#13373: PyTorch Tracing Sweeps - Eltwise set 2 (#13421)

* #13373: PyTorch Tracing Sweeps for set 2

* #13373: Update golden function

* #13421: Update

* #13421: Update where sweep

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

@@ -28,6 +28,8 @@ on:
           - eltwise.unary.sin.sin_pytorch2
           - eltwise.unary.tril.tril_pytorch2
           - eltwise.unary.clamp.clamp
+          - eltwise.unary.clamp.clamp_pytorch2
+          - eltwise.unary.clamp.clamp_min_pytorch2
           - eltwise.unary.clip.clip
           - eltwise.unary.cbrt.cbrt
           - eltwise.unary.rsub.rsub
@@ -70,6 +72,8 @@ on:
           - eltwise.unary.logical_not.logical_not_
           - eltwise.unary.logical_not.logical_not
           - eltwise.unary.logical_not.logical_not_output
+          - eltwise.unary.logical_not.logical_not_pytorch2
+          - eltwise.unary.neg.neg_pytorch2
           - eltwise.unary.erf.erf
           - eltwise.unary.erfinv.erfinv
           - eltwise.unary.i0.i0
@@ -148,10 +152,15 @@ on:
           - eltwise.unary.lez.lez
           - eltwise.unary.nez.nez
           - eltwise.unary.prelu.prelu
+          - eltwise.unary.hardswish.hardswish_pytorch2
+          - eltwise.unary.hardtanh.hardtanh_pytorch2
           - eltwise.binary.subtract.subtract
+          - eltwise.binary.subtract.subtract_tensor_pytorch2
           - eltwise.binary.multiply.multiply
           - eltwise.binary.multiply.mul_tensor_pytorch2
+          - eltwise.binary.multiply.multiply_scalar_pytorch2
           - eltwise.binary.div.div
+          - eltwise.binary.div.div_tensor_pytorch2
           - eltwise.binary.div_no_nan.div_no_nan
           - eltwise.binary.logical_or.logical_or_
           - eltwise.binary.logical_or.logical_or
@@ -175,6 +184,9 @@ on:
           - eltwise.binary.floor_divide.floor_divide_pytorch2
           - eltwise.binary.logaddexp.logaddexp
           - eltwise.binary.ldexp.ldexp
+          - eltwise.binary.lt.lt_tensor_pytorch2
+          - eltwise.binary.lt.lt_scalar_pytorch2
+          - eltwise.binary.ne.ne_scalar_pytorch2
           - eltwise.binary.hypot.hypot
           - eltwise.binary.xlogy.xlogy
           - eltwise.binary_backward.embedding_bw.embedding_bw

