comfy-aimdo 0.2 - Improved pytorch allocator integration (Comfy-Org#12557)

Integrate comfy-aimdo 0.2 which takes a different approach to
installing the memory allocator hook. Instead of using the complicated
and buggy pytorch MemPool+CudaPluggableAlloctor, cuda is directly hooked
making the process much more transparent to both comfy and pytorch. As
far as pytorch knows, aimdo doesnt exist anymore, and just operates
behind the scenes.

Remove all the mempool setup stuff for dynamic_vram and bump the
comfy-aimdo version. Remove the allocator object from memory_management
and demote its use as an enablment check to a boolean flag.

Comfy-aimdo 0.2 also support the pytorch cuda async allocator, so
remove the dynamic_vram based force disablement of cuda_malloc and
just go back to the old settings of allocators based on command line
input.

Author: rattus128

comfy/memory_management.py

@@ -78,4 +78,4 @@ def interpret_gathered_like(tensors, gathered):
 
     return dest_views
 
-aimdo_allocator = None
+aimdo_enabled = False

comfy/model_management.py

@@ -836,7 +836,7 @@ def unet_inital_load_device(parameters, dtype):
 
     mem_dev = get_free_memory(torch_dev)
     mem_cpu = get_free_memory(cpu_dev)
-    if mem_dev > mem_cpu and model_size < mem_dev and comfy.memory_management.aimdo_allocator is None:
+    if mem_dev > mem_cpu and model_size < mem_dev and comfy.memory_management.aimdo_enabled:
         return torch_dev
     else:
         return cpu_dev
@@ -1121,7 +1121,6 @@ def get_cast_buffer(offload_stream, device, size, ref):
             synchronize()
             del STREAM_CAST_BUFFERS[offload_stream]
             del cast_buffer
-            #FIXME: This doesn't work in Aimdo because mempool cant clear cache
             soft_empty_cache()
         with wf_context:
             cast_buffer = torch.empty((size), dtype=torch.int8, device=device)

comfy/utils.py

@@ -1154,7 +1154,7 @@ def tiled_scale(samples, function, tile_x=64, tile_y=64, overlap = 8, upscale_am
     return tiled_scale_multidim(samples, function, (tile_y, tile_x), overlap=overlap, upscale_amount=upscale_amount, out_channels=out_channels, output_device=output_device, pbar=pbar)
 
 def model_trange(*args, **kwargs):
-    if comfy.memory_management.aimdo_allocator is None:
+    if not comfy.memory_management.aimdo_enabled:
         return trange(*args, **kwargs)
 
     pbar = trange(*args, **kwargs, smoothing=1.0)

cuda_malloc.py

@@ -1,10 +1,8 @@
 import os
 import importlib.util
-from comfy.cli_args import args, PerformanceFeature, enables_dynamic_vram
+from comfy.cli_args import args, PerformanceFeature
 import subprocess
 
-import comfy_aimdo.control
-
 #Can't use pytorch to get the GPU names because the cuda malloc has to be set before the first import.
 def get_gpu_names():
     if os.name == 'nt':
@@ -87,10 +85,6 @@ def cuda_malloc_supported():
     except:
         pass
 
-if enables_dynamic_vram() and comfy_aimdo.control.init():
-    args.cuda_malloc = False
-    os.environ['PYTORCH_CUDA_ALLOC_CONF'] = ""
-
 if args.disable_cuda_malloc:
     args.cuda_malloc = False
 

execution.py

@@ -9,7 +9,6 @@
 from enum import Enum
 from typing import List, Literal, NamedTuple, Optional, Union
 import asyncio
-from contextlib import nullcontext
 
 import torch
 
@@ -521,19 +520,14 @@ def pre_execute_cb(call_index):
                 # TODO - How to handle this with async functions without contextvars (which requires Python 3.12)?
                 GraphBuilder.set_default_prefix(unique_id, call_index, 0)
 
-            #Do comfy_aimdo mempool chunking here on the per-node level. Multi-model workflows
-            #will cause all sorts of incompatible memory shapes to fragment the pytorch alloc
-            #that we just want to cull out each model run.
-            allocator = comfy.memory_management.aimdo_allocator
-            with nullcontext() if allocator is None else torch.cuda.use_mem_pool(torch.cuda.MemPool(allocator.allocator())):
-                try:
-                    output_data, output_ui, has_subgraph, has_pending_tasks = await get_output_data(prompt_id, unique_id, obj, input_data_all, execution_block_cb=execution_block_cb, pre_execute_cb=pre_execute_cb, v3_data=v3_data)
-                finally:
-                    if allocator is not None:
-                        if args.verbose == "DEBUG":
-                            comfy_aimdo.model_vbar.vbars_analyze()
-                        comfy.model_management.reset_cast_buffers()
-                        comfy_aimdo.model_vbar.vbars_reset_watermark_limits()
+            try:
+                output_data, output_ui, has_subgraph, has_pending_tasks = await get_output_data(prompt_id, unique_id, obj, input_data_all, execution_block_cb=execution_block_cb, pre_execute_cb=pre_execute_cb, v3_data=v3_data)
+            finally:
+                if comfy.memory_management.aimdo_enabled:
+                    if args.verbose == "DEBUG":
+                        comfy_aimdo.control.analyze()
+                    comfy.model_management.reset_cast_buffers()
+                    comfy_aimdo.model_vbar.vbars_reset_watermark_limits()
 
             if has_pending_tasks:
                 pending_async_nodes[unique_id] = output_data

main.py

@@ -173,6 +173,10 @@ def execute_script(script_path):
 if 'torch' in sys.modules:
     logging.warning("WARNING: Potential Error in code: Torch already imported, torch should never be imported before this point.")
 
+import comfy_aimdo.control
+
+if enables_dynamic_vram():
+    comfy_aimdo.control.init()
 
 import comfy.utils
 
@@ -188,13 +192,9 @@ def execute_script(script_path):
 import comfy.memory_management
 import comfy.model_patcher
 
-import comfy_aimdo.control
-import comfy_aimdo.torch
-
 if enables_dynamic_vram():
     if comfy.model_management.torch_version_numeric < (2, 8):
         logging.warning("Unsupported Pytorch detected. DynamicVRAM support requires Pytorch version 2.8 or later. Falling back to legacy ModelPatcher. VRAM estimates may be unreliable especially on Windows")
-        comfy.memory_management.aimdo_allocator = None
     elif comfy_aimdo.control.init_device(comfy.model_management.get_torch_device().index):
         if args.verbose == 'DEBUG':
             comfy_aimdo.control.set_log_debug()
@@ -208,11 +208,10 @@ def execute_script(script_path):
             comfy_aimdo.control.set_log_info()
 
         comfy.model_patcher.CoreModelPatcher = comfy.model_patcher.ModelPatcherDynamic
-        comfy.memory_management.aimdo_allocator = comfy_aimdo.torch.get_torch_allocator()
+        comfy.memory_management.aimdo_enabled = True
         logging.info("DynamicVRAM support detected and enabled")
     else:
         logging.warning("No working comfy-aimdo install detected. DynamicVRAM support disabled. Falling back to legacy ModelPatcher. VRAM estimates may be unreliable especially on Windows")
-        comfy.memory_management.aimdo_allocator = None
 
 
 def cuda_malloc_warning():

requirements.txt

@@ -22,7 +22,7 @@ alembic
 SQLAlchemy
 av>=14.2.0
 comfy-kitchen>=0.2.7
-comfy-aimdo>=0.1.8
+comfy-aimdo>=0.2.0
 requests
 
 #non essential dependencies: