search

siliconflow / onediff

OneDiff: An out-of-the-box acceleration library for diffusion models.
https://github.com/siliconflow/onediff/wiki
Apache License 2.0
1.58k stars 97 forks source link

[Bug] ComfyUI中使用 Model Graph Saver 节点不能保存正确的graph #1063

Closed LingXuanYin closed 4 days ago

LingXuanYin commented 1 month ago

Your current environment information

Collecting environment information... PyTorch version: 2.1.0+cu121 Is debug build: False CUDA used to build PyTorch: 12.1 ROCM used to build PyTorch: N/A

OneFlow version: path: ['/root/miniconda3/lib/python3.10/site-packages/oneflow'], version: 0.9.1.dev20240701+cu121, git_commit: ec7b682, cmake_build_type: Release, rdma: True, mlir: True, enterprise: False Nexfort version: none OneDiff version: unknown version OneDiffX version: none

OS: Ubuntu 22.04.1 LTS (x86_64) GCC version: (Ubuntu 11.3.0-1ubuntu1~22.04) 11.3.0 Clang version: Could not collect CMake version: version 3.27.7 Libc version: glibc-2.35

Python version: 3.10.8 (main, Nov 24 2022, 14:13:03) [GCC 11.2.0] (64-bit runtime) Python platform: Linux-5.15.0-83-generic-x86_64-with-glibc2.35 Is CUDA available: True CUDA runtime version: 11.8.89 CUDA_MODULE_LOADING set to: LAZY GPU models and configuration: GPU 0: NVIDIA GeForce RTX 4090 Nvidia driver version: 535.104.05 cuDNN version: Probably one of the following: /usr/lib/x86_64-linux-gnu/libcudnn.so.8.6.0 /usr/lib/x86_64-linux-gnu/libcudnn_adv_infer.so.8.6.0 /usr/lib/x86_64-linux-gnu/libcudnn_adv_train.so.8.6.0 /usr/lib/x86_64-linux-gnu/libcudnn_cnn_infer.so.8.6.0 /usr/lib/x86_64-linux-gnu/libcudnn_cnn_train.so.8.6.0 /usr/lib/x86_64-linux-gnu/libcudnn_ops_infer.so.8.6.0 /usr/lib/x86_64-linux-gnu/libcudnn_ops_train.so.8.6.0 HIP runtime version: N/A MIOpen runtime version: N/A Is XNNPACK available: True

CPU: Architecture: x86_64 CPU op-mode(s): 32-bit, 64-bit Address sizes: 46 bits physical, 57 bits virtual Byte Order: Little Endian CPU(s): 128 On-line CPU(s) list: 0-127 Vendor ID: GenuineIntel Model name: Intel(R) Xeon(R) Platinum 8352V CPU @ 2.10GHz CPU family: 6 Model: 106 Thread(s) per core: 2 Core(s) per socket: 32 Socket(s): 2 Stepping: 6 CPU max MHz: 3500.0000 CPU min MHz: 800.0000 BogoMIPS: 4000.00 Flags: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx pdpe1gb rdtscp lm constant_tsc art arch_perfmon pebs bts rep_good nopl xtopology nonstop_tsc cpuid aperfmperf pni pclmulqdq dtes64 monitor ds_cpl vmx smx est tm2 ssse3 sdbg fma cx16 xtpr pdcm pcid dca sse4_1 sse4_2 x2apic movbe popcnt tsc_deadline_timer aes xsave avx f16c rdrand lahf_lm abm 3dnowprefetch cpuid_fault epb cat_l3 invpcid_single ssbd mba ibrs ibpb stibp ibrs_enhanced tpr_shadow vnmi flexpriority ept vpid ept_ad fsgsbase tsc_adjust bmi1 avx2 smep bmi2 erms invpcid cqm rdt_a avx512f avx512dq rdseed adx smap avx512ifma clflushopt clwb intel_pt avx512cd sha_ni avx512bw avx512vl xsaveopt xsavec xgetbv1 xsaves cqm_llc cqm_occup_llc cqm_mbm_total cqm_mbm_local split_lock_detect wbnoinvd dtherm ida arat pln pts avx512vbmi umip pku ospke avx512_vbmi2 gfni vaes vpclmulqdq avx512_vnni avx512_bitalg tme avx512_vpopcntdq la57 rdpid fsrm md_clear pconfig flush_l1d arch_capabilities Virtualization: VT-x L1d cache: 3 MiB (64 instances) L1i cache: 2 MiB (64 instances) L2 cache: 80 MiB (64 instances) L3 cache: 108 MiB (2 instances) NUMA node(s): 2 NUMA node0 CPU(s): 0-31,64-95 NUMA node1 CPU(s): 32-63,96-127 Vulnerability Gather data sampling: Mitigation; Microcode Vulnerability Itlb multihit: Not affected Vulnerability L1tf: Not affected Vulnerability Mds: Not affected Vulnerability Meltdown: Not affected Vulnerability Mmio stale data: Mitigation; Clear CPU buffers; SMT vulnerable Vulnerability Retbleed: Not affected Vulnerability Spec store bypass: Mitigation; Speculative Store Bypass disabled via prctl and seccomp Vulnerability Spectre v1: Mitigation; usercopy/swapgs barriers and __user pointer sanitization Vulnerability Spectre v2: Mitigation; Enhanced IBRS, IBPB conditional, RSB filling, PBRSB-eIBRS SW sequence Vulnerability Srbds: Not affected Vulnerability Tsx async abort: Not affected

Versions of relevant libraries: [pip3] diffusers==0.27.2 [pip3] numpy==1.24.4 [pip3] onnx==1.16.1 [pip3] onnxruntime==1.18.0 [pip3] onnxruntime-gpu==1.18.0 [pip3] open-clip-torch==2.23.0 [pip3] torch==2.1.0 [pip3] torchaudio==2.1.0+cu121 [pip3] torchsde==0.2.6 [pip3] torchvision==0.16.0+cu121 [pip3] transformers==4.42.3 [pip3] triton==2.1.0 [conda] numpy 1.24.4 pypi_0 pypi [conda] open-clip-torch 2.23.0 pypi_0 pypi [conda] torch 2.1.0 pypi_0 pypi [conda] torchaudio 2.1.0+cu121 pypi_0 pypi [conda] torchsde 0.2.6 pypi_0 pypi [conda] torchvision 0.16.0+cu121 pypi_0 pypi [conda] triton 2.1.0 pypi_0 pypi

