import gc
import sys
import torch
from torch.testing._internal.common_utils import NoTest, run_tests, TestCase
if not torch.accelerator.is_available():
print("No available accelerator detected, skipping tests", file=sys.stderr)
TestCase = NoTest
sys.exit()
class TestAccelerator(TestCase):
def test_current_accelerator(self):
self.assertTrue(torch.accelerator.is_available())
accelerators = ["cuda", "xpu", "mps", "npu"]
for accelerator in accelerators:
if torch.get_device_module(accelerator).is_available():
self.assertEqual(
torch.accelerator.current_accelerator().type, accelerator
)
self.assertIsNone(torch.accelerator.current_accelerator().index)
with self.assertRaisesRegex(
ValueError, "doesn't match the current accelerator"
):
torch.accelerator.set_device_index("cpu")
def test_generic_stream_behavior(self):
s1 = torch.Stream()
s2 = torch.Stream()
torch.accelerator.set_stream(s1)
self.assertEqual(torch.accelerator.current_stream(), s1)
event = torch.Event()
a = torch.randn(1000)
b = torch.randn(1000)
c = a + b
torch.accelerator.set_stream(s2)
self.assertEqual(torch.accelerator.current_stream(), s2)
a_acc = a.to(torch.accelerator.current_accelerator(), non_blocking=True)
b_acc = b.to(torch.accelerator.current_accelerator(), non_blocking=True)
torch.accelerator.set_stream(s1)
self.assertEqual(torch.accelerator.current_stream(), s1)
event.record(s2)
event.synchronize()
c_acc = a_acc + b_acc
event.record(s2)
torch.accelerator.synchronize()
self.assertTrue(event.query())
self.assertEqual(c_acc.cpu(), c)
def test_current_stream_query(self):
s = torch.accelerator.current_stream()
self.assertEqual(torch.accelerator.current_stream(s.device), s)
self.assertEqual(torch.accelerator.current_stream(s.device.index), s)
self.assertEqual(torch.accelerator.current_stream(str(s.device)), s)
other_device = torch.device("cpu")
with self.assertRaisesRegex(
ValueError, "doesn't match the current accelerator"
):
torch.accelerator.current_stream(other_device)
def test_stream_context_manager(self):
prev_stream = torch.accelerator.current_stream()
with torch.Stream() as s:
self.assertEqual(torch.accelerator.current_stream(), s)
self.assertEqual(torch.accelerator.current_stream(), prev_stream)
def test_pin_memory_on_non_blocking_copy(self):
t_acc = torch.randn(100).to(torch.accelerator.current_accelerator())
t_host = t_acc.to("cpu", non_blocking=True)
torch.accelerator.synchronize()
self.assertTrue(t_host.is_pinned())
self.assertEqual(t_acc.cpu(), t_host)
def test_generic_event_behavior(self):
event1 = torch.Event(enable_timing=False)
event2 = torch.Event(enable_timing=False)
with self.assertRaisesRegex(
ValueError,
"Both events must be created with argument 'enable_timing=True'",
):
event1.elapsed_time(event2)
event1 = torch.Event(enable_timing=True)
event2 = torch.Event(enable_timing=True)
with self.assertRaisesRegex(
ValueError,
"Both events must be recorded before calculating elapsed time",
):
event1.elapsed_time(event2)
event1 = torch.Event()
event2 = torch.Event()
with self.assertRaisesRegex(
ValueError,
"Both events must be created with argument 'enable_timing=True'",
):
event1.elapsed_time(event2)
def test_event_elapsed_time(self):
start_event = torch.Event(enable_timing=True)
end_event = torch.Event(enable_timing=True)
start_event.record()
x = torch.randn(1000, 1000, device='npu')
y = torch.randn(1000, 1000, device='npu')
z = torch.matmul(x, y)
end_event.record()
torch.npu.synchronize()
ms = start_event.elapsed_time(end_event)
self.assertGreater(ms, 0)
def test_device_context_manager(self):
prev_device = torch.accelerator.current_device_index()
with torch.accelerator.device_index(None):
self.assertEqual(torch.accelerator.current_device_index(), prev_device)
self.assertEqual(torch.accelerator.current_device_index(), prev_device)
with torch.accelerator.device_index(0):
self.assertEqual(torch.accelerator.current_device_index(), 0)
self.assertEqual(torch.accelerator.current_device_index(), prev_device)
def test_stream_context_manager_reentrance(self):
prev_stream = torch.accelerator.current_stream()
s0 = torch.Stream()
with s0, s0:
self.assertEqual(torch.accelerator.current_stream(), s0)
self.assertEqual(torch.accelerator.current_stream(), prev_stream)
s1 = torch.Stream()
with s0:
self.assertEqual(torch.accelerator.current_stream(), s0)
with s1:
self.assertEqual(torch.accelerator.current_stream(), s1)
with s0:
self.assertEqual(torch.accelerator.current_stream(), s0)
self.assertEqual(torch.accelerator.current_stream(), prev_stream)
def test_memory_interfaces_smoke(self):
"""All 10 torch.accelerator.memory APIs exist and return correct types."""
