Merge 562e355f97 into c05a08ae66

* mp: fix full dynamic unloading

This was not unloading dynamic models when requesting a full unload via
the unpatch() code path.

This was ok, i your workflow was all dynamic models but fails with big
VRAM leaks if you need to fully unload something for a regular ModelPatcher

It also fices the "unload models" button.

* mm: load models outside of Aimdo Mempool

In dynamic_vram mode, escape the Aimdo mempool and load into the regular
mempool. Use a dummy thread to do it.

comfy/model_management.py

@@ -19,7 +19,8 @@
 import psutil
 import logging
 from enum import Enum
-from comfy.cli_args import args, PerformanceFeature
+from comfy.cli_args import args, PerformanceFeature, enables_dynamic_vram
+import threading
 import torch
 import sys
 import platform
@@ -650,7 +651,7 @@ def free_memory(memory_required, device, keep_loaded=[], for_dynamic=False, ram_
                 soft_empty_cache()
     return unloaded_models
 
-def load_models_gpu(models, memory_required=0, force_patch_weights=False, minimum_memory_required=None, force_full_load=False):
+def load_models_gpu_orig(models, memory_required=0, force_patch_weights=False, minimum_memory_required=None, force_full_load=False):
     cleanup_models_gc()
     global vram_state
 
@@ -746,6 +747,26 @@ def load_models_gpu(models, memory_required=0, force_patch_weights=False, minimu
         current_loaded_models.insert(0, loaded_model)
     return
 
+def load_models_gpu_thread(models, memory_required, force_patch_weights, minimum_memory_required, force_full_load):
+    with torch.inference_mode():
+        load_models_gpu_orig(models, memory_required, force_patch_weights, minimum_memory_required, force_full_load)
+        soft_empty_cache()
+
+def load_models_gpu(models, memory_required=0, force_patch_weights=False, minimum_memory_required=None, force_full_load=False):
+    #Deliberately load models outside of the Aimdo mempool so they can be retained accross
+    #nodes. Use a dummy thread to do it as pytorch documents that mempool contexts are
+    #thread local. So exploit that to escape context
+    if enables_dynamic_vram():
+        t = threading.Thread(
+            target=load_models_gpu_thread,
+            args=(models, memory_required, force_patch_weights, minimum_memory_required, force_full_load)
+        )
+        t.start()
+        t.join()
+    else:
+        load_models_gpu_orig(models, memory_required=memory_required, force_patch_weights=force_patch_weights,
+                             minimum_memory_required=minimum_memory_required, force_full_load=force_full_load)
+
 def load_model_gpu(model):
     return load_models_gpu([model])
 
@@ -1112,11 +1133,11 @@ def get_cast_buffer(offload_stream, device, size, ref):
             return None
         if cast_buffer is not None and cast_buffer.numel() > 50 * (1024 ** 2):
             #I want my wrongly sized 50MB+ of VRAM back from the caching allocator right now
-            torch.cuda.synchronize()
+            synchronize()
             del STREAM_CAST_BUFFERS[offload_stream]
             del cast_buffer
             #FIXME: This doesn't work in Aimdo because mempool cant clear cache
-            torch.cuda.empty_cache()
+            soft_empty_cache()
         with wf_context:
             cast_buffer = torch.empty((size), dtype=torch.int8, device=device)
             STREAM_CAST_BUFFERS[offload_stream] = cast_buffer
@@ -1132,9 +1153,7 @@ def reset_cast_buffers():
     for offload_stream in STREAM_CAST_BUFFERS:
         offload_stream.synchronize()
     STREAM_CAST_BUFFERS.clear()
-    if comfy.memory_management.aimdo_allocator is None:
-        #Pytorch 2.7 and earlier crashes if you try and empty_cache when mempools exist
-        torch.cuda.empty_cache()
+    soft_empty_cache()
 
 def get_offload_stream(device):
     stream_counter = stream_counters.get(device, 0)
@@ -1284,7 +1303,7 @@ def discard_cuda_async_error():
         a = torch.tensor([1], dtype=torch.uint8, device=get_torch_device())
         b = torch.tensor([1], dtype=torch.uint8, device=get_torch_device())
         _ = a + b
-        torch.cuda.synchronize()
+        synchronize()
     except torch.AcceleratorError:
         #Dump it! We already know about it from the synchronous return
         pass
@@ -1688,6 +1707,12 @@ def lora_compute_dtype(device):
     LORA_COMPUTE_DTYPES[device] = dtype
     return dtype
 
