mirror of
https://github.com/hpcaitech/ColossalAI.git
synced 2025-10-11 08:58:34 +00:00
[model checkpoint] add gloo groups for cpu tensor communication (#589)
This commit is contained in:
@@ -34,17 +34,20 @@ class Initializer_Tensor(ProcessGroupInitializer):
|
||||
local_rank = None
|
||||
ranks_in_group = None
|
||||
process_group = None
|
||||
cpu_group = None
|
||||
group_world_size = None
|
||||
mode = ParallelMode.TENSOR
|
||||
|
||||
for i in range(self.num_tensor_parallel_group):
|
||||
ranks = [i * self.tensor_parallel_size + j for j in range(self.tensor_parallel_size)]
|
||||
group = dist.new_group(ranks)
|
||||
group_cpu = dist.new_group(ranks, backend='gloo') if dist.get_backend() != 'gloo' else group
|
||||
|
||||
if self.rank in ranks:
|
||||
local_rank = ranks.index(self.rank)
|
||||
group_world_size = len(ranks)
|
||||
process_group = group
|
||||
cpu_group = group_cpu
|
||||
ranks_in_group = ranks
|
||||
|
||||
return local_rank, group_world_size, process_group, ranks_in_group, mode
|
||||
return local_rank, group_world_size, process_group, cpu_group, ranks_in_group, mode
|
||||
|
Reference in New Issue
Block a user