🐛 Describe the bug

  1. 在ComfyUI中使用Load Checkpoint - OneDiff节点加速一个带IPA的自定义的工作流，运行，发现IPA无法生效。删除graph文件并重启comfyUI，更换工作流为ipadapter_advanced.json并使用上述onediff节点进行加速，运行，发现IPA生效。对比两次生成的graph文件，IPA生效时生成的graph文件大小为不生效时的两倍，但其文件名一致。

  2. 删除graph文件并重启comfyUI，使用onediff加速的工作流ipadapter_advanced.json，运行，IPA生效，删除graph文件（此graph文件为较大的），不重启comfyUI，使用onediff加速上述我自定义的工作流，在KSampler节点后添加了节点Model Graph Saver，其中输入samplers来自KSampler节点，输入model与KSampler节点的输入相同，运行，报错如下

    
Warning: ModelGraphSaver will be deleted. Please use it with caution.
Graph saving...
!!! Exception during processing!!! Can't pickle <class 'onediff.infer_compiler.backends.oneflow.transform.builtin_transform.<class 'register_comfy.CrossAttentionPatch.Attn2Replace'>'>: attribute lookup <class 'register_comfy.CrossAttentionPatch.Attn2Replace'> on onediff.infer_compiler.backends.oneflow.transform.builtin_transform failed
Traceback (most recent call last):
File "/root/ComfyUI/execution.py", line 151, in recursive_execute
output_data, output_ui = get_output_data(obj, input_data_all)
File "/root/ComfyUI/execution.py", line 81, in get_output_data
return_values = map_node_over_list(obj, input_data_all, obj.FUNCTION, allow_interrupt=True)
File "/root/ComfyUI/execution.py", line 74, in map_node_over_list
results.append(getattr(obj, func)(**slice_dict(input_data_all, i)))
File "/root/ComfyUI/custom_nodes/onediff_comfy_nodes/extras_nodes/nodes_oneflow_booster.py", line 524, in save_graph
save_graph(diffusion_model, filename_prefix, "cuda", subfolder="unet")
File "/root/ComfyUI/custom_nodes/onediff_comfy_nodes/modules/oneflow/utils/__init__.py", line 64, in save_graph
deploy_module.save_graph(graph_filename)
File "/root/onediff/src/onediff/infer_compiler/backends/oneflow/deployable_module.py", line 212, in save_graph
self.get_graph().save_graph(file_path, process_state_dict=process_state_dict)
File "/root/onediff/src/onediff/infer_compiler/backends/oneflow/utils/cost_util.py", line 49, in clocked
return func(*args, **kwargs)
File "/root/onediff/src/onediff/infer_compiler/backends/oneflow/graph.py", line 65, in save_graph
flow.save(state_dict, file_path)
File "/root/miniconda3/lib/python3.10/site-packages/oneflow/framework/check_point_v2.py", line 720, in save
pickled_bytes = pickle.dumps(obj)
_pickle.PicklingError: Can't pickle <class 'onediff.infer_compiler.backends.oneflow.transform.builtin_transform.<class 'register_comfy.CrossAttentionPatch.Attn2Replace'>'>: attribute lookup <class 'register_comfy.CrossAttentionPatch.Attn2Replace'> on onediff.infer_compiler.backends.oneflow.transform.builtin_transform failed

Prompt executed in 89.25 seconds

注意：此时依然生成了一个graph文件，但其大小是上述较大的graph文件的一半，且此次运行结果显示IPA未生效

2. 删除graph并重启comfyUI，重复` 1. ` 但不加入节点` Model Graph Saver `且不删除较大的graph文件，运行，IPA生效，再次运行，IPA生效

3.  `1.` 和`2.` 的所有控制台输出。上述所有graph文件具有相同的文件名`SDXL_7ecb88_d49b38b9.graph`

got prompt model_type EPS Using xformers attention in VAE Using xformers attention in VAE Cache lookup: Key='b303a9a0-6293-4280-86af-aaf69abe240c', Cached Model Type='<class 'NoneType'>' Requested to load CLIPVisionModelProjection Loading 1 new model Requested to load SDXLClipModel Loading 1 new model Requested to load SDXL Loading 1 new model 100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████| 30/30 [01:08<00:00, 2.29s/it] Requested to load AutoencoderKL Loading 1 new model Prompt executed in 113.55 seconds

got prompt model_type EPS Using xformers attention in VAE Using xformers attention in VAE Cache lookup: Key='fdb5fad2-2eef-4905-822f-82d40b08e6fc', Cached Model Type='<class 'NoneType'>' Requested to load CLIPVisionModelProjection Loading 1 new model CLIP: [1girl, Seraphina Noir,

