8000 update xeon launch script for Intel(R) Xeon 6 support by jingxu10 · Pull Request #133835 · pytorch/pytorch · GitHub
[go: up one dir, main page]
Skip to content

update xeon launch script for Intel(R) Xeon 6 support #133835

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
wants to merge 1 commit into from
Closed

update xeon launch script for Intel(R) Xeon 6 support #133835

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion docs/source/backends.rst
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -234,4 +234,4 @@ torch.backends.opt_einsum
torch.backends.xeon
^^^^^^^^^^^^^^^^^^^
.. automodule:: torch.backends.xeon
.. py:module:: torch.backends.xeon.run_cpu
.. automodule:: torch.backends.xeon.run_cpu
1 change: 1 addition & 0 deletions setup.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1098,6 +1098,7 @@ def make_relative_rpath_args(path):
"convert-caffe2-to-onnx = caffe2.python.onnx.bin.conversion:caffe2_to_onnx",
"convert-onnx-to-caffe2 = caffe2.python.onnx.bin.conversion:onnx_to_caffe2",
"torchrun = torch.distributed.run:main",
"torch-xeon-launcher = torch.backends.xeon.run_cpu:main",
],
"torchrun.logs_specs": [
"default = torch.distributed.elastic.multiprocessing:DefaultLogsSpecs",
Expand Down
115 changes: 77 additions & 38 deletions test/backends/xeon/test_launch.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -17,38 +17,57 @@ def tearDown(self):
shutil.rmtree(self._test_dir)

def test_cpu_info(self):
lscpu_info = """# The following is the parsable format, which can be fed to other
# programs. Each different item in every column has an unique ID
# starting from zero.
# CPU,Core,Socket,Node
0,0,0,0
1,1,0,0
2,2,0,0
3,3,0,0
4,4,1,1
5,5,1,1
6,6,1,1
7,7,1,1
8,0,0,0
9,1,0,0
10,2,0,0
11,3,0,0
12,4,1,1
13,5,1,1
14,6,1,1
15,7,1,1
lscpu_info = """
CPU NODE SOCKET CORE L1d:L1i:L2:L3 ONLINE MAXMHZ MINMHZ MHZ
0 0 0 0 0:0:0:0 yes 5000.0000 800.0000 2400.000
1 0 0 1 0:0:0:0 yes 5000.0000 800.0000 2400.000
2 0 0 2 0:0:0:0 yes 5000.0000 800.0000 2400.000
3 0 0 3 0:0:0:0 yes 5000.0000 800.0000 2400.000
4 1 1 4 0:0:0:0 yes 5000.0000 800.0000 2400.000
5 1 1 5 0:0:0:0 yes 5000.0000 800.0000 2400.000
6 1 1 6 0:0:0:0 yes 5000.0000 800.0000 2400.000
7 1 1 7 0:0:0:0 yes 5000.0000 800.0000 2400.000
8 0 0 0 0:0:0:0 yes 5000.0000 800.0000 2400.000
9 0 0 1 0:0:0:0 yes 5000.0000 800.0000 2400.000
10 0 0 2 0:0:0:0 yes 5000.0000 800.0000 2400.000
11 0 0 3 0:0:0:0 yes 5000.0000 800.0000 2400.000
12 1 1 4 0:0:0:0 yes 5000.0000 800.0000 2400.000
13 1 1 5 0:0:0:0 yes 5000.0000 800.0000 2400.000
14 1 1 6 0:0:0:0 yes 5000.0000 800.0000 2400.000
15 1 1 7 0:0:0:0 yes 5000.0000 800.0000 2400.000
"""
from torch.backends.xeon.run_cpu import _CPUinfo
from torch.backends.xeon._cpu_info import CPUPoolList

cpuinfo = _CPUinfo(lscpu_info)
assert cpuinfo._physical_core_nums() == 8
assert cpuinfo._logical_core_nums() == 16
assert cpuinfo.get_node_physical_cores(0) == [0, 1, 2, 3]
assert cpuinfo.get_node_physical_cores(1) == [4, 5, 6, 7]
assert cpuinfo.get_node_logical_cores(0) == [0, 1, 2, 3, 8, 9, 10, 11]
assert cpuinfo.get_node_logical_cores(1) == [4, 5, 6, 7, 12, 13, 14, 15]
assert cpuinfo.get_all_physical_cores() == [0, 1, 2, 3, 4, 5, 6, 7]
assert cpuinfo.get_all_logical_cores() == [
cpupool = CPUPoolList(lscpu_txt=lscpu_info)
assert [c.cpu for c in cpupool.pool_all] == [
0,
1,
2,
3,
4,
5,
6,
7,
8, # noqa: Q003
9,
10,
11,
12,
13,
14,
15,
]
assert [c.cpu for c in cpupool.pool_all if c.is_physical_core] == [
0,
1,
2,
3,
4,
5, # noqa: Q003
6,
7,
]
assert [c.cpu for c in cpupool.pool_all if c.node == 0] == [
0,
1,
2,
Expand All @@ -57,6 +76,8 @@ def test_cpu_info(self):
9,
10,
11,
]
assert [c.cpu for c in cpupool.pool_all if c.node == 1] == [
4,
5,
6,
Expand All @@ -66,22 +87,40 @@ def test_cpu_info(self):
14,
15,
]
assert cpuinfo.numa_aware_check([0, 1, 2, 3]) == [0]
assert cpuinfo.numa_aware_check([4, 5, 6, 7]) == [1]
assert cpuinfo.numa_aware_check([2, 3, 4, 5]) == [0, 1]
assert [
c.cpu for c in cpupool.pool_all if c.node == 0 and c.is_physical_core
] == [0, 1, 2, 3]
assert [
c.cpu for c in cpupool.pool_all if c.node == 1 and c.is_physical_core
] == [4, 5, 6, 7]

def test_multi_threads_module(self):
num = 0
with subprocess.Popen(
f'python -m torch.backends.xeon.run_cpu --ninstances 4 --memory-allocator default \
--omp-runtime default --multi-task-manager none --log-dir {self._test_dir} --no-python echo "test"',
shell=True,
stdout=subprocess.PIPE,
stderr=subprocess.STDOUT,
) as p:
for line in p.stdout.readlines():
segs = str(line, "utf-8").strip().split(":")
if segs[-1].strip() == "test":
num += 1
assert num == 4, "Failed to launch multiple instances for inference"

def test_multi_threads(self):
def test_multi_threads_command(self):
num = 0
with subprocess.Popen(
f"python -m torch.backends.xeon.run_cpu --ninstances 4 --use-default-allocator \
--disable-iomp --disable-numactl --disable-taskset --log-path {self._test_dir} --no-python pwd",
f'torch-xeon-launcher --ninstances 4 --memory-allocator default \
--omp-runtime default --multi-task-manager none --log-dir {self._test_dir} --no-python echo "test"',
shell=True,
stdout=subprocess.PIPE,
stderr=subprocess.STDOUT,
) as p:
for line in p.stdout.readlines():
segs = str(line, "utf-8").strip().split("-")
if segs[-1].strip() == "pwd":
segs = str(line, "utf-8").strip().split(":")
if segs[-1].strip() == "test":
num += 1
assert num == 4, "Failed to launch multiple instances for inference"

Expand Down
1 change: 1 addition & 0 deletions torch/backends/__init__.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -69,4 +69,5 @@ def __getattr__(self, attr):
nnpack as nnpack,
openmp as openmp,
quantized as quantized,
xeon as xeon,
)
1 change: 1 addition & 0 deletions torch/backends/xeon/__init__.py
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1 @@
from . import run_cpu
Loading
Loading
0