#13373: PyTorch Tracing Sweeps for set 2 ops

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

@@ -18,6 +18,7 @@ on:
           - eltwise.unary.cos.cos
           - eltwise.unary.sin.sin
           - eltwise.unary.clamp.clamp
+          - eltwise.unary.clamp.clamp_pytorch2
           - eltwise.unary.clip.clip
           - eltwise.unary.cbrt.cbrt
           - eltwise.unary.rsub.rsub
@@ -47,6 +48,8 @@ on:
           - eltwise.unary.bitwise.bitwise_xor
           - eltwise.unary.log_sigmoid.log_sigmoid
           - eltwise.unary.logical_not_.logical_not_
+          - eltwise.unary.logical_not.logical_not_pytorch2
+          - eltwise.unary.neg.neg_pytorch2
           - eltwise.unary.erf.erf
           - eltwise.unary.erfinv.erfinv
           - eltwise.unary.i0.i0
@@ -55,6 +58,8 @@ on:
           - eltwise.unary.lgamma.lgamma
           - eltwise.unary.sigmoid.sigmoid
           - eltwise.unary.sigmoid_accurate.sigmoid_accurate
+          - eltwise.unary.hardswish.hardswish_pytorch2
+          - eltwise.unary.hardtanh.hardtanh_pytorch2
           - eltwise.binary.subtract.subtract
           - eltwise.binary.multiply.multiply
           - eltwise.binary.div.div

tests/sweep_framework/sweeps/eltwise/unary/clamp/clamp_pytorch2.py

@@ -0,0 +1,132 @@
+# 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_low_high_scalars
+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 = {
+    "clamp_1": {
+        "input_specs": [
+            {"shape": [0, 1], "max": 4.135166556742356},
+            {"shape": [0, 2], "min": 0, "max": 1066},
+            {"shape": [0, 2], "min": 0, "max": 800},
+            {"shape": [1, 1, 1, 42], "min": 0, "max": 82},
+            {"shape": [1, 1, 32, 1], "min": 0, "max": 49},
+            {"shape": [1066], "min": 0.0},
+            {"shape": [1066], "max": 639},
+            {"shape": [12, 1, 1], "max": 4.605170185988092},
+            {"shape": [120], "min": 0.0},
+            {"shape": [120], "max": 59},
+            {"shape": [128], "min": 0.0},
+            {"shape": [128], "max": 127},
+            {"shape": [128], "max": 15},
+            {"shape": [128], "max": 31},
+            {"shape": [128], "max": 63},
+            {"shape": [16, 1, 1], "max": 4.605170185988092},
+            {"shape": [160], "min": 0.0},
+            {"shape": [160], "max": 79},
+            {"shape": [24, 1, 1], "max": 4.605170185988092},
+            {"shape": [240], "min": 0.0},
+            {"shape": [240], "max": 119},
+            {"shape": [3, 1, 1], "max": 4.605170185988092},
+            {"shape": [300], "min": 0.0},
+            {"shape": [300], "max": 479},
+            {"shape": [300], "max": 639},
+            {"shape": [30], "min": 0.0},
+            {"shape": [30], "max": 14},
+            {"shape": [32, 1, 1], "max": 4.605170185988092},
+            {"shape": [320], "min": 0.0},
+            {"shape": [320], "max": 159},
+            {"shape": [320], "max": 319},
+            {"shape": [320], "max": 479},
+            {"shape": [320], "max": 639},
+            {"shape": [3234, 1], "max": 4.135166556742356},
+            {"shape": [3234, 2], "min": 0, "max": 320},
+            {"shape": [4, 1, 1], "max": 4.605170185988092},
+            {"shape": [4, 2], "min": 0, "max": 1},
+            {"shape": [40], "min": 0.0},
+            {"shape": [40], "max": 19},
+            {"shape": [480], "min": 0.0},
+            {"shape": [480], "max": 239},
+            {"shape": [6, 1, 1], "max": 4.605170185988092},
+            {"shape": [6, 2], "min": 0, "max": 1},
+            {"shape": [60], "min": 0.0},
+            {"shape": [60], "max": 29},
+            {"shape": [640], "min": 0.0},
+            {"shape": [640], "max": 319},
+            {"shape": [8, 1, 1], "max": 4.605170185988092},
+            {"shape": [800], "min": 0.0},
+            {"shape": [800], "max": 479},
+            {"shape": [80], "min": 0.0},
+            {"shape": [80], "max": 39},
+            {"shape": [8732, 1], "max": 4.135166556742356},
+            {"shape": [8732, 2], "min": 0, "max": 300},
+            {"shape": [1, 1, 38, 38], "min": 1e-12},
+            {"shape": [1, 1], "min": 1e-12},
+            {"shape": [1, 24, 64, 1], "min": 1e-12},
+            {"shape": [1, 32, 64, 1], "min": 1e-12},
+            {"shape": [16, 6, 64, 1], "min": 1e-12},
+            {"shape": [16, 8, 64, 1], "min": 1e-12},
+            {"shape": [4, 12, 64, 1], "min": 1e-12},
+            {"shape": [4, 16, 64, 1], "min": 1e-12},
+            {"shape": [64, 3, 64, 1], "min": 1e-12},
+            {"shape": [64, 4, 64, 1], "min": 1e-12},
+        ],
+        "input_a_dtype": [ttnn.bfloat16, ttnn.bfloat8_b],
+        "input_a_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],
+    },
+}
+
+
+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"])
+
+    min_val = input_specs.get("min", None)
+    max_val = input_specs.get("max", None)
+
+    torch_output_tensor = torch.clamp(torch_input_tensor_a, min_val, max_val)
+
+    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.clamp(input_tensor_a, min_val, max_val, 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/hardswish/hardswish_pytorch2.py