Brown almond-shaped eyes, Long black flowing hair, Fair skin, Blue strapless dress with earrings.] Requested to load SDXLClipModel Loading 1 new model Requested to load SDXL Loading 1 new model 100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████| 24/24 [00:48<00:00, 2.04s/it] Requested to load AutoencoderKL Loading 1 new model WAS Node Suite: Image file saved to: /root/ComfyUI/output/ComfyUI_0001.jpeg Warning: ModelGraphSaver will be deleted. Please use it with caution. Graph saving... !!! Exception during processing!!! Can't pickle <class 'onediff.infer_compiler.backends.oneflow.transform.builtin_transform.<class 'register_comfy.CrossAttentionPatch.Attn2Replace'>'>: attribute lookup <class 'register_comfy.CrossAttentionPatch.Attn2Replace'> on onediff.infer_compiler.backends.oneflow.transform.builtin_transform failed Traceback (most recent call last): File "/root/ComfyUI/execution.py", line 151, in recursive_execute output_data, output_ui = get_output_data(obj, input_data_all) File "/root/ComfyUI/execution.py", line 81, in get_output_data return_values = map_node_over_list(obj, input_data_all, obj.FUNCTION, allow_interrupt=True) File "/root/ComfyUI/execution.py", line 74, in map_node_over_list results.append(getattr(obj, func)(*slice_dict(input_data_all, i))) File "/root/ComfyUI/custom_nodes/onediff_comfy_nodes/extras_nodes/nodes_oneflow_booster.py", line 524, in save_graph save_graph(diffusion_model, filename_prefix, "cuda", subfolder="unet") File "/root/ComfyUI/custom_nodes/onediff_comfy_nodes/modules/oneflow/utils/init.py", line 64, in save_graph deploy_module.save_graph(graph_filename) File "/root/onediff/src/onediff/infer_compiler/backends/oneflow/deployable_module.py", line 212, in save_graph self.get_graph().save_graph(file_path, process_state_dict=process_state_dict) File "/root/onediff/src/onediff/infer_compiler/backends/oneflow/utils/cost_util.py", line 49, in clocked return func(args, **kwargs) File "/root/onediff/src/onediff/infer_compiler/backends/oneflow/graph.py", line 65, in save_graph flow.save(state_dict, file_path) File "/root/miniconda3/lib/python3.10/site-packages/oneflow/framework/check_point_v2.py", line 720, in save pickled_bytes = pickle.dumps(obj) _pickle.PicklingError: Can't pickle <class 'onediff.infer_compiler.backends.oneflow.transform.builtin_transform.<class 'register_comfy.CrossAttentionPatch.Attn2Replace'>'>: attribute lookup <class 'register_comfy.CrossAttentionPatch.Attn2Replace'> on onediff.infer_compiler.backends.oneflow.transform.builtin_transform failed

Prompt executed in 93.72 seconds

got prompt model_type EPS Using xformers attention in VAE Using xformers attention in VAE Cache lookup: Key='f9098578-f38c-4fa8-b5fe-47cad9a44cd5', Cached Model Type='<class 'NoneType'>' Requested to load CLIPVisionModelProjection Loading 1 new model Requested to load SDXLClipModel Loading 1 new model Requested to load SDXL Loading 1 new model 100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████| 30/30 [01:01<00:00, 2.06s/it] Requested to load AutoencoderKL Loading 1 new model Prompt executed in 100.32 seconds

got prompt model_type EPS Using xformers attention in VAE Using xformers attention in VAE Cache lookup: Key='c75095ff-76ae-47a7-854d-51cd394529a6', Cached Model Type='<class 'NoneType'>' Requested to load CLIPVisionModelProjection Loading 1 new model CLIP: [1girl, Seraphina Noir,

Brown almond-shaped eyes, Long black flowing hair, Fair skin, Blue strapless dress with earrings.] Requested to load SDXLClipModel Loading 1 new model Requested to load SDXL Loading 1 new model 100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████| 24/24 [00:04<00:00, 5.57it/s] Requested to load AutoencoderKL Loading 1 new model WAS Node Suite: Image file saved to: /root/ComfyUI/output/ComfyUI_0001.jpeg Prompt executed in 43.03 seconds

got prompt CLIP: [1girl, Seraphina Noir,

Brown almond-shaped eyes, Long black flowing hair, Fair skin, Blue strapless dress with earrings.] 100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████| 24/24 [00:02<00:00, 11.00it/s] WAS Node Suite: Image file saved to: /root/ComfyUI/output/ComfyUI_0002.jpeg Prompt executed in 3.22 seconds

ccssu commented 1 month ago

hello 感谢反馈， @LingXuanYin 请 使用 Load Checkpoint - OneDiff 节点，这个节点会自动加载和保存图文件， 不要使用 Model Graph Saver 节点，这个文件已经没有维护，会导致保存错误的图文件，

关于编译缓存参考文档： 🔗 https://github.com/siliconflow/onediff/tree/main/onediff_comfy_nodes#compiler-cache

LingXuanYin commented 1 month ago

感谢您的耐心和回复 @ccssu ，我在第0步时，只使用 Load Checkpoint - OneDiff 节点加载了工作流使用的模型，但其运行结果显示工作流中带有的IPA节点并没有生效（并且保存的graph较生效时小一倍）。 重启comfyui（清除了硬盘和内存（显存）中的graph）后，我尝试使用了官方工作流ipadapter_advanced.json，其IPA生效并生成了一个较大的graph，使用我的工作流加载这个graph并运行，运行时间表明其进行了重新编译，同时结果显示IPA生效，但这时Load Checkpoint - OneDiff 节点并没有保存这个graph，因为当我重启comfyui（清除了内存（显存）中的graph，没有删除graph文件）后，再次运行，IPA生效，但其运行时间表明进行了重新编译，这时不重启再次运行，则运行时间明显缩短且效果与上一次一致

ccssu commented 1 month ago

我在第0步时，只使用 Load Checkpoint - OneDiff 节点加载了工作流使用的模型，但其运行结果显示工作流中带有的IPA节点并没有生效（并且保存的graph较生效时小一倍）。

没有生效这个比较奇怪，可以打开一下debug 模式看下运行日志吗 @LingXuanYin ： 

ONEDIFF_DEBUG=1 python main.py --gpu-only
LingXuanYin commented 1 month ago

@ccssu
启动时：

Total VRAM 24210 MB, total RAM 773288 MB
pytorch version: 2.1.0+cu121
xformers version: 0.0.22.post4
Set vram state to: NORMAL_VRAM
Device: cuda:0 NVIDIA GeForce RTX 4090 D : <module 'oneflow.cuda.get_allocator_backend(, )' from None>
Using xformers cross attention
DEBUG [2024-08-05 17:46:18] /root/onediff/src/onediff/infer_compiler/backends/oneflow/transform/manager.py:57 - Updated class proxies: len(class_proxy_dict)=1
{'comfy.ldm.modules.attention.CrossAttention': <class 'onediff_comfy_nodes.modules.oneflow.infer_compiler_registry.register_comfy.attention.CrossAttention'>}