acc = torch.accelerator.current_accelerator()
tmp = torch.randn(10, device=acc)
torch.accelerator.empty_cache()
torch.accelerator.memory.empty_host_cache()
torch.accelerator.reset_peak_memory_stats()
torch.accelerator.reset_accumulated_memory_stats()
self.assertIsInstance(torch.accelerator.memory_stats(), dict)
for fn in [
torch.accelerator.memory_allocated,
torch.accelerator.max_memory_allocated,
torch.accelerator.memory_reserved,
torch.accelerator.max_memory_reserved,
]:
self.assertIsInstance(fn(), int)
free_mem, total_mem = torch.accelerator.get_memory_info()
self.assertIsInstance(free_mem, int)
self.assertIsInstance(total_mem, int)
del tmp
gc.collect()
torch.accelerator.empty_cache()
def test_memory_stats(self):
acc = torch.accelerator.current_accelerator()
tmp = torch.randn(100, device=acc)
del tmp
gc.collect()
self.assertTrue(torch._C._accelerator_isAllocatorInitialized())
torch.accelerator.empty_cache()
pool_type = ["all", "small_pool", "large_pool"]
metric_type = ["peak", "current", "allocated", "freed"]
stats_type = [
"allocated_bytes",
"reserved_bytes",
"active_bytes",
"requested_bytes",
]
mem_stats = torch.accelerator.memory_stats()
expected_stats = [
f"{st}.{pt}.{mt}"
for st in stats_type
for pt in pool_type
for mt in metric_type
]
missing_stats = [stat for stat in expected_stats if stat not in mem_stats]
self.assertEqual(
len(missing_stats),
0,
f"Missing expected memory statistics: {missing_stats}",
)
self.assertIn("num_alloc_retries", mem_stats)
self.assertIn("num_ooms", mem_stats)
self.assertIn("num_sync_all_streams", mem_stats)
self.assertIn("num_device_alloc", mem_stats)
self.assertIn("num_device_free", mem_stats)
def test_memory_allocated_reserved(self):
acc = torch.accelerator.current_accelerator()
tmp = torch.randn(100, device=acc)
del tmp
gc.collect()
self.assertTrue(torch._C._accelerator_isAllocatorInitialized())
torch.accelerator.empty_cache()
prev_allocated = torch.accelerator.memory_allocated()
prev_reserved = torch.accelerator.memory_reserved()
prev_max_allocated = torch.accelerator.max_memory_allocated()
prev_max_reserved = torch.accelerator.max_memory_reserved()
self.assertGreaterEqual(prev_allocated, 0)
self.assertGreaterEqual(prev_reserved, 0)
self.assertGreater(prev_max_allocated, 0)
self.assertGreater(prev_max_reserved, 0)
tmp = torch.ones(256, device=acc)
self.assertGreater(torch.accelerator.memory_allocated(), prev_allocated)
self.assertGreaterEqual(torch.accelerator.memory_reserved(), prev_reserved)
del tmp
gc.collect()
torch.accelerator.empty_cache()
torch.accelerator.reset_peak_memory_stats()
self.assertEqual(torch.accelerator.memory_allocated(), prev_allocated)
self.assertEqual(torch.accelerator.memory_reserved(), prev_reserved)
def test_memory_stats_active_bytes(self):
acc = torch.accelerator.current_accelerator()
tmp = torch.empty(100, device=acc)
del tmp
gc.collect()
self.assertTrue(torch._C._accelerator_isAllocatorInitialized())
torch.accelerator.empty_cache()
torch.accelerator.reset_accumulated_memory_stats()
torch.accelerator.reset_peak_memory_stats()
self.assertEqual(torch.accelerator.memory_stats()["active_bytes.all.freed"], 0)
prev_max_allocated = torch.accelerator.max_memory_allocated()
prev_max_reserved = torch.accelerator.max_memory_reserved()
prev_active_current = torch.accelerator.memory_stats()[
