joapolarbear / dpro Goto Github PK
View Code? Open in Web Editor NEWAnalysis for the traces from byteprofile
Analysis for the traces from byteprofile
dPRO's replayer uses a multi-frontier topological sorting algorithm. Why not just use one global queue as in Daydream.
I tried to run the script but got this error. Here is my command.
python3 analyze.py --option optimize --platform TENSORFLOW --comm_backend NCCL --nccl_algo RING --pretty --path capture_file_tf --workspace capture_file_tf
log
[2021-01-03 14:49:35] [analyze.py:16] INFO - Namespace(ckpt=False, clean=False, comm_backend='NCCL', cost_model_tmp_dir='./', debug_traces=False, del_queue=False, delay_ratio=1.1, disable_revise=False, filter=None, force=False, full_trace=False, head=None, heat_window_size=5, logging_level='INFO', mcmc_beta=100, metadata_path=None, nccl_algo='RING', no_mutation=False, optimizer='MCMC', option='optimize', path='capture_file_tf', pcap_file_path=None, platform='TENSORFLOW', pretty=True, profile_duration=None, profile_start_step=None, progress=False, relabel=False, server_log_path=None, show_queue=False, simulate=False, sort=False, step_num=1, sub_option=None, trace_level='info', ucb_gamma=0.1, ucb_type='AVG', ucb_visual=False, update_barrier=False, workspace='capture_file_tf', xlsx=False, zmq_log_path=None)
[2021-01-03 14:49:37] [dataloader.py:19] INFO - Use TENSORFLOW metadata
WARNING:tensorflow:From /home/yuchen/repos/byteprofile-analysis/cost_model_xla/gen_dataset_utils.py:11: The name tf.GraphDef is deprecated. Please use tf.compat.v1.GraphDef instead.
WARNING:tensorflow:From /home/yuchen/repos/byteprofile-analysis/cost_model_xla/gen_dataset_utils.py:15: The name tf.NodeDef is deprecated. Please use tf.compat.v1.NodeDef instead.
WARNING:tensorflow:From /home/yuchen/repos/byteprofile-analysis/cost_model_xla/gen_dataset_utils.py:23: The name tf.Session is deprecated. Please use tf.compat.v1.Session instead.
2021-01-03 14:49:38.839543: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcuda.so.1
2021-01-03 14:49:38.875380: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1618] Found device 0 with properties:
name: Tesla P100-PCIE-12GB major: 6 minor: 0 memoryClockRate(GHz): 1.3285
pciBusID: 0000:02:00.0
2021-01-03 14:49:38.876548: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1618] Found device 1 with properties:
name: Tesla P100-PCIE-12GB major: 6 minor: 0 memoryClockRate(GHz): 1.3285
pciBusID: 0000:03:00.0
2021-01-03 14:49:38.877014: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudart.so.10.0
2021-01-03 14:49:38.878811: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcublas.so.10.0
2021-01-03 14:49:38.880434: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcufft.so.10.0
2021-01-03 14:49:38.880841: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcurand.so.10.0
2021-01-03 14:49:38.882994: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcusolver.so.10.0
2021-01-03 14:49:38.884649: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcusparse.so.10.0
2021-01-03 14:49:38.889754: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudnn.so.7
2021-01-03 14:49:38.893663: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1746] Adding visible gpu devices: 0, 1
Set cost model to use GPU 1
/home/yuchen/repos/byteprofile-analysis/capture_file_tf
Traceback (most recent call last):
File "analyze.py", line 126, in <module>
clct = Collector(path_list[0], comm_backend=args_.comm_backend, platform=args.platform)
File "/home/yuchen/repos/byteprofile-analysis/collect.py", line 70, in __init__
self.pm = PathManager(root_path)
File "/home/yuchen/repos/byteprofile-analysis/trace_utils.py", line 751, in __init__
self.dir_level = self.get_dir_level(self.path)
File "/home/yuchen/repos/byteprofile-analysis/trace_utils.py", line 772, in get_dir_level
level = recur_look_up(_dir)
File "/home/yuchen/repos/byteprofile-analysis/trace_utils.py", line 770, in recur_look_up
return 1 + recur_look_up(os.path.join(root, target_dir))
File "/home/yuchen/repos/byteprofile-analysis/trace_utils.py", line 769, in recur_look_up
assert target_dir is not None, "No explicit directory found under {}".format(root)
AssertionError: No explicit directory found under /home/yuchen/repos/byteprofile-analysis/capture_file_tf/server_logs
Here is my trace directory structure.
capture_file_tf
├── collect_data.sh
├── comm_traces
│ ├── server_0.pcap
│ ├── server_1.pcap
│ ├── worker_0.pcap
│ └── worker_1.pcap
├── log_option-optimize.txt
├── run_0
│ ├── bps_cache.pickle
│ ├── bps_comm_aligned.json
│ ├── bps_trace_final.json
│ ├── comm_timeline.json
│ ├── ip_to_rank.txt
│ ├── log_option-replay.txt
│ ├── server_timeline.json
│ ├── statistic.txt
│ ├── synthetic.json
│ ├── traces_0
│ │ ├── 0
│ │ │ ├── comm.json
│ │ │ ├── dag.gml
│ │ │ ├── final_graph.json
│ │ │ ├── final_graph.pbtxt
│ │ │ ├── run_meta.json
│ │ │ ├── temp.json
│ │ │ ├── temp_json.tar
│ │ │ ├── tensor_shapes.json
│ │ │ └── variables_meta.json
│ │ └── key_dict.txt
│ ├── traces_1
│ │ └── 0
│ │ ├── comm.json
│ │ ├── dag.gml
│ │ ├── final_graph.json
│ │ ├── graph.json
│ │ ├── run_meta.json
│ │ ├── temp.json
│ │ ├── temp_json.tar
│ │ ├── tensor_shapes.json
│ │ └── variables_meta.json
│ └── trail_dag.gml
└── server_logs
├── server_log_0.txt
└── server_log_1.txt
7 directories, 37 files
A declarative, efficient, and flexible JavaScript library for building user interfaces.
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google ❤️ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.