DEBUG [2024-08-05 17:46:18] /root/onediff/src/onediff/infer_compiler/backends/oneflow/transform/manager.py:57 - Updated class proxies: len(class_proxy_dict)=1
{'comfy.ldm.modules.attention.attention_pytorch': <function attention_pytorch_oneflow at 0x7f8d73d1d1b0>}

DEBUG [2024-08-05 17:46:18] /root/onediff/src/onediff/infer_compiler/backends/oneflow/transform/manager.py:57 - Updated class proxies: len(class_proxy_dict)=1
{'comfy.ldm.modules.attention.SpatialTransformer': <class 'onediff_comfy_nodes.modules.oneflow.infer_compiler_registry.register_comfy.attention.SpatialTransformer'>}

DEBUG [2024-08-05 17:46:18] /root/onediff/src/onediff/infer_compiler/backends/oneflow/transform/manager.py:57 - Updated class proxies: len(class_proxy_dict)=1
{'comfy.ldm.modules.attention.SpatialVideoTransformer': <class 'onediff_comfy_nodes.modules.oneflow.infer_compiler_registry.register_comfy.attention.SpatialVideoTransformer'>}

DEBUG [2024-08-05 17:46:18] /root/onediff/src/onediff/infer_compiler/backends/oneflow/transform/manager.py:57 - Updated class proxies: len(class_proxy_dict)=1
{'comfy.ldm.modules.diffusionmodules.util.AlphaBlender': <class 'onediff_comfy_nodes.modules.oneflow.infer_compiler_registry.register_comfy.util.AlphaBlender'>}

DEBUG [2024-08-05 17:46:18] /root/onediff/src/onediff/infer_compiler/backends/oneflow/transform/manager.py:57 - Updated class proxies: len(class_proxy_dict)=1
{'comfy.ops.disable_weight_init.Linear': <class 'onediff_comfy_nodes.modules.oneflow.infer_compiler_registry.register_comfy.linear.Linear'>}

DEBUG [2024-08-05 17:46:18] /root/onediff/src/onediff/infer_compiler/backends/oneflow/transform/manager.py:57 - Updated class proxies: len(class_proxy_dict)=1
{'comfy.ldm.modules.diffusionmodules.model.AttnBlock': <class 'onediff_comfy_nodes.modules.oneflow.infer_compiler_registry.register_comfy.vae_patch.AttnBlock'>}

DEBUG [2024-08-05 17:46:18] /root/onediff/src/onediff/infer_compiler/backends/oneflow/transform/manager.py:57 - Updated class proxies: len(class_proxy_dict)=1
{'comfy.ldm.modules.diffusionmodules.openaimodel.Upsample': <class 'onediff_comfy_nodes.modules.oneflow.infer_compiler_registry.register_comfy.openaimodel.Upsample'>}

DEBUG [2024-08-05 17:46:18] /root/onediff/src/onediff/infer_compiler/backends/oneflow/transform/manager.py:57 - Updated class proxies: len(class_proxy_dict)=1
{'comfy.ldm.modules.diffusionmodules.openaimodel.UNetModel': <class 'onediff_comfy_nodes.modules.oneflow.infer_compiler_registry.register_comfy.openaimodel.UNetModel'>}

DEBUG [2024-08-05 17:46:18] /root/onediff/src/onediff/infer_compiler/backends/oneflow/transform/manager.py:57 - Updated class proxies: len(class_proxy_dict)=1
{'comfy.ldm.modules.diffusionmodules.openaimodel.VideoResBlock': <class 'onediff_comfy_nodes.modules.oneflow.infer_compiler_registry.register_comfy.openaimodel.VideoResBlock'>}

DEBUG [2024-08-05 17:46:19] /root/onediff/src/onediff/infer_compiler/backends/oneflow/transform/manager.py:57 - Updated class proxies: len(class_proxy_dict)=1
{'comfy.ldm.modules.attention.CrossAttention': <class 'register_comfy.attention.CrossAttention'>}

DEBUG [2024-08-05 17:46:19] /root/onediff/src/onediff/infer_compiler/backends/oneflow/transform/manager.py:57 - Updated class proxies: len(class_proxy_dict)=1
{'comfy.ldm.modules.attention.attention_pytorch': <function attention_pytorch_oneflow at 0x7f8d73dd2440>}

DEBUG [2024-08-05 17:46:19] /root/onediff/src/onediff/infer_compiler/backends/oneflow/transform/manager.py:57 - Updated class proxies: len(class_proxy_dict)=1
{'comfy.ldm.modules.attention.SpatialTransformer': <class 'register_comfy.attention.SpatialTransformer'>}

DEBUG [2024-08-05 17:46:19] /root/onediff/src/onediff/infer_compiler/backends/oneflow/transform/manager.py:57 - Updated class proxies: len(class_proxy_dict)=1
{'comfy.ldm.modules.attention.SpatialVideoTransformer': <class 'register_comfy.attention.SpatialVideoTransformer'>}

DEBUG [2024-08-05 17:46:19] /root/onediff/src/onediff/infer_compiler/backends/oneflow/transform/manager.py:57 - Updated class proxies: len(class_proxy_dict)=1
{'comfy.ldm.modules.diffusionmodules.util.AlphaBlender': <class 'register_comfy.util.AlphaBlender'>}

DEBUG [2024-08-05 17:46:19] /root/onediff/src/onediff/infer_compiler/backends/oneflow/transform/manager.py:57 - Updated class proxies: len(class_proxy_dict)=1
{'comfy.ops.disable_weight_init.Linear': <class 'register_comfy.linear.Linear'>}

DEBUG [2024-08-05 17:46:19] /root/onediff/src/onediff/infer_compiler/backends/oneflow/transform/manager.py:57 - Updated class proxies: len(class_proxy_dict)=1
{'comfy.ldm.modules.diffusionmodules.model.AttnBlock': <class 'register_comfy.vae_patch.AttnBlock'>}

