diff --git a/README.md b/README.md index 786a14166..61036bd81 100644 --- a/README.md +++ b/README.md @@ -382,11 +382,7 @@ For AMD 7600 and maybe other RDNA3 cards: ```HSA_OVERRIDE_GFX_VERSION=11.0.0 pyt ### AMD ROCm Tips -You can enable experimental memory efficient attention on recent pytorch in ComfyUI on some AMD GPUs using this command, it should already be enabled by default on RDNA3. If this improves speed for you on latest pytorch on your GPU please report it so that I can enable it by default. - -```TORCH_ROCM_AOTRITON_ENABLE_EXPERIMENTAL=1 python main.py --use-pytorch-cross-attention``` - -You can also try setting this env variable `PYTORCH_TUNABLEOP_ENABLED=1` which might speed things up at the cost of a very slow initial run. +You can try setting this env variable `PYTORCH_TUNABLEOP_ENABLED=1` which might speed things up at the cost of a very slow initial run. # Notes diff --git a/comfy/cli_args.py b/comfy/cli_args.py index e7ee0d5eb..e3099a230 100644 --- a/comfy/cli_args.py +++ b/comfy/cli_args.py @@ -145,6 +145,7 @@ vram_group.add_argument("--novram", action="store_true", help="When lowvram isn' vram_group.add_argument("--cpu", action="store_true", help="To use the CPU for everything (slow).") parser.add_argument("--reserve-vram", type=float, default=None, help="Set the amount of vram in GB you want to reserve for use by your OS/other software. By default some amount is reserved depending on your OS.") +parser.add_argument("--vram-headroom", type=float, default=0, help="Set the amount of vram in GB for DynamicVRAM to maintain as extra headroom above default. ComfyUI will try and keep this much VRAM completely free and unused, even counting VRAM from other apps.") parser.add_argument("--async-offload", nargs='?', const=2, type=int, default=None, metavar="NUM_STREAMS", help="Use async weight offloading. An optional argument controls the amount of offload streams. Default is 2. Enabled by default on Nvidia.") parser.add_argument("--disable-async-offload", action="store_true", help="Disable async weight offloading.") diff --git a/comfy_api/latest/_input_impl/video_types.py b/comfy_api/latest/_input_impl/video_types.py index 92a1298c0..6c69256ab 100644 --- a/comfy_api/latest/_input_impl/video_types.py +++ b/comfy_api/latest/_input_impl/video_types.py @@ -325,21 +325,25 @@ class VideoFromFile(VideoInput): checked_alpha = True # Fix non-deterministic video decode when the video width is not a multiple of 32 - # For non-yuvj pixel formats (all H.264/H.265 video) + # For non-yuvj pixel formats: most H.264/H.265 video and static images (e.g. lossy WebP via LoadImage) + # Pad both axes to a multiple of 32 and smear the border so the alignment padding never bleeds into the cropped edges if image_format in ('gbrpf32le', 'gbrapf32le') and frame.width % 32 != 0: if align_graph is None: pad_w = ((frame.width + 31) // 32) * 32 + pad_h = ((frame.height + 31) // 32) * 32 g = av.filter.Graph() g_src = g.add_buffer(width=frame.width, height=frame.height, format=frame.format.name, time_base=video_stream.time_base) - g_pad = g.add('pad', f'{pad_w}:{frame.height}:0:0') + g_pad = g.add('pad', f'{pad_w}:{pad_h}:0:0') + g_fill = g.add('fillborders', f'left=0:right={pad_w - frame.width}:top=0:bottom={pad_h - frame.height}:mode=smear') g_sink = g.add('buffersink') g_src.link_to(g_pad) - g_pad.link_to(g_sink) + g_pad.link_to(g_fill) + g_fill.link_to(g_sink) g.configure() align_graph = (g, g_src, g_sink) align_graph[1].push(frame) - img = np.ascontiguousarray(align_graph[2].pull().to_ndarray(format=image_format)[:, :frame.width]) + img = np.ascontiguousarray(align_graph[2].pull().to_ndarray(format=image_format)[:frame.height, :frame.width]) else: img = frame.to_ndarray(format=image_format) if frame.rotation != 0: diff --git a/comfy_api_nodes/nodes_sonilo.py