Session misc. #9502

leaves-zwx · 2022-12-01T13:02:39Z

MultiClientSession (python/oneflow/framework/multi_client_session.py) 增加 Reset 接口
删除 SessionGlobalObjectsScope (oneflow/core/job/session_global_objects_scope.h)
删除 Cluster (oneflow/core/job/cluster.h)
删除 session 不再使用的 api (oneflow/api/python/session/session.cpp)
删除 session_util 中不再使用的 api (oneflow/api/python/framework/session_util.cpp)

github-actions · 2022-12-01T16:54:36Z

Speed stats:

GPU Name: GeForce GTX 1080 









❌ OneFlow resnet50 time: 140.5ms (= 14053.4ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 163.1ms (= 16305.4ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.16 (= 163.1ms / 140.5ms)

OneFlow resnet50 time: 86.3ms (= 8627.1ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 105.6ms (= 10557.2ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.22 (= 105.6ms / 86.3ms)

OneFlow resnet50 time: 57.9ms (= 11579.7ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 77.8ms (= 15563.0ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.34 (= 77.8ms / 57.9ms)

OneFlow resnet50 time: 45.7ms (= 9130.6ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 70.4ms (= 14084.0ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.54 (= 70.4ms / 45.7ms)

OneFlow resnet50 time: 40.5ms (= 8092.1ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 68.5ms (= 13699.9ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.69 (= 68.5ms / 40.5ms)

github-actions · 2022-12-01T17:00:43Z

View latest API docs preview at: https://staging.oneflow.info/docs/Oneflow-Inc/oneflow/pr/9502/

github-actions · 2022-12-02T02:43:25Z

Speed stats:

GPU Name: GeForce GTX 1080 









❌ OneFlow resnet50 time: 140.8ms (= 14081.4ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 163.9ms (= 16391.1ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.16 (= 163.9ms / 140.8ms)

OneFlow resnet50 time: 85.6ms (= 8560.2ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 111.3ms (= 11128.4ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.30 (= 111.3ms / 85.6ms)

OneFlow resnet50 time: 57.9ms (= 11574.7ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 77.8ms (= 15569.8ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.35 (= 77.8ms / 57.9ms)

OneFlow resnet50 time: 44.1ms (= 8813.1ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 77.7ms (= 15549.3ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.76 (= 77.7ms / 44.1ms)

OneFlow resnet50 time: 39.5ms (= 7907.6ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 67.3ms (= 13461.7ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.70 (= 67.3ms / 39.5ms)

github-actions · 2022-12-02T02:55:50Z

View latest API docs preview at: https://staging.oneflow.info/docs/Oneflow-Inc/oneflow/pr/9502/

strint · 2022-12-02T10:16:22Z

oneflow/api/python/session/session.cpp

@@ -43,6 +34,8 @@ ONEFLOW_API_PYBIND11_MODULE("", m) {
           [](MultiClientSessionContext& session, const std::string& config_proto_str) {
             return session.TryInit(config_proto_str).GetOrThrow();
           })
+      .def("try_close",
+           [](MultiClientSessionContext& session) { return session.TryClose().GetOrThrow(); })


这里关闭 Session 的操作，只是调用了 try close，没有析构 session 本身对吧。

然后再 try init 把之前释放的资源再 init 出来？

session 本身没法析构，以 shared_ptr 形式被 nn_graph 共享，我这里控制不了它的析构。

strint · 2022-12-02T10:20:31Z

python/oneflow/test/expensive/test_session_reset_stream_index_state.py

+        return torch.nn.functional.relu(x)
+
+    def setUp(self):
+        session_ctx.GetDefaultSession().Reset()


Reset 是不是会有个 bad case，主动调用 Rest 时， graph 虽然当前执行完成了，但是后面可能还要用，这时是不是会触发错误了？

所以这个 Reset 使用有个前置条件，就是调用 reset 之前，之前创建的 graph 都得析构了？

这个需要用户控制了，用户也不需要这个接口，因为实际场景不会创建那么多 graph。

github-actions · 2022-12-02T10:44:34Z

Speed stats:

