#11512: Modify existing sharded sweeps to use new sharding funciton

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

@@ -0,0 +1,87 @@
+# 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, sanitize_shape_rm
+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, 1, 1], [6, 12, 128, 128], [1, 1, 1, 1], 32)
+        + gen_shapes([1, 1, 1], [12, 256, 256], [1, 1, 1], 32)
+        + gen_shapes([1, 1], [256, 256], [1, 1], 32),
+        "input_a_dtype": [ttnn.int32],
+        "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],
+    },
+}
+
+
+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_layout,
+    input_a_memory_config,
+    output_memory_config,
+    *,
+    device,
+) -> list:
+    data_seed = random.randint(0, 20000000)
+    torch.manual_seed(data_seed)
+
+    if input_layout == ttnn.ROW_MAJOR_LAYOUT:
+        input_shape = sanitize_shape_rm(input_shape)
+
+    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_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_not/bitwise_not.py

@@ -0,0 +1,96 @@
+# 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, sanitize_shape_rm
+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, 1, 1], [6, 12, 512, 512], [1, 1, 1, 1], 16)
+        + gen_shapes([1, 1, 1], [12, 1024, 1024], [1, 1, 1], 16)
+        + gen_shapes([1, 1], [1024, 1024], [1, 1], 16),
+        "input_a_dtype": [ttnn.int32],
+        "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],
+    },
+}
+
+
+# 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:
+        return True, "Row Major layout is not supported"
+    return False, None
+
+
+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_layout,
+    input_a_memory_config,
+    output_memory_config,
+    *,
+    device,
+) -> list:
+    data_seed = random.randint(0, 20000000)
+    torch.manual_seed(data_seed)
+
+    if input_layout == ttnn.ROW_MAJOR_LAYOUT:
+        input_shape = sanitize_shape_rm(input_shape)
+
+    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)
+
+    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_layout,
+        device=device,
+        memory_config=input_a_memory_config,
+    )
+
+    start_time = start_measuring_time()
+    result = ttnn.bitwise_not(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]

tests/sweep_framework/sweeps/eltwise/unary/hardshrink/hardshrink_sharded.py

@@ -4,30 +4,28 @@
 
 from typing import Optional, Tuple
 from functools import partial
-import itertools
 
+import json
 import torch
 import random
 import ttnn
 import math
-from tests.sweep_framework.sweep_utils.utils import (
-    gen_shapes,
-    sanitize_shape_rm,
-    get_device_grid_size,
-)
+from tests.sweep_framework.sweep_utils.utils import gen_shapes, sanitize_shape_rm
 from tests.sweep_framework.sweep_utils.sharding_utils import (
-    gen_unary_sharded_spec,
+    gen_sharded_spec_unary,
     parse_sharding_spec,
+    invalidate_vector_sharding,
+    roundup,
+    divup,
 )
 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 = 120
 
-Y, X = get_device_grid_size()
-
 random.seed(0)
 
 
@@ -36,43 +34,42 @@
 # 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_spec": list(gen_unary_sharded_spec(8, 4, "ROW_MAJOR", "BLOCK", True))
-        + list(gen_unary_sharded_spec(8, 4, "COL_MAJOR", "BLOCK", True))
-        + list(gen_unary_sharded_spec(8, 4, "ROW_MAJOR", "HEIGHT", True))
-        + list(gen_unary_sharded_spec(8, 4, "COL_MAJOR", "HEIGHT", True))
-        + list(gen_unary_sharded_spec(8, 4, "ROW_MAJOR", "WIDTH", True))
-        + list(gen_unary_sharded_spec(8, 4, "COL_MAJOR", "WIDTH", True))
-        + list(gen_unary_sharded_spec(8, 4, "ROW_MAJOR", "TENSOR_HW", True))
-        + list(gen_unary_sharded_spec(8, 4, "COL_MAJOR", "TENSOR_HW", True))
-        + list(gen_unary_sharded_spec(8, 4, "ROW_MAJOR", "BLOCK", False))
-        + list(gen_unary_sharded_spec(8, 4, "COL_MAJOR", "BLOCK", False))
-        + list(gen_unary_sharded_spec(8, 4, "ROW_MAJOR", "HEIGHT", False))
-        + list(gen_unary_sharded_spec(8, 4, "COL_MAJOR", "HEIGHT", False))
-        + list(gen_unary_sharded_spec(8, 4, "ROW_MAJOR", "WIDTH", False))
-        + list(gen_unary_sharded_spec(8, 4, "COL_MAJOR", "WIDTH", False))
-        + list(gen_unary_sharded_spec(8, 4, "ROW_MAJOR", "TENSOR_HW", False))
-        + list(gen_unary_sharded_spec(8, 4, "COL_MAJOR", "TENSOR_HW", False)),
+    "nightly": {
+        "input_spec": gen_sharded_spec_unary(16, 4),
         "input_a_dtype": [ttnn.bfloat16, ttnn.bfloat8_b],
-        "input_layout": [ttnn.TILE_LAYOUT, ttnn.ROW_MAJOR_LAYOUT],
     },
 }
 
 
+# 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 or test_vector["input_a_dtype"] == ttnn.bfloat8_b:
+    input_spec = parse_sharding_spec(test_vector["input_spec"])
+    (
+        input_shape,
+        core_grid_size,
+        shard_orientation,
+        sharding_strategy,
+        tensor_hw_as_shard_shape,
+        shard_height_mul_of_32,
+        input_layout,
+    ) = input_spec
+    invlidated, output_str = invalidate_vector_sharding(*input_spec)
+    if invlidated:
+        return invlidated, output_str
+    if input_layout == ttnn.ROW_MAJOR_LAYOUT or test_vector["input_a_dtype"] == ttnn.bfloat8_b:
         return True, "Row Major layout and bfloat8_b are not supported"
     return False, None
 
 
-# This is the run instructions forgit a the test, defined by the developer.
+# 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_spec,
     input_a_dtype,
-    input_layout,
     *,
     device,
 ) -> list:
@@ -86,6 +83,7 @@ def run(
         sharding_strategy,
         tensor_hw_as_shard_shape,
         shard_height_mul_of_32,
+        input_layout,
     ) = parse_sharding_spec(input_spec)
     y, x = core_grid_size
     device_grid_size = ttnn.CoreGrid(y=y, x=x)