DEBUG [2024-08-05 17:46:19] /root/onediff/src/onediff/infer_compiler/backends/oneflow/transform/manager.py:57 - Updated class proxies: len(class_proxy_dict)=1
{'comfy.ldm.modules.diffusionmodules.openaimodel.Upsample': <class 'register_comfy.openaimodel.Upsample'>}

DEBUG [2024-08-05 17:46:19] /root/onediff/src/onediff/infer_compiler/backends/oneflow/transform/manager.py:57 - Updated class proxies: len(class_proxy_dict)=1
{'comfy.ldm.modules.diffusionmodules.openaimodel.UNetModel': <class 'register_comfy.openaimodel.UNetModel'>}

DEBUG [2024-08-05 17:46:19] /root/onediff/src/onediff/infer_compiler/backends/oneflow/transform/manager.py:57 - Updated class proxies: len(class_proxy_dict)=1
{'comfy.ldm.modules.diffusionmodules.openaimodel.VideoResBlock': <class 'register_comfy.openaimodel.VideoResBlock'>}

DEBUG [2024-08-05 17:46:19] /root/onediff/src/onediff/infer_compiler/backends/oneflow/transform/custom_transform.py:33 - Register func=<function convert_fused_self_attn_processor at 0x7f8d73dd1000> first_param_type=<class 'onediff.optimization.attention_processor.FusedSelfAttnProcessor'>
DEBUG [2024-08-05 17:46:20] /root/onediff/src/onediff/infer_compiler/backends/oneflow/transform/manager.py:57 - Updated class proxies: len(class_proxy_dict)=1
{'ComfyUI-AnimateDiff-Evolved.animatediff.motion_module_ad.TemporalTransformer3DModel': <class 'onediff_comfy_nodes.modules.oneflow.hijack_animatediff.motion_module_ad.TemporalTransformer3DModel_OF'>}

DEBUG [2024-08-05 17:46:20] /root/onediff/src/onediff/infer_compiler/backends/oneflow/transform/manager.py:57 - Updated class proxies: len(class_proxy_dict)=1
{'ComfyUI-AnimateDiff-Evolved.animatediff.motion_module_ad.VersatileAttention': <class 'onediff_comfy_nodes.modules.oneflow.hijack_animatediff.motion_module_ad.VersatileAttention_OF'>}

DEBUG [2024-08-05 17:46:21] /root/onediff/src/onediff/infer_compiler/backends/oneflow/transform/manager.py:57 - Updated class proxies: len(class_proxy_dict)=1
{'ComfyUI-AnimateDiff-Evolved.animatediff.utils_motion.CrossAttentionMM': <class 'onediff_comfy_nodes.modules.oneflow.hijack_animatediff.utils_motion.CrossAttentionMM_OF'>}

Total VRAM 24210 MB, total RAM 773288 MB
pytorch version: 2.1.0+cu121
xformers version: 0.0.22.post4
Set vram state to: HIGH_VRAM
Device: cuda:0 NVIDIA GeForce RTX 4090 D : cudaMallocAsync

运行工作流

got prompt
model_type EPS
Using xformers attention in VAE
Using xformers attention in VAE
loaded straight to GPU
Requested to load SDXL
Loading 1 new model
/root/miniconda3/lib/python3.10/site-packages/oneflow/_dynamo/__init__.py:30: UserWarning: The oneflow._dynamo.allow_in_graph interface is just to align the torch._dynamo.allow_in_graph interface and has no practical significance.
  warnings.warn(
/root/miniconda3/lib/python3.10/site-packages/diffusers/models/dual_transformer_2d.py:20: FutureWarning: `DualTransformer2DModel` is deprecated and will be removed in version 0.29. Importing `DualTransformer2DModel` from `diffusers.models.dual_transformer_2d` is deprecated and this will be removed in a future version. Please use `from diffusers.models.transformers.dual_transformer_2d import DualTransformer2DModel`, instead.
  deprecate("DualTransformer2DModel", "0.29", deprecation_message)
/root/miniconda3/lib/python3.10/site-packages/diffusers/models/dual_transformer_2d.py:20: FutureWarning: `DualTransformer2DModel` is deprecated and will be removed in version 0.29. Importing `DualTransformer2DModel` from `diffusers.models.dual_transformer_2d` is deprecated and this will be removed in a future version. Please use `from diffusers.models.transformers.dual_transformer_2d import DualTransformer2DModel`, instead.
  deprecate("DualTransformer2DModel", "0.29", deprecation_message)
DEBUG [2024-08-05 17:49:36] /root/onediff/src/onediff/infer_compiler/backends/oneflow/transform/manager.py:57 - Updated class proxies: len(class_proxy_dict)=1
{'diffusers.models.attention_processor.Attention': <class 'register_diffusers.attention_processor_oflow.Attention'>}

DEBUG [2024-08-05 17:49:36] /root/onediff/src/onediff/infer_compiler/backends/oneflow/transform/manager.py:57 - Updated class proxies: len(class_proxy_dict)=1
{'diffusers.models.attention_processor.AttnProcessor': <class 'register_diffusers.attention_processor_oflow.AttnProcessor'>}

DEBUG [2024-08-05 17:49:36] /root/onediff/src/onediff/infer_compiler/backends/oneflow/transform/manager.py:57 - Updated class proxies: len(class_proxy_dict)=1
{'diffusers.models.attention_processor.AttnProcessor2_0': <class 'register_diffusers.attention_processor_oflow.AttnProcessor'>}

DEBUG [2024-08-05 17:49:36] /root/onediff/src/onediff/infer_compiler/backends/oneflow/transform/manager.py:57 - Updated class proxies: len(class_proxy_dict)=1
{'diffusers.models.attention_processor.LoRAAttnProcessor2_0': <class 'register_diffusers.attention_processor_oflow.LoRAAttnProcessor2_0'>}

