+
+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
+
+
+# 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 device_mesh_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_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=0, high=2147483648, dtype=torch.int64), input_a_dtype
+    )(input_shape)
+
+    scalar = torch.randint(0, 2147483648, (1,), dtype=torch.int32).item()
+
+    torch_output_tensor = torch.bitwise_or(torch_input_tensor_a, scalar)
+
+    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.bitwise_or(input_tensor_a, value=scalar, 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]