change chunking scheme for full-allreduce case, add parameter order argument,...

change chunking scheme for full-allreduce case, add parameter order argument, both to enable contiguous chunking of allgather (#1190)

apex/contrib/optimizers/distributed_fused_lamb.py

+import os
 import math
 import torch
 import importlib
@@ -89,7 +90,7 @@ class DistributedFusedLAMB(torch.optim.Optimizer):
                  dwu_num_rs_pg=1, dwu_num_ar_pg=4, dwu_num_ag_pg=0, fused_norm=False,
                  e5m2_allgather=False, verbose=False, clip_after_ar=True,
                  full_ar=False, set_param_views_to_flat_buffer=False, skip_allgather=False,
-                 fuse_scale=False):
+                 fuse_scale=False, param_order=None, nccl_allgather_channels=0):
         defaults = dict(lr=lr, bias_correction=bias_correction,
                         betas=betas, eps=eps, weight_decay=weight_decay,
                         grad_averaging=grad_averaging,
@@ -170,7 +171,8 @@ class DistributedFusedLAMB(torch.optim.Optimizer):
                 if torch.distributed.get_rank() in ranks:
                     self._ar_pg.append(grp)
             self._ar_st = [torch.cuda.Stream() for _ in range(self._num_ar_pg)]
-
+            if nccl_allgather_channels > 0:
+                os.putenv('NCCL_MAX_NCHANNELS', str(nccl_allgather_channels))
             if self._num_ag_pg == 0:
                 self._ag_pg = self._ar_pg
                 self._ag_st = self._ar_st
@@ -283,6 +285,10 @@ class DistributedFusedLAMB(torch.optim.Optimizer):
                     ))
                 p_grads_size = p.numel()
                 if self._set_flat_param_view:
+                    if param_order:
+                        # this is executed when param_order is specified by the user
+                        self._param_order.add(param_order[p])
+                    else:
                         self._param_order.add(p_i)
                 p_offset += p_grads_size
                 # Only enforce 128b alignment (64 * fp16) for non-consecutive parameters
@@ -292,6 +298,8 @@ class DistributedFusedLAMB(torch.optim.Optimizer):
                     p_offset = ((p_offset + 63) // 64) * 64
                 prev = p
                 p_i += 1
+        if param_order:
+            self._param_order.order = torch.argsort(torch.tensor(self._param_order.order)).tolist()
         self._grads_generated = [False]*len(self._model_params)
         self._grads_fp16, self._grads_fp32 = [], []
         if self._overlap_reductions:
@@ -330,7 +338,8 @@ class DistributedFusedLAMB(torch.optim.Optimizer):
                                 self._do_overlapped_reduction(idx, param)
                         else:
                             if not self._first_step:
-                                self._do_overlapped_reduction(param_i, param)
+                                idx = self._param_order.order.index(param_i)
+                                self._do_overlapped_reduction(idx, param)
                     grad_acc.register_hook(allreduce_hook)
                     self._grad_accs.append(grad_acc)
                 wrapper(p, p_i)
@@ -359,7 +368,7 @@ class DistributedFusedLAMB(torch.optim.Optimizer):
                 return [p[chunk_id*self._chunk_size:(chunk_id+1)*self._chunk_size] for chunk_id in range(self._num_chunks)]
             def __shardify(p):
                 return [p[shard_id*self._shard_size:(shard_id+1)*self._shard_size] for shard_id in range(self._group_size)]
-            list_of_blocks = __blockify(self._flat_grads)
+            list_of_blocks = __blockify(p)
             list_of_list_of_chunks = [__chunkify(block) for block in list_of_blocks]
             list_of_list_of_list_of_shards = [[__shardify(chunk) for chunk in chunks] for chunks in list_of_list_of_chunks]
             return list_of_blocks, list_of_list_of_chunks, list_of_list_of_list_of_shards
@@ -392,8 +401,20 @@ class DistributedFusedLAMB(torch.optim.Optimizer):
             list_of_blocks = __packed_blockify(p)
             list_of_list_of_chunks = [__packed_chunkify(block) for block in list_of_blocks]
             return list_of_blocks, list_of_list_of_chunks
+        def _split_assign(shards):
+            packed_block_size = self._num_chunks*self._shard_size
+            list_of_list_of_chunks=[]
+            for block_id in range(self._num_blocks):
+                list_of_chunks=[]
+                for chunk_id in range(self._num_chunks):
+                    #self._fp16_g[block_id*packed_block_size+chunk_id*self._shard_size:block_id*packed_block_size+(chunk_id+1)*self._shard_size] = shards[block_id][chunk_id][self._rank_in_group]
+                    list_of_chunks.append( shards[block_id][chunk_id][self._rank_in_group])
+                list_of_list_of_chunks.append(list_of_chunks)
+            return list_of_list_of_chunks
 