DEBUG [2024-08-05 17:49:36] /root/onediff/src/onediff/infer_compiler/backends/oneflow/transform/manager.py:57 - Updated class proxies: len(class_proxy_dict)=1
{'diffusers.models.resnet.SpatioTemporalResBlock': <class 'register_diffusers.spatio_temporal_oflow.SpatioTemporalResBlock'>}

DEBUG [2024-08-05 17:49:36] /root/onediff/src/onediff/infer_compiler/backends/oneflow/transform/manager.py:57 - Updated class proxies: len(class_proxy_dict)=1
{'diffusers.models.autoencoders.autoencoder_kl_temporal_decoder.TemporalDecoder': <class 'register_diffusers.spatio_temporal_oflow.TemporalDecoder'>}

DEBUG [2024-08-05 17:49:36] /root/onediff/src/onediff/infer_compiler/backends/oneflow/transform/manager.py:57 - Updated class proxies: len(class_proxy_dict)=1
{'diffusers.models.transformers.transformer_temporal.TransformerSpatioTemporalModel': <class 'register_diffusers.spatio_temporal_oflow.TransformerSpatioTemporalModel'>}

DEBUG [2024-08-05 17:49:36] /root/onediff/src/onediff/infer_compiler/backends/oneflow/transform/manager.py:57 - Updated class proxies: len(class_proxy_dict)=1
{'diffusers.models.attention.TemporalBasicTransformerBlock': <class 'register_diffusers.spatio_temporal_oflow.TemporalBasicTransformerBlock'>}

DEBUG [2024-08-05 17:49:36] /root/onediff/src/onediff/infer_compiler/backends/oneflow/transform/manager.py:57 - Updated class proxies: len(class_proxy_dict)=1
{'diffusers.models.unets.unet_spatio_temporal_condition.UNetSpatioTemporalConditionModel': <class 'register_diffusers.spatio_temporal_oflow.UNetSpatioTemporalConditionModel'>}

DEBUG [2024-08-05 17:49:36] /root/onediff/src/onediff/infer_compiler/backends/oneflow/transform/manager.py:57 - Updated class proxies: len(class_proxy_dict)=1
{'diffusers.models.transformers.transformer_2d.Transformer2DModel': <class 'register_diffusers.transformer_2d_oflow.Transformer2DModel'>}

DEBUG [2024-08-05 17:49:36] /root/onediff/src/onediff/infer_compiler/backends/oneflow/transform/manager.py:57 - Updated class proxies: len(class_proxy_dict)=1
{'diffusers.models.unets.unet_2d_condition.UNet2DConditionModel': <class 'register_diffusers.unet_2d_condition_oflow.UNet2DConditionModel'>}

DEBUG [2024-08-05 17:49:36] /root/onediff/src/onediff/infer_compiler/backends/oneflow/transform/manager.py:57 - Updated class proxies: len(class_proxy_dict)=1
{'diffusers.models.unets.unet_2d_blocks.AttnUpBlock2D': <class 'register_diffusers.unet_2d_blocks_oflow.AttnUpBlock2D'>}

DEBUG [2024-08-05 17:49:36] /root/onediff/src/onediff/infer_compiler/backends/oneflow/transform/manager.py:57 - Updated class proxies: len(class_proxy_dict)=1
{'diffusers.models.unets.unet_2d_blocks.CrossAttnUpBlock2D': <class 'register_diffusers.unet_2d_blocks_oflow.CrossAttnUpBlock2D'>}

DEBUG [2024-08-05 17:49:36] /root/onediff/src/onediff/infer_compiler/backends/oneflow/transform/manager.py:57 - Updated class proxies: len(class_proxy_dict)=1
{'diffusers.models.unets.unet_2d_blocks.UpBlock2D': <class 'register_diffusers.unet_2d_blocks_oflow.UpBlock2D'>}

DEBUG [2024-08-05 17:49:36] /root/onediff/src/onediff/infer_compiler/backends/oneflow/transform/manager.py:57 - Updated class proxies: len(class_proxy_dict)=1
{'diffusers.models.upsampling.Upsample2D': <class 'register_diffusers.resnet_oflow.Upsample2D'>}

Requested to load CLIPVisionModelProjection
Loading 1 new model
DEBUG [2024-08-05 17:49:41] /root/talesofai_comfy/ComfyUI/custom_nodes/onediff_comfy_nodes/modules/oneflow/hijack_ipadapter_plus/IPAdapterPlus.py:30 - Caching new IPAdapter instance
DEBUG [2024-08-05 17:49:41] /root/talesofai_comfy/ComfyUI/custom_nodes/onediff_comfy_nodes/modules/oneflow/hijack_ipadapter_plus/IPAdapterPlus.py:30 - Caching new IPAdapter instance
DEBUG [2024-08-05 17:49:41] /root/talesofai_comfy/ComfyUI/custom_nodes/onediff_comfy_nodes/modules/oneflow/hijack_ipadapter_plus/IPAdapterPlus.py:30 - Caching new IPAdapter instance
DEBUG [2024-08-05 17:49:41] /root/talesofai_comfy/ComfyUI/custom_nodes/onediff_comfy_nodes/modules/oneflow/hijack_ipadapter_plus/IPAdapterPlus.py:30 - Caching new IPAdapter instance
DEBUG [2024-08-05 17:49:41] /root/talesofai_comfy/ComfyUI/custom_nodes/onediff_comfy_nodes/modules/oneflow/hijack_ipadapter_plus/IPAdapterPlus.py:30 - Caching new IPAdapter instance
DEBUG [2024-08-05 17:49:41] /root/talesofai_comfy/ComfyUI/custom_nodes/onediff_comfy_nodes/modules/oneflow/hijack_ipadapter_plus/IPAdapterPlus.py:30 - Caching new IPAdapter instance
DEBUG [2024-08-05 17:49:41] /root/talesofai_comfy/ComfyUI/custom_nodes/onediff_comfy_nodes/modules/oneflow/hijack_ipadapter_plus/IPAdapterPlus.py:30 - Caching new IPAdapter instance
DEBUG [2024-08-05 17:49:41] /root/talesofai_comfy/ComfyUI/custom_nodes/onediff_comfy_nodes/modules/oneflow/hijack_ipadapter_plus/IPAdapterPlus.py:30 - Caching new IPAdapter instance
DEBUG [2024-08-05 17:49:41] /root/talesofai_comfy/ComfyUI/custom_nodes/onediff_comfy_nodes/modules/oneflow/hijack_ipadapter_plus/IPAdapterPlus.py:30 - Caching new IPAdapter instance
DEBUG [2024-08-05 17:49:41] /root/onediff/src/onediff/infer_compiler/backends/oneflow/transform/manager.py:57 - Updated class proxies: len(class_proxy_dict)=1
{'comfy.ldm.modules.attention.CrossAttention': <class 'register_comfy.attention.CrossAttention'>}