GPU Name: GeForce GTX 1080 









❌ OneFlow resnet50 time: 139.8ms (= 13979.6ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 162.0ms (= 16197.6ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.16 (= 162.0ms / 139.8ms)

OneFlow resnet50 time: 84.8ms (= 8477.6ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 102.1ms (= 10208.0ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.20 (= 102.1ms / 84.8ms)

OneFlow resnet50 time: 57.5ms (= 11495.9ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 78.3ms (= 15659.6ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.36 (= 78.3ms / 57.5ms)

OneFlow resnet50 time: 44.6ms (= 8924.4ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 81.2ms (= 16231.3ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.82 (= 81.2ms / 44.6ms)

OneFlow resnet50 time: 40.1ms (= 8012.8ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 67.6ms (= 13518.0ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.69 (= 67.6ms / 40.1ms)

github-actions · 2022-12-02T12:26:34Z

Speed stats:

GPU Name: GeForce GTX 1080 









❌ OneFlow resnet50 time: 139.7ms (= 13969.5ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 161.0ms (= 16100.9ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.15 (= 161.0ms / 139.7ms)

OneFlow resnet50 time: 85.2ms (= 8517.5ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 102.2ms (= 10223.7ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.20 (= 102.2ms / 85.2ms)

OneFlow resnet50 time: 57.3ms (= 11469.4ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 87.4ms (= 17471.5ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.52 (= 87.4ms / 57.3ms)

OneFlow resnet50 time: 45.1ms (= 9025.0ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 69.4ms (= 13885.6ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.54 (= 69.4ms / 45.1ms)

OneFlow resnet50 time: 40.1ms (= 8012.3ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 69.2ms (= 13840.4ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.73 (= 69.2ms / 40.1ms)

github-actions · 2022-12-02T12:30:28Z

View latest API docs preview at: https://staging.oneflow.info/docs/Oneflow-Inc/oneflow/pr/9502/

chengtbf · 2022-12-09T07:04:16Z

oneflow/api/cpp/env_impl.cpp

  config_proto.set_session_id(session_id);
-
+  CHECK(of::RegsterSessionId(session_id));


session id 未来也可以优化掉。这个概念应该没有实际的用处。

这个 session_id 我尝试删过了，删不掉。scope 还用到了 session_id。而且系统中全局搜索了 session id，还有很多地方使用，它们以什么方式取的得 session id，路劲上全部要修改。牵扯的改动挺多。所以还是另外的 pr 再去改了。

可以，我原本想的就是删掉 scope 中的 session id。不过可以后面再搞。 scope 本来也很重度，还有 symbol，后面一块儿重构吧。

chengtbf · 2022-12-09T07:08:43Z

oneflow/core/framework/session_util.cpp

-  static HashMap<int64_t, std::shared_ptr<Session>> id2session_map;
-  return &id2session_map;
-}
-
 std::vector<int64_t>* RegsiteredSessionIds() {


要不顺道直接删掉 session id 吧，不需要记录 session id 2 session 的映射，没有多 session 同时存在的场景， scope 中的 session id 也可以删掉。

只保留 default session 的全局对象就行。

chengtbf · 2022-12-09T07:09:58Z

python/oneflow/framework/session_context.py

-    new_default_sess = MultiClientSession(env, oneflow._oneflow_internal.NewSessionId())
+    session_id = oneflow._oneflow_internal.NewSessionId()
+    assert oneflow._oneflow_internal.RegsterSessionId(session_id)
+    new_default_sess = MultiClientSession(env, session_id)
    global _sess_id2sess


不需要这个 global，可以直接用 global sess，去掉 id 相关的所有函数。

strint · 2022-12-09T08:52:41Z

是不是可以用这个功能来试验下用这个来清理 CI 中的 session。

然后直接就加到 CI 的 unittest 里面。

chengtbf · 2022-12-13T02:52:24Z

python/oneflow/framework/unittest.py

+            os.environ.get("ONEFLOW_TEST_RESET_SESSION_PERIOD", "10")
+        )
+        if RESET_SESSION_COUNT >= reset_session_period:
+            oneflow.framework.session_context.GetDefaultSession().Reset()