+def synchronize():
+    if is_intel_xpu():
+        torch.xpu.synchronize()
+    elif torch.cuda.is_available():
+        torch.cuda.synchronize()
+
 def soft_empty_cache(force=False):
     global cpu_state
     if cpu_state == CPUState.MPS:
@@ -1713,9 +1738,6 @@ def debug_memory_summary():
         return torch.cuda.memory.memory_summary()
     return ""
 
-#TODO: might be cleaner to put this somewhere else
-import threading
-
 class InterruptProcessingException(Exception):
     pass
 

comfy/model_patcher.py

@@ -1597,7 +1597,7 @@ class ModelPatcherDynamic(ModelPatcher):
 
         if unpatch_weights:
             self.partially_unload_ram(1e32)
-            self.partially_unload(None)
+            self.partially_unload(None, 1e32)
 
     def partially_load(self, device_to, extra_memory=0, force_patch_weights=False):
         assert not force_patch_weights #See above

comfy_extras/nodes_resolution.py

@@ -0,0 +1,82 @@
+from __future__ import annotations
+import math
+from enum import Enum
+from typing_extensions import override
+from comfy_api.latest import ComfyExtension, io
+
+
+class AspectRatio(str, Enum):
+    SQUARE = "1:1 (Square)"
+    PHOTO_H = "3:2 (Photo)"
+    STANDARD_H = "4:3 (Standard)"
+    WIDESCREEN_H = "16:9 (Widescreen)"
+    ULTRAWIDE_H = "21:9 (Ultrawide)"
+    PHOTO_V = "2:3 (Portrait Photo)"
+    STANDARD_V = "3:4 (Portrait Standard)"
+    WIDESCREEN_V = "9:16 (Portrait Widescreen)"
+
+
+ASPECT_RATIOS: dict[str, tuple[int, int]] = {
+    "1:1 (Square)": (1, 1),
+    "3:2 (Photo)": (3, 2),
+    "4:3 (Standard)": (4, 3),
+    "16:9 (Widescreen)": (16, 9),
+    "21:9 (Ultrawide)": (21, 9),
+    "2:3 (Portrait Photo)": (2, 3),
+    "3:4 (Portrait Standard)": (3, 4),
+    "9:16 (Portrait Widescreen)": (9, 16),
+}
+
+
+class ResolutionSelector(io.ComfyNode):
+    """Calculate width and height from aspect ratio and megapixel target."""
+
+    @classmethod
+    def define_schema(cls):
+        return io.Schema(
+            node_id="ResolutionSelector",
+            display_name="Resolution Selector",
+            category="utils",
+            description="Calculate width and height from aspect ratio and megapixel target. Useful for setting up Empty Latent Image dimensions.",
+            inputs=[
+                io.Combo.Input(
+                    "aspect_ratio",
+                    options=AspectRatio,
+                    default=AspectRatio.SQUARE,
+                    tooltip="The aspect ratio for the output dimensions.",
+                ),
+                io.Float.Input(
+                    "megapixels",
+                    default=1.0,
+                    min=0.1,
+                    max=16.0,
+                    step=0.1,
+                    tooltip="Target total megapixels. 1.0 MP ≈ 1024×1024 for square.",
+                ),
+            ],
+            outputs=[
+                io.Int.Output("width", tooltip="Calculated width in pixels (multiple of 8)."),
+                io.Int.Output("height", tooltip="Calculated height in pixels (multiple of 8)."),
+            ],
+        )
+
+    @classmethod
+    def execute(cls, aspect_ratio: str, megapixels: float) -> io.NodeOutput:
+        w_ratio, h_ratio = ASPECT_RATIOS[aspect_ratio]
+        total_pixels = megapixels * 1024 * 1024
+        scale = math.sqrt(total_pixels / (w_ratio * h_ratio))
+        width = round(w_ratio * scale / 8) * 8
+        height = round(h_ratio * scale / 8) * 8
+        return io.NodeOutput(width, height)
+
+
+class ResolutionExtension(ComfyExtension):
+    @override
+    async def get_node_list(self) -> list[type[io.ComfyNode]]:
+        return [
+            ResolutionSelector,
+        ]
+
+
+async def comfy_entrypoint() -> ResolutionExtension:
+    return ResolutionExtension()

nodes.py

@@ -2427,6 +2427,7 @@ async def init_builtin_extra_nodes():
         "nodes_audio_encoder.py",
         "nodes_rope.py",
         "nodes_logic.py",
+        "nodes_resolution.py",
         "nodes_nop.py",
         "nodes_kandinsky5.py",
         "nodes_wanmove.py",