b/comfy_api_nodes/nodes_sonilo.py index 24a9a0b06..d146f63ea 100644 --- a/comfy_api_nodes/nodes_sonilo.py +++ b/comfy_api_nodes/nodes_sonilo.py @@ -111,11 +111,10 @@ class SoniloTextToMusic(IO.ComfyNode): ), IO.Int.Input( "duration", - default=0, - min=0, + default=30, + min=1, max=360, - tooltip="Target duration in seconds. Set to 0 to let the model " - "infer the duration from the prompt. Maximum: 6 minutes.", + tooltip="Target duration in seconds. Maximum: 6 minutes.", ), IO.Int.Input( "seed", @@ -150,14 +149,13 @@ class SoniloTextToMusic(IO.ComfyNode): async def execute( cls, prompt: str, - duration: int = 0, + duration: int = 1, seed: int = 0, ) -> IO.NodeOutput: - validate_string(prompt, strip_whitespace=True, min_length=1) + validate_string(prompt, strip_whitespace=True, min_length=1, max_length=1000) form = aiohttp.FormData() form.add_field("prompt", prompt) - if duration > 0: - form.add_field("duration", str(duration)) + form.add_field("duration", str(duration)) audio_bytes = await _stream_sonilo_music( cls, ApiEndpoint(path="/proxy/sonilo/t2m/generate", method="POST"), diff --git a/main.py b/main.py index 0ad660376..82f2bab64 100644 --- a/main.py +++ b/main.py @@ -55,7 +55,11 @@ if __name__ == "__main__" and args.debug_hang: import comfy_aimdo.control if enables_dynamic_vram(): - comfy_aimdo.control.init() + try: + comfy_aimdo.control.init(simple_vram_headroom=None if args.reserve_vram is None else int(args.reserve_vram * 1024 ** 3)) + except TypeError: + # comfy-aimdo 0.4.9 protocol. + comfy_aimdo.control.init() if os.name == "nt": os.environ['MIMALLOC_PURGE_DELAY'] = '0' @@ -231,23 +235,30 @@ import comfy.model_patcher if args.enable_dynamic_vram or (enables_dynamic_vram() and comfy.model_management.is_nvidia() and not comfy.model_management.is_wsl()): if (not args.enable_dynamic_vram) and (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") - elif comfy_aimdo.control.init_devices(d.index for d in comfy.model_management.get_all_torch_devices()): - if args.verbose == 'DEBUG': - comfy_aimdo.control.set_log_debug() - elif args.verbose == 'CRITICAL': - comfy_aimdo.control.set_log_critical() - elif args.verbose == 'ERROR': - comfy_aimdo.control.set_log_error() - elif args.verbose == 'WARNING': - comfy_aimdo.control.set_log_warning() - else: #INFO - comfy_aimdo.control.set_log_info() - - comfy.model_patcher.CoreModelPatcher = comfy.model_patcher.ModelPatcherDynamic - 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") + try: + aimdo_initialized = comfy_aimdo.control.init_devices((d.index, int(args.vram_headroom * 1024 ** 3)) for d in comfy.model_management.get_all_torch_devices()) + except TypeError: + # comfy-aimdo 0.4.9 protocol. + aimdo_initialized = comfy_aimdo.control.init_devices(d.index for d in comfy.model_management.get_all_torch_devices()) + + if aimdo_initialized: + if args.verbose == 'DEBUG': + comfy_aimdo.control.set_log_debug() + elif args.verbose == 'CRITICAL': + comfy_aimdo.control.set_log_critical() + elif args.verbose == 'ERROR': + comfy_aimdo.control.set_log_error() + elif args.verbose == 'WARNING': + comfy_aimdo.control.set_log_warning() + else: #INFO + comfy_aimdo.control.set_log_info() + + comfy.model_patcher.CoreModelPatcher = comfy.model_patcher.ModelPatcherDynamic + 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") def cuda_malloc_warning(): diff --git a/requirements.txt b/requirements.txt index a49d968af..392709e64 100644 --- a/requirements.txt +++ b/requirements.txt @@ -1,6 +1,6 @@ comfyui-frontend-package==1.45.15 -comfyui-workflow-templates==0.9.98 -comfyui-embedded-docs==0.5.3 +comfyui-workflow-templates==0.10.0 +comfyui-embedded-docs==0.5.4 torch torchsde torchvision @@ -23,7 +23,7 @@ SQLAlchemy>=2.0.0 filelock av>=16.0.0 comfy-kitchen==0.2.10 -comfy-aimdo==0.4.9 +comfy-aimdo==0.4.10 requests simpleeval>=1.0.0 blake3