File size: 3,172 Bytes
ecf08bc |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 |
# Copyright 2020 Division of Medical Image Computing, German Cancer Research Center (DKFZ), Heidelberg, Germany
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import torch
from torch import distributed
from torch import autograd
from torch.nn.parallel import DistributedDataParallel as DDP
def print_if_rank0(*args):
if distributed.get_rank() == 0:
print(*args)
class awesome_allgather_function(autograd.Function):
@staticmethod
def forward(ctx, input):
world_size = distributed.get_world_size()
# create a destination list for the allgather. I'm assuming you're gathering from 3 workers.
allgather_list = [torch.empty_like(input) for _ in range(world_size)]
#if distributed.get_rank() == 0:
# import IPython;IPython.embed()
distributed.all_gather(allgather_list, input)
return torch.cat(allgather_list, dim=0)
@staticmethod
def backward(ctx, grad_output):
#print_if_rank0("backward grad_output len", len(grad_output))
#print_if_rank0("backward grad_output shape", grad_output.shape)
grads_per_rank = grad_output.shape[0] // distributed.get_world_size()
rank = distributed.get_rank()
# We'll receive gradients for the entire catted forward output, so to mimic DataParallel,
# return only the slice that corresponds to this process's input:
sl = slice(rank * grads_per_rank, (rank + 1) * grads_per_rank)
#print("worker", rank, "backward slice", sl)
return grad_output[sl]
if __name__ == "__main__":
import torch.distributed as dist
import argparse
from torch import nn
from torch.optim import Adam
argumentparser = argparse.ArgumentParser()
argumentparser.add_argument("--local_rank", type=int)
args = argumentparser.parse_args()
torch.cuda.set_device(args.local_rank)
dist.init_process_group(backend='nccl', init_method='env://')
rnd = torch.rand((5, 2)).cuda()
rnd_gathered = awesome_allgather_function.apply(rnd)
print("gathering random tensors\nbefore\b", rnd, "\nafter\n", rnd_gathered)
# so far this works as expected
print("now running a DDP model")
c = nn.Conv2d(2, 3, 3, 1, 1, 1, 1, True).cuda()
c = DDP(c)
opt = Adam(c.parameters())
bs = 5
if dist.get_rank() == 0:
bs = 4
inp = torch.rand((bs, 2, 5, 5)).cuda()
out = c(inp)
print("output_shape", out.shape)
out_gathered = awesome_allgather_function.apply(out)
print("output_shape_after_gather", out_gathered.shape)
# this also works
loss = out_gathered.sum()
loss.backward()
opt.step()
|