DEBUG [2024-08-05 17:49:41] /root/onediff/src/onediff/infer_compiler/backends/oneflow/transform/manager.py:57 - Updated class proxies: len(class_proxy_dict)=1
{'comfy.ldm.modules.attention.attention_pytorch': <function attention_pytorch_oneflow at 0x7f89dc22b0a0>}

DEBUG [2024-08-05 17:49:41] /root/onediff/src/onediff/infer_compiler/backends/oneflow/transform/manager.py:57 - Updated class proxies: len(class_proxy_dict)=1
{'comfy.ldm.modules.attention.SpatialTransformer': <class 'register_comfy.attention.SpatialTransformer'>}

DEBUG [2024-08-05 17:49:41] /root/onediff/src/onediff/infer_compiler/backends/oneflow/transform/manager.py:57 - Updated class proxies: len(class_proxy_dict)=1
{'comfy.ldm.modules.attention.SpatialVideoTransformer': <class 'register_comfy.attention.SpatialVideoTransformer'>}

DEBUG [2024-08-05 17:49:41] /root/onediff/src/onediff/infer_compiler/backends/oneflow/transform/manager.py:57 - Updated class proxies: len(class_proxy_dict)=1
{'comfy.ldm.modules.diffusionmodules.util.AlphaBlender': <class 'register_comfy.util.AlphaBlender'>}

DEBUG [2024-08-05 17:49:41] /root/onediff/src/onediff/infer_compiler/backends/oneflow/transform/manager.py:57 - Updated class proxies: len(class_proxy_dict)=1
{'comfy.ops.disable_weight_init.Linear': <class 'register_comfy.linear.Linear'>}

DEBUG [2024-08-05 17:49:41] /root/onediff/src/onediff/infer_compiler/backends/oneflow/transform/manager.py:57 - Updated class proxies: len(class_proxy_dict)=1
{'comfy.ldm.modules.diffusionmodules.model.AttnBlock': <class 'register_comfy.vae_patch.AttnBlock'>}

DEBUG [2024-08-05 17:49:41] /root/onediff/src/onediff/infer_compiler/backends/oneflow/transform/manager.py:57 - Updated class proxies: len(class_proxy_dict)=1
{'comfy.ldm.modules.diffusionmodules.openaimodel.Upsample': <class 'register_comfy.openaimodel.Upsample'>}

DEBUG [2024-08-05 17:49:41] /root/onediff/src/onediff/infer_compiler/backends/oneflow/transform/manager.py:57 - Updated class proxies: len(class_proxy_dict)=1
{'comfy.ldm.modules.diffusionmodules.openaimodel.UNetModel': <class 'register_comfy.openaimodel.UNetModel'>}

DEBUG [2024-08-05 17:49:41] /root/onediff/src/onediff/infer_compiler/backends/oneflow/transform/manager.py:57 - Updated class proxies: len(class_proxy_dict)=1
{'comfy.ldm.modules.diffusionmodules.openaimodel.VideoResBlock': <class 'register_comfy.openaimodel.VideoResBlock'>}

Requested to load SDXLClipModel
Loading 1 new model
Requested to load SDXL
Loading 1 new model
  0%|                                                                                                                                                                                                                                                                                             | 0/24 [00:00<?, ?it/s]DEBUG [2024-08-05 17:50:20] /root/onediff/src/onediff/infer_compiler/backends/oneflow/dual_module.py:29 - Convert <class 'comfy.ldm.modules.diffusionmodules.openaimodel.UNetModel'> ...
DEBUG [2024-08-05 17:50:28] /root/onediff/src/onediff/infer_compiler/backends/oneflow/dual_module.py:31 - Convert <class 'comfy.ldm.modules.diffusionmodules.openaimodel.UNetModel'> done!
INFO [2024-08-05 17:50:28] /root/onediff/src/onediff/infer_compiler/backends/oneflow/graph_management_utils.py:84 - Graph file /root/autodl-tmp/graphs/SDXL_7ecb88_d49b38b9.graph does not exist! Generating graph.
INFO [2024-08-05 17:50:28] /root/onediff/src/onediff/infer_compiler/backends/oneflow/graph.py:14 - Building a graph for <class 'register_comfy.openaimodel.UNetModel'> ...
DEBUG [2024-08-05 17:51:07] /root/onediff/src/onediff/infer_compiler/backends/oneflow/utils/cost_util.py:51 - ==> function onediff.infer_compiler.backends.oneflow.graph.save_graph  try to run...
DEBUG [2024-08-05 17:51:07] /root/onediff/src/onediff/infer_compiler/backends/oneflow/utils/cost_util.py:57 - save_graph cuda mem before 10734.5625 MB
DEBUG [2024-08-05 17:51:07] /root/onediff/src/onediff/infer_compiler/backends/oneflow/utils/cost_util.py:60 - save_graph host mem before 10174.0 MB
DEBUG [2024-08-05 17:51:08] /root/onediff/src/onediff/infer_compiler/backends/oneflow/utils/cost_util.py:67 - save_graph run time 1.3020057678222656 seconds
DEBUG [2024-08-05 17:51:08] /root/onediff/src/onediff/infer_compiler/backends/oneflow/utils/cost_util.py:70 - save_graph cuda mem after 10958.5625 MB
DEBUG [2024-08-05 17:51:08] /root/onediff/src/onediff/infer_compiler/backends/oneflow/utils/cost_util.py:72 - save_graph cuda mem diff 224.0 MB
DEBUG [2024-08-05 17:51:08] /root/onediff/src/onediff/infer_compiler/backends/oneflow/utils/cost_util.py:74 - save_graph host mem after 10898.0 MB
DEBUG [2024-08-05 17:51:08] /root/onediff/src/onediff/infer_compiler/backends/oneflow/utils/cost_util.py:75 - save_graph host mem diff 724.0 MB
DEBUG [2024-08-05 17:51:08] /root/onediff/src/onediff/infer_compiler/backends/oneflow/utils/cost_util.py:79 - <== function save_graph finish run.
DEBUG [2024-08-05 17:51:08] /root/onediff/src/onediff/infer_compiler/backends/oneflow/utils/cost_util.py:80 - 
INFO [2024-08-05 17:51:08] /root/onediff/src/onediff/infer_compiler/backends/oneflow/graph_management_utils.py:113 - Saved graph file: /root/autodl-tmp/graphs/SDXL_7ecb88_d49b38b9.graph
INFO [2024-08-05 17:51:08] /root/onediff/src/onediff/infer_compiler/backends/oneflow/param_utils.py:189 - generate constant folding info
100%|████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 24/24 [00:49<00:00,  2.06s/it]
Requested to load AutoencoderKL
Loading 1 new model
Prompt executed in 137.83 seconds
ccssu commented 1 month ago