这里会影响 CI 吗？比如 CI 里自动进行 reset

这个就是让 ci 每测试若干用例，就 reset 一次来重置 stream index 计数，避免出现 stream_index > 4096 的报错。

github-actions · 2022-12-13T08:19:12Z

Speed stats:

GPU Name: GeForce GTX 1080 









❌ OneFlow resnet50 time: 139.8ms (= 13982.7ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 161.2ms (= 16124.6ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.15 (= 161.2ms / 139.8ms)

OneFlow resnet50 time: 84.7ms (= 8466.1ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 100.7ms (= 10073.1ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.19 (= 100.7ms / 84.7ms)

OneFlow resnet50 time: 57.9ms (= 11575.6ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 77.6ms (= 15528.7ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.34 (= 77.6ms / 57.9ms)

OneFlow resnet50 time: 45.0ms (= 9007.4ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 80.0ms (= 15992.5ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.78 (= 80.0ms / 45.0ms)

OneFlow resnet50 time: 40.8ms (= 8163.6ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 74.3ms (= 14858.3ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.82 (= 74.3ms / 40.8ms)

github-actions · 2022-12-13T08:25:03Z

View latest API docs preview at: https://staging.oneflow.info/docs/Oneflow-Inc/oneflow/pr/9502/

github-actions · 2022-12-13T08:28:20Z

CI failed when running job: cuda-speed-test. PR label automerge has been removed

github-actions · 2022-12-19T05:25:42Z

Speed stats:

GPU Name: GeForce GTX 1080 









❌ OneFlow resnet50 time: 139.4ms (= 13939.5ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 160.9ms (= 16092.9ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.15 (= 160.9ms / 139.4ms)

OneFlow resnet50 time: 85.2ms (= 8516.9ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 101.6ms (= 10155.7ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.19 (= 101.6ms / 85.2ms)

OneFlow resnet50 time: 57.7ms (= 11534.8ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 77.9ms (= 15570.1ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.35 (= 77.9ms / 57.7ms)

OneFlow resnet50 time: 44.0ms (= 8804.4ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 80.3ms (= 16052.9ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.82 (= 80.3ms / 44.0ms)

OneFlow resnet50 time: 40.7ms (= 8146.1ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 74.4ms (= 14877.5ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.83 (= 74.4ms / 40.7ms)

github-actions · 2022-12-19T05:32:39Z

View latest API docs preview at: https://staging.oneflow.info/docs/Oneflow-Inc/oneflow/pr/9502/

github-actions · 2022-12-19T14:03:22Z

Speed stats:

GPU Name: GeForce GTX 1080 









❌ OneFlow resnet50 time: 139.6ms (= 13963.9ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 160.7ms (= 16065.6ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.15 (= 160.7ms / 139.6ms)

OneFlow resnet50 time: 84.6ms (= 8459.5ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 101.5ms (= 10145.8ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.20 (= 101.5ms / 84.6ms)

OneFlow resnet50 time: 57.1ms (= 11427.0ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 77.1ms (= 15411.3ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.35 (= 77.1ms / 57.1ms)

OneFlow resnet50 time: 43.9ms (= 8787.4ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 78.7ms (= 15749.1ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.79 (= 78.7ms / 43.9ms)

OneFlow resnet50 time: 39.1ms (= 7821.4ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 68.7ms (= 13737.8ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.76 (= 68.7ms / 39.1ms)

github-actions · 2022-12-19T14:17:41Z

View latest API docs preview at: https://staging.oneflow.info/docs/Oneflow-Inc/oneflow/pr/9502/

Fixes https://github.com/Oneflow-Inc/OneTeam/issues/1861 ，在 TryCloseDefaultSession 之前设置 is_shutting_down 为 true（#9502 这个 PR 在 TryCloseDefaultSession 里引入了 Sync 操作），并把 python c 对象的指针保存在 PyFrame 对象里，GetCurrentFrame 里不再需要重复获取，修复因为没有拿到 gil 锁导致的 segfault Signed-off-by: daquexian <daquexian566@gmail.com> Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>

add Reset for MultiClientSession

b17b06c

leaves-zwx requested review from chengtbf, strint, BBuf, daquexian and jackalcooper as code owners December 1, 2022 13:02