tests/sweep_framework/sweeps/eltwise/binary/div/div_tensor_pytorch2.py

@@ -0,0 +1,193 @@
+# 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 = {
+    "nightly": {
+        "input_specs": [
+            {"shape": [0, 1], "other": 1.0},
+            {"shape": [1, 1, 16384, 256], "other": 5.656854249492381},
+            {"shape": [1, 1, 19200, 300], "other": 8.0},
+            {"shape": [1, 1, 256], "other": 0.5},
+            {"shape": [1, 1024, 640], "other": 1.0},
+            {"shape": [1, 12, 10, 10], "other": 8.0},
+            {"shape": [1, 12, 12, 12], "other": 8.0},
+            {"shape": [1, 12, 14, 14], "other": 8.0},
+            {"shape": [1, 12, 16, 64], "other": 8.0},
+            {"shape": [1, 12, 197, 197], "other": 8.0},
+            {"shape": [1, 12, 201, 201], "other": 8.0},
+            {"shape": [1, 12, 25, 25], "other": 8.0},
+            {"shape": [1, 12, 7, 7], "other": []},
+            {"shape": [1, 12, 9, 9], "other": 8.0},
+            {"shape": [1, 128, 1536], "other": 3},
+            {"shape": [1, 1280, 16, 16], "other": 1.0},
+            {"shape": [1, 1280, 8, 8], "other": 1},
+            {"shape": [1, 1280, 8, 8], "other": 1.0},
+            {"shape": [1, 16, 1, 6], "other": []},
+            # {"shape": [1, 16, 1, s10 + 1], "other": []},
+            {"shape": [1, 16, 197, 197], "other": 8.0},
+            {"shape": [1, 16, 256, 256], "other": 8.0},
+            {"shape": [1, 16, 5, 5], "other": []},
+            {"shape": [1, 16, 9, 9], "other": 11.313708498984761},
+            {"shape": [1, 16, 9, 9], "other": 8.0},
+            {"shape": [1, 1], "other": 16},
+            {"shape": [1, 1], "other": 2.0794415416798357},
+            {"shape": [1, 2, 4096, 256], "other": 5.656854249492381},
+            {"shape": [1, 2, 4800, 300], "other": 8.0},
+            {"shape": [1, 23, 40, 1], "other": [128]},
+            {"shape": [1, 23, 40], "other": [1, 1, 40]},
+            {"shape": [1, 23, 40], "other": [1, 23, 1]},
+            {"shape": [1, 24, 64, 32], "other": [1, 24, 64, 32]},
+            {"shape": [1, 256, 1280], "other": 1.0},
+            {"shape": [1, 256, 384], "other": 3},
+            {"shape": [1, 3, 1445, 1445], "other": 8.0},
+            {"shape": [1, 32, 24576], "other": 3},
+            {"shape": [1, 32, 64, 32], "other": [1, 32, 64, 32]},
+            {"shape": [1, 320, 64, 64], "other": 1.0},
+            {"shape": [1, 4096, 320], "other": 1.0},
+            {"shape": [1, 5, 1024, 256], "other": 5.656854249492381},
+            {"shape": [1, 5, 1200, 300], "other": 8.0},
+            {"shape": [1, 50257], "other": 0.9},
+            {"shape": [1, 512, 38, 38], "other": [1, 512, 38, 38]},
+            {"shape": [1, 512], "other": [1, 1]},
+            {"shape": [1, 512], "other": [1, 512]},
+            {"shape": [1, 64, 1280], "other": 1.0},
+            {"shape": [1, 64, 6144], "other": 3},
+            {"shape": [1, 64, 9, 9], "other": 8.0},
+            {"shape": [1, 640, 32, 32], "other": 1.0},
+            {"shape": [1, 8, 2048, 256], "other": 5.656854249492381},
+            {"shape": [1, 8, 256, 2048], "other": 5.656854249492381},
+            {"shape": [1, 8, 256, 256], "other": 5.656854249492381},
+            {"shape": [1, 8, 300, 300], "other": 8.0},
+            # {"shape": [1, s0, 256], "other": 0.5 },
+            {"shape": [10, 10], "other": 2.772588722239781},
+            {"shape": [10, 10], "other": 8},
+            {"shape": [10], "other": 10},
+            {"shape": [10], "other": 9.375},
+            {"shape": [128], "other": 128},
+            {"shape": [15, 15], "other": 2.772588722239781},
+            {"shape": [15, 15], "other": 8},
+            {"shape": [16, 6, 64, 32], "other": [16, 6, 64, 32]},
+            {"shape": [16, 8, 64, 32], "other": [16, 8, 64, 32]},
+            {"shape": [160], "other": 160},
+            {"shape": [17, 17], "other": 16},
+            {"shape": [17, 17], "other": 2.0794415416798357},
+            {"shape": [19], "other": 18.75},
+            {"shape": [1], "other": 1},
+            {"shape": [1], "other": 1.0},
+            {"shape": [2, 2], "other": 16},
+            {"shape": [2, 2], "other": 2.0794415416798357},
+            {"shape": [2, 512], "other": [2, 1]},
+            {"shape": [20], "other": 20},
+            {"shape": [2], "other": 2},
+            {"shape": [3, 320, 320], "other": [3, 1, 1]},
+            {"shape": [3, 480, 640], "other": [3, 1, 1]},
+            {"shape": [3234, 1], "other": 10.0},
+            {"shape": [3234, 1], "other": 5.0},
+            {"shape": [38], "other": 37.5},
+            {"shape": [3], "other": 3},
+            {"shape": [3], "other": 3.0},
+            {"shape": [4, 12, 64, 32], "other": [4, 12, 64, 32]},
+            {"shape": [4, 16, 64, 32], "other": [4, 16, 64, 32]},
+            {"shape": [5], "other": 4.6875},
+            {"shape": [5], "other": 5},
+            {"shape": [64, 3, 64, 32], "other": [64, 3, 64, 32]},
+            {"shape": [64, 4, 64, 32], "other": [64, 4, 64, 32]},
+            {"shape": [8, 100, 32], "other": 5.656854249492381},
+            {"shape": [8, 920, 32], "other": 5.656854249492381},
+            {"shape": [8732, 1], "other": 10.0},
+            {"shape": [8732, 1], "other": 5.0},
+            {"shape": [96, 80], "other": [80]},
+            {"shape": [], "other": []},
+            # {"shape": [s0 + 1, s0 + 1], "other": 16 },
+            # {"shape": [s0 + 1, s0 + 1], "other": 2.0794415416798357  },
+        ],
+        "input_a_dtype": [ttnn.bfloat16],
+        "input_b_dtype": [ttnn.bfloat16],
+        "input_a_layout": [ttnn.TILE_LAYOUT],
+        "input_b_layout": [ttnn.TILE_LAYOUT],
+        "input_a_memory_config": [ttnn.DRAM_MEMORY_CONFIG, ttnn.L1_MEMORY_CONFIG],
+        "input_b_memory_config": [ttnn.DRAM_MEMORY_CONFIG, ttnn.L1_MEMORY_CONFIG],
+        "output_memory_config": [ttnn.DRAM_MEMORY_CONFIG, ttnn.L1_MEMORY_CONFIG],
+    },
+}
+
+
+def run(
+    input_specs,
+    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:
+    data_seed = random.randint(0, 20000000)
+    torch.manual_seed(data_seed)
+
+    input_shape = input_specs["shape"]
+    if len(input_shape) == 0:
+        torch_input_tensor_a = torch.empty([])
+    else:
+        torch_input_tensor_a = gen_func_with_cast_tt(
+            partial(torch_random, low=-100, high=100, dtype=torch.float32), input_a_dtype
+        )(input_shape)
+
+    other = input_specs["other"]
+    if isinstance(other, (int, float)):
+        torch_other_tensor = torch.tensor(other, dtype=torch.float32)
+    elif len(other) == 0:
+        torch_other_tensor = torch.empty([])
+    else:
+        torch_other_tensor = gen_func_with_cast_tt(
+            partial(torch_random, low=-100, high=100, dtype=torch.float32), input_b_dtype
+        )(other)
+
+    golden_function = ttnn.get_golden_function(ttnn.divide)
+    torch_output_tensor = golden_function(torch_input_tensor_a, torch_other_tensor)
+
+    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_other_tensor,
+        dtype=input_b_dtype,
+        layout=input_b_layout,
+        device=device,
+        memory_config=input_b_memory_config,
+    )
+
+    start_time = start_measuring_time()
+
+    output_tensor = ttnn.divide(input_tensor_a, input_tensor_b, memory_config=output_memory_config)
+    output_tensor = ttnn.to_torch(output_tensor)
+
+    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/lt/lt_scalar_pytorch2.py

