Skip to content

add a more example of gpu_util usage and fix main.py #90

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

Merged
merged 2 commits into from
Jan 8, 2021
Merged

add a more example of gpu_util usage and fix main.py #90

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
39a22ce
add a more example of gpu_util usage
wizard1203 Jan 8, 2021
4359c0f
fix launch/main.py
wizard1203 Jan 8, 2021
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
11 changes: 11 additions & 0 deletions fedml_experiments/distributed/fed_launch/README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -47,6 +47,17 @@ config_11:
This example is also used for 11 process. But the mapping is different: Server process -- host1:GPU:0, client 1 -- host1:GPU:0, client 2 -- host1:GPU:1, client 3 -- host1:GPU:1, client 4 -- host2:GPU:0, client 5 -- host2:GPU:1, client 6 -- host2:GPU:2, client 7 -- host3:GPU:0, client 8 -- host3:GPU:1, client 9 -- host3:GPU:2, client 10 -- host3:GPU:3


Sometimes one may want to use some GPUs in one machine, instead of all GPUs. Then you can use this:
```
config_11:
host1: [0, 2]
host2: [1, 0, 1]
host3: [1, 1, 0, 1]
host4: [0, 1, 0, 0, 0, 1, 0, 2]
```
Now the mapping become: Server process -- host1:GPU:1, client 1 -- host1:GPU:1, client 2 -- host1:GPU:0, client 3 -- host1:GPU:2, client 4 -- host3:GPU:0, client 5 -- host3:GPU:1, client 6 -- host3:GPU:3, client 7 -- host4:GPU:1, client 8 -- host4:GPU:6, client 9 -- host4:GPU:7, client 10 -- host4:GPU:7.


And you also can add many mappings in one yaml file like this:
```
config_11:
Expand Down
25 changes: 0 additions & 25 deletions fedml_experiments/distributed/fed_launch/main.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -467,31 +467,6 @@ def init_training_device_from_gpu_util_file(process_id, worker_number, gpu_util_
FedML_FedAvg_distributed(process_id, worker_number, device, comm,
model, train_data_num, train_data_global, test_data_global,
train_data_local_num_dict, train_data_local_dict, test_data_local_dict, args)
elif args.algorithm == 'PSGD':
from fedml_api.distributed.PSGD.PSGD_API import FedML_init, FedML_PSGD_distributed
FedML_PSGD_distributed(process_id, worker_number, device, comm,
model, train_data_num, train_data_global, test_data_global,
train_data_local_num_dict, train_data_local_dict, test_data_local_dict, args)
elif args.algorithm == 'DPSGD':
from fedml_api.distributed.DPSGD.DPSGD_API import FedML_init, FedML_DPSGD
FedML_DPSGD(process_id, worker_number, device, comm,
model, train_data_num, train_data_global, test_data_global,
train_data_local_num_dict, train_data_local_dict, test_data_local_dict, args)
elif args.algorithm == 'DCD_PSGD':
from fedml_api.distributed.DCD_PSGD.DCD_PSGD_API import FedML_init, FedML_DCD_PSGD
FedML_DCD_PSGD(process_id, worker_number, device, comm,
model, train_data_num, train_data_global, test_data_global,
train_data_local_num_dict, train_data_local_dict, test_data_local_dict, args)
elif args.algorithm == 'CHOCO_SGD':
from fedml_api.distributed.CHOCO_SGD.CHOCO_SGD_API import FedML_init, FedML_CHOCO_SGD
FedML_CHOCO_SGD(process_id, worker_number, device, comm,
model, train_data_num, train_data_global, test_data_global,
train_data_local_num_dict, train_data_local_dict, test_data_local_dict, args)
elif args.algorithm == 'CHOCO_SGD':
from fedml_api.distributed.SAPS_FL.SAPS_FL_API import FedML_init, FedML_SAPS_FL
FedML_SAPS_FL(process_id, worker_number, device, comm,
model, train_data_num, train_data_global, test_data_global,
train_data_local_num_dict, train_data_local_dict, test_data_local_dict, args)
else:
raise NotImplementedError

Expand Down