leaves-zwx added 2 commits December 1, 2022 14:18

test

edb5dd6

reset session in unittest

0fe1b21

leaves-zwx self-assigned this Dec 1, 2022

leaves-zwx added bug enhancement python and removed bug labels Dec 1, 2022

leaves-zwx requested a review from oneflow-ci-bot December 1, 2022 15:19

move reset to setUp

209908d

leaves-zwx added 3 commits December 2, 2022 05:45

delete SessionGlobalObjectsScope and related

99b3da6

delete out-of-date api in session util

56d5e0e

RegsterSessionId

8fec3ea

strint reviewed Dec 2, 2022

View reviewed changes

fix reg session_id

40c2931

chengtbf reviewed Dec 9, 2022

View reviewed changes

chengtbf added the graph graph mode label Dec 9, 2022

strint approved these changes Dec 9, 2022

View reviewed changes

Merge branch 'master' into session_reset

55f6215

leaves-zwx added the automerge label Dec 13, 2022

chengtbf approved these changes Dec 13, 2022

View reviewed changes

chengtbf reviewed Dec 13, 2022

View reviewed changes

Merge branch 'master' into session_reset

ce2f223

github-actions bot removed the automerge label Dec 13, 2022

lixiang007666 mentioned this pull request Dec 14, 2022

check graph op global test #8614

Merged

1 task

Merge branch 'master' into session_reset

e6a2779

leaves-zwx added the automerge label Dec 19, 2022

leaves-zwx and others added 2 commits December 19, 2022 14:49

Merge branch 'master' into session_reset

9249b6f

Merge branch 'master' into session_reset

1c33add

mergify bot merged commit 658c6f4 into master Dec 19, 2022

mergify bot deleted the session_reset branch December 19, 2022 14:50

daquexian mentioned this pull request Jan 3, 2023

Fix stack getter segfault caused by is_shutting_down and gil #9681

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Session misc. #9502

Session misc. #9502

leaves-zwx commented Dec 1, 2022 •

edited

Loading

github-actions bot commented Dec 1, 2022

github-actions bot commented Dec 1, 2022

github-actions bot commented Dec 2, 2022

github-actions bot commented Dec 2, 2022

strint Dec 2, 2022

leaves-zwx Dec 2, 2022

strint Dec 2, 2022

leaves-zwx Dec 2, 2022

github-actions bot commented Dec 2, 2022

github-actions bot commented Dec 2, 2022

github-actions bot commented Dec 2, 2022

chengtbf Dec 9, 2022

leaves-zwx Dec 9, 2022

chengtbf Dec 9, 2022

chengtbf Dec 9, 2022

chengtbf Dec 9, 2022

chengtbf Dec 9, 2022

strint commented Dec 9, 2022

chengtbf Dec 13, 2022

leaves-zwx Dec 13, 2022

github-actions bot commented Dec 13, 2022

github-actions bot commented Dec 13, 2022

github-actions bot commented Dec 13, 2022

github-actions bot commented Dec 19, 2022

github-actions bot commented Dec 19, 2022

github-actions bot commented Dec 19, 2022

github-actions bot commented Dec 19, 2022

		config_proto.set_session_id(session_id);

		CHECK(of::RegsterSessionId(session_id));

Session misc. #9502

Session misc. #9502

Conversation

leaves-zwx commented Dec 1, 2022 • edited Loading

github-actions bot commented Dec 1, 2022

github-actions bot commented Dec 1, 2022

github-actions bot commented Dec 2, 2022

github-actions bot commented Dec 2, 2022

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

github-actions bot commented Dec 2, 2022

github-actions bot commented Dec 2, 2022

github-actions bot commented Dec 2, 2022

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

strint commented Dec 9, 2022

Choose a reason for hiding this comment

Choose a reason for hiding this comment

github-actions bot commented Dec 13, 2022

github-actions bot commented Dec 13, 2022

github-actions bot commented Dec 13, 2022

github-actions bot commented Dec 19, 2022

github-actions bot commented Dec 19, 2022

github-actions bot commented Dec 19, 2022

github-actions bot commented Dec 19, 2022

leaves-zwx commented Dec 1, 2022 •

edited

Loading