@@ -0,0 +1,92 @@
+# 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_low_high_scalars
+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 = {
+    "hardswish_1": {
+        "input_shape": [
+            [1, 1024],
+            [1, 120, 14, 14],
+            [1, 1280],
+            [1, 144, 14, 14],
+            [1, 16, 112, 112],
+            [1, 16, 160, 160],
+            [1, 184, 14, 14],
+            [1, 184, 20, 20],
+            [1, 200, 14, 14],
+            [1, 200, 20, 20],
+            [1, 240, 14, 14],
+            [1, 240, 20, 20],
+            [1, 240, 28, 28],
+            [1, 240, 40, 40],
+            [1, 288, 14, 14],
+            [1, 288, 7, 7],
+            [1, 480, 10, 10],
+            [1, 480, 14, 14],
+            [1, 480, 20, 20],
+            [1, 576, 7, 7],
+            [1, 672, 10, 10],
+            [1, 672, 14, 14],
+            [1, 672, 20, 20],
+            [1, 672, 7, 7],
+            [1, 96, 14, 14],
+            [1, 96, 28, 28],
+            [1, 960, 7, 7],
+        ],
+        "input_dtype": [ttnn.bfloat16, ttnn.bfloat8_b],
+        "input_layout": [ttnn.TILE_LAYOUT, ttnn.ROW_MAJOR_LAYOUT],
+        "input_memory_config": [ttnn.DRAM_MEMORY_CONFIG, ttnn.L1_MEMORY_CONFIG],
+        "output_memory_config": [ttnn.DRAM_MEMORY_CONFIG, ttnn.L1_MEMORY_CONFIG],
+    },
+}
+
+
+def run(
+    input_shape,
+    input_dtype,
+    input_layout,
+    input_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_dtype
+    )(input_shape)
+
+    torch_output_tensor = torch.nn.functional.hardswish(torch_input_tensor_a)
+
+    input_tensor_a = ttnn.from_torch(
+        torch_input_tensor_a,
+        dtype=input_dtype,
+        layout=input_layout,
+        device=device,
+        memory_config=input_memory_config,
+    )
+
+    start_time = start_measuring_time()
+    result = ttnn.hardswish(input_tensor_a, 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]