"active_bytes.all.current"
]
self.assertEqual(prev_max_allocated, 0)
self.assertEqual(prev_active_current, 0)
tmp = torch.randn(256, device=acc)
active_delta = (
torch.accelerator.memory_stats()["active_bytes.all.current"]
- prev_active_current
)
self.assertGreater(active_delta, 0)
del tmp
gc.collect()
torch.accelerator.synchronize()
torch.accelerator.empty_cache()
self.assertEqual(
torch.accelerator.memory_stats()["active_bytes.all.current"],
prev_active_current,
)
torch.accelerator.reset_peak_memory_stats()
self.assertEqual(torch.accelerator.max_memory_allocated(), prev_max_allocated)
self.assertEqual(torch.accelerator.max_memory_reserved(), prev_max_reserved)
def test_get_memory_info(self):
"""getMemoryInfo is an independent aclrtGetMemInfo + NPUGuard path."""
free_bytes, total_bytes = torch.accelerator.get_memory_info()
self.assertGreaterEqual(free_bytes, 0)
self.assertGreaterEqual(total_bytes, 0)
self.assertGreater(total_bytes, free_bytes)
def test_device_capability_supported_dtypes(self):
try:
caps = torch.accelerator.get_device_capability()
except RuntimeError:
self.skipTest("Backend doesn't support get_device_capability")
supported_dtypes = caps["supported_dtypes"]
self.assertIsInstance(supported_dtypes, set)
self.assertGreater(len(supported_dtypes), 0)
def test_memory_stats_consistency_with_npu(self):
"""accelerator memory_stats must match torch.npu.memory_stats exactly."""
acc_stats = torch.accelerator.memory_stats()
npu_stats = torch.npu.memory_stats()
for key in acc_stats:
self.assertIn(key, npu_stats, f"Key '{key}' missing from npu stats")
self.assertEqual(
acc_stats[key],
npu_stats[key],
f"Mismatch for key '{key}': accelerator={acc_stats[key]}, "
f"npu={npu_stats[key]}",
)
self.assertEqual(
torch.accelerator.memory_allocated(),
torch.npu.memory_allocated(),
)
self.assertEqual(
torch.accelerator.max_memory_allocated(),
torch.npu.max_memory_allocated(),
)
self.assertEqual(
torch.accelerator.memory_reserved(),
torch.npu.memory_reserved(),
)
self.assertEqual(
torch.accelerator.max_memory_reserved(),
torch.npu.max_memory_reserved(),
)
def test_graph_capture_replay(self):
if torch.accelerator.current_accelerator().type != "npu":
self.skipTest("NPU-only accelerator graph backend test")
x = torch.ones(4, device="npu")
y = torch.empty_like(x)
g = torch.accelerator.Graph()
s = torch.Stream()
with s, g:
y.copy_(x + 1)
y.zero_()
g.replay()
torch.accelerator.synchronize()
self.assertEqual(y.cpu(), torch.full((4,), 2.0))
def test_graph_capture_with_rng(self):
if torch.accelerator.current_accelerator().type != "npu":
self.skipTest("NPU-only accelerator graph backend test")
out = torch.empty(4, device="npu")
g = torch.accelerator.Graph()
s = torch.Stream()
with s, g:
out.copy_(torch.randn(4, device="npu"))
g.replay()
torch.accelerator.synchronize()
def test_graph_capture_rejects_python_npugraph_getter(self):
if torch.accelerator.current_accelerator().type != "npu":
self.skipTest("NPU-only accelerator graph backend test")
x = torch.ones(1, device="npu")
y = torch.empty_like(x)
g = torch.accelerator.Graph()
s = torch.Stream()
with s, g:
y.copy_(x)
with self.assertRaisesRegex(
RuntimeError,
"torch.accelerator.Graph capture",
):
torch.npu.NPUGraph.get_currently_capturing_graph()
if __name__ == "__main__":
run_tests()