@@ -0,0 +1,73 @@
+# 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 = {
+    "nightly": {
+        "input_specs": [
+            {"shape": [1, 1], "other": 16},
+            {"shape": [10, 10], "other": 8},
+            {"shape": [15, 15], "other": 8},
+            {"shape": [17, 17], "other": 16},
+            {"shape": [2, 2], "other": 16},
+            # {"shape": [s0 + 1, s0 + 1], "other": 16},
+        ],
+        "input_a_dtype": [ttnn.bfloat16],
+        "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 run(
+    input_specs,
+    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_specs["shape"])
+
+    golden_function = ttnn.get_golden_function(ttnn.lt)
+    torch_output_tensor = golden_function(torch_input_tensor_a, input_specs["other"])
+
+    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()
+    output_tensor = ttnn.lt(input_tensor_a, input_specs["other"], memory_config=output_memory_config)
+    output_tensor = ttnn.to_torch(output_tensor)
+    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/lt/lt_tensor_pytorch2.py

@@ -0,0 +1,80 @@
+# 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 assert_equal, 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 = {
+    "nightly": {
+        "input_shape_a": [[1, 50257]],
+        "input_shape_b": [[1, 1]],
+        "input_a_dtype": [ttnn.bfloat16],
+        "input_b_dtype": [ttnn.bfloat16],
+        "input_a_layout": [ttnn.TILE_LAYOUT],
+        "input_b_layout": [ttnn.TILE_LAYOUT],
+        "input_a_memory_config": [ttnn.DRAM_MEMORY_CONFIG, ttnn.L1_MEMORY_CONFIG],
+        "input_b_memory_config": [ttnn.DRAM_MEMORY_CONFIG, ttnn.L1_MEMORY_CONFIG],
+    },
+}
+
+
+def run(
+    input_shape_a,
+    input_shape_b,
+    input_a_dtype,
+    input_b_dtype,
+    input_a_layout,
+    input_b_layout,
+    input_a_memory_config,
+    input_b_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_a)
+    torch_input_tensor_b = gen_func_with_cast_tt(
+        partial(torch_random, low=-100, high=100, dtype=torch.float32), input_b_dtype
+    )(input_shape_b)
+
+    golden_function = ttnn.get_golden_function(ttnn.lt)
+    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.lt(input_tensor_a, input_tensor_b)
+    output_tensor = ttnn.to_torch(result)
+    e2e_perf = stop_measuring_time(start_time)
+
+    return [assert_equal(torch_output_tensor, output_tensor), e2e_perf]