         self._new_params_mega_shards, self._new_params_mega_blocks, self._new_params_mega_chunks = _full_packed_split(self._new_params)
+        # this splitting scheme is needed when allgather needs to be split into multiple chunks in a contiguous way
+        self._new_params2_blocks, self._new_params2_chunks, self._new_params2_shards = _flat_split(self._new_params)
 
         self._fp32_p_blocks, self._fp32_p_chunks = _packed_split(self._fp32_p)
         self._fp32_m_blocks, self._fp32_m_chunks = _packed_split(self._fp32_m)
@@ -403,10 +424,9 @@ class DistributedFusedLAMB(torch.optim.Optimizer):
 
         if self._full_ar:
             # for gradient all-reduce
-            self._flat_grads_blocks, self._flat_grads_chunks = _flat_split_no_shards(self._flat_grads)
+            self._flat_grads_blocks, self._flat_grads_chunks, self._flat_grads_shards = _flat_split(self._flat_grads)
             # for weight update
-            self._flat_grads_shards, _, _ = _full_packed_split(self._flat_grads)
-            self._fp16_g_blocks, self._fp16_g_chunks = _packed_split(self._flat_grads_shards[self._rank_in_group])
+            self._fp16_g_chunks = _split_assign(self._flat_grads_shards)
         else:
             self._flat_grads_blocks, self._flat_grads_chunks, self._flat_grads_shards = _flat_split(self._flat_grads)
             self._fp16_g_blocks, self._fp16_g_chunks = _packed_split(self._fp16_g)
@@ -416,6 +436,7 @@ class DistributedFusedLAMB(torch.optim.Optimizer):
     def _lazy_init_stage2(self):
         if self._lazy_init_stage2_done: return
         if not self._set_flat_param_view: 
+            # reversing is needed for overlapping allreduce and backprop, but currently not supported for flat param view
             self._param_order.order.reverse()
 
             # re-order model_params, grad_accs, group_properties lists
@@ -570,13 +591,21 @@ class DistributedFusedLAMB(torch.optim.Optimizer):
 
     def _full_all_reduce_scale(self, block_id, scale):
         works = [None]*self._num_chunks
-
+        if  self._clip_after_ar:    
             for chunk_id in range(self._num_chunks):
                 glob_chunk_id = block_id * self._num_chunks + chunk_id
                 ar_stream = self._ar_st[glob_chunk_id%self._num_ar_pg]
                 ar_stream.wait_stream(torch.cuda.current_stream())
                 with torch.cuda.stream(ar_stream):
                     works[chunk_id] = torch.distributed.all_reduce(self._flat_grads_chunks[block_id][chunk_id],group=self._ar_pg[glob_chunk_id%self._num_ar_pg],async_op=True,op=torch.distributed.make_nccl_premul_sum((scale,)))
+        else:
+            glob_chunk_id = block_id
+            ar_stream = self._ar_st[glob_chunk_id%self._num_ar_pg]
+            ar_stream.wait_stream(torch.cuda.current_stream())
+            with torch.cuda.stream(ar_stream):
+                    works0 = torch.distributed.all_reduce(self._flat_grads_blocks[block_id],group=self._ar_pg[glob_chunk_id%self._num_ar_pg],async_op=True,op=torch.distributed.make_nccl_premul_sum((scale,)))
+            for i in range(self._num_chunks):
+                works[i]=works0
         self._reductions_works[block_id] = works
 
     def _full_all_reduce(self, block_id):
@@ -653,7 +682,7 @@ class DistributedFusedLAMB(torch.optim.Optimizer):
                             self._reductions_works[block_id][chunk_id].wait()
                     # Since the packed format is contiguous after reductions, only one norm is needed
                     l2_grad_norm_sq = torch.empty([1], device='cuda')
-                    if self._full_ar:
+                    if 0:#self._full_ar:
                         l2_grad_norm_sq = self._flat_grads_shards[self._rank_in_group].norm(dtype=torch.float32, p=2)**2
                     else:
                         l2_grad_norm_sq = self._fp16_g.norm(dtype=torch.float32, p=2)**2
@@ -775,6 +804,12 @@ class DistributedFusedLAMB(torch.optim.Optimizer):
                     global_grad_norm,
                     self._use_nvlamb)
             if not self._skip_ag:
+                # allgather chunking is currently not supported for clip after allreduce 
+                if not self._clip_after_ar:
+                    for block in range(self._num_blocks):
+                        for chunk in range(self._num_chunks):
+                            torch.distributed.all_gather(self._new_params2_shards[block][chunk], self._fp16_p_chunks[block][chunk], group=self._ag_pg[0], no_copy=True)
+                else:
                     torch.distributed.all_gather(self._new_params_mega_shards, self._fp16_p, group=self._ag_pg[0], no_copy=True)
 
     def _flatten_grad_mt(self, scale):