感谢您的耐心和回复 @ccssu ，我在第0步时，只使用 Load Checkpoint - OneDiff 节点加载了工作流使用的模型，但其运行结果显示工作流中带有的IPA节点并没有生效（并且保存的graph较生效时小一倍）。 重启comfyui（清除了硬盘和内存（显存）中的graph）后，我尝试使用了官方工作流ipadapter_advanced.json，其IPA生效并生成了一个较大的graph，使用我的工作流加载这个graph并运行，运行时间表明其进行了重新编译，同时结果显示IPA生效，但这时Load Checkpoint - OneDiff 节点并没有保存这个graph，因为当我重启comfyui（清除了内存（显存）中的graph，没有删除graph文件）后，再次运行，IPA生效，但其运行时间表明进行了重新编译，这时不重启再次运行，则运行时间明显缩短且效果与上一次一致

上面这个看起来第一次运行时成功保存了图 符合预期：INFO [2024-08-05 17:51:08] /root/onediff/src/onediff/infer_compiler/backends/oneflow/graph_management_utils.py:113 - Saved graph file: /root/autodl-tmp/graphs/SDXL_7ecb88_d49b38b9.graph

看上面有提到的问题，目前

  1. oad Checkpoint - OneDiff 节点加载了工作流使用的模型，但其运行结果显示工作流中带有的IPA节点并没有生效（并且保存的graph较生效时小一倍） 这个估计是图名称冲突了，可以通过指定目录解决下。@LingXuanYin 

    # Set custom directory for saving graphs in ComfyUI with OneFlow backend
export COMFYUI_ONEDIFF_SAVE_GRAPH_DIR="/path/to/save/graphs"

  2. 重启comfyui（不要清除了硬盘和内存（显存）中的graph），预计应该是会触发 load graph 不是重新编译 。 可在 debug 模式观察

tzwm commented 1 month ago

感谢您的耐心和回复 @ccssu ，我在第0步时，只使用 Load Checkpoint - OneDiff 节点加载了工作流使用的模型，但其运行结果显示工作流中带有的IPA节点并没有生效（并且保存的graph较生效时小一倍）。 重启comfyui（清除了硬盘和内存（显存）中的graph）后，我尝试使用了官方工作流ipadapter_advanced.json，其IPA生效并生成了一个较大的graph，使用我的工作流加载这个graph并运行，运行时间表明其进行了重新编译，同时结果显示IPA生效，但这时Load Checkpoint - OneDiff 节点并没有保存这个graph，因为当我重启comfyui（清除了内存（显存）中的graph，没有删除graph文件）后，再次运行，IPA生效，但其运行时间表明进行了重新编译，这时不重启再次运行，则运行时间明显缩短且效果与上一次一致

上面这个看起来第一次运行时成功保存了图 符合预期：INFO [2024-08-05 17:51:08] /root/onediff/src/onediff/infer_compiler/backends/oneflow/graph_management_utils.py:113 - Saved graph file: /root/autodl-tmp/graphs/SDXL_7ecb88_d49b38b9.graph

看上面有提到的问题，目前

  1. oad Checkpoint - OneDiff 节点加载了工作流使用的模型，但其运行结果显示工作流中带有的IPA节点并没有生效（并且保存的graph较生效时小一倍） 这个估计是图名称冲突了，可以通过指定目录解决下。@LingXuanYin
# Set custom directory for saving graphs in ComfyUI with OneFlow backend
export COMFYUI_ONEDIFF_SAVE_GRAPH_DIR="/path/to/save/graphs"
  1. 重启comfyui（不要清除了硬盘和内存（显存）中的graph），预计应该是会触发 load graph 不是重新编译 。 可在 debug 模式观察
  1. 指定了一个全新的 COMFYUI_ONEDIFF_SAVE_GRAPH_DIR，直接跑 IPA 工作流，还是同样的结果 IPA 没有生效。生成的 graph 是个不带 IPA 的
  2. 没有重启 comfyui 尝试的，主要还是上面那个 IPA 无法生效(保存的 graph 不带 IPA)的问题
ccssu commented 1 month ago

_指定了一个全新的 COMFYUI_ONEDIFF_SAVE_GRAPHDIR，直接跑 IPA 工作流，还是同样的结果 IPA 没有生效。生成的 graph 是个不带 IPA 的

这个比较奇怪，方便提供个开源可复现工作流么， 我们实验下。

ccssu commented 1 month ago

Progress @tzwm @LingXuanYin : Fixed and awaiting merge PR https://github.com/siliconflow/onediff/pull/1076