Commit Graph

Select branches

Hide Pull Requests

burndown/issue-144-clean-base

dvram

issue-178

main

Mono Color

• 04b5bb0a6f refactor: Enhance block swap memory analysis report John Pollock 2025-08-11 19:19:58 -05:00
• bde78780da Flux block swap manager, self contained for now. We'll do the same for wanvideo and qwen then re-evaluate point-solutions and opportunities to synergize John Pollock 2025-08-11 17:09:30 -05:00
• d666205fd9 Refactor: Overhaul BlockSwap with hook-based manager John Pollock 2025-08-11 14:28:20 -05:00
• 62718ea12f Refactor: Simplify block swap analysis and cleanup .gitignore John Pollock 2025-08-11 12:47:25 -05:00
• 372217e901 fixing inconsistent variable naming choices and defaults John Pollock 2025-08-11 12:25:46 -05:00
• a70ce10960 Replace block-swap-v3 with aa49ad11 John Pollock 2025-08-11 09:45:26 -05:00
• 02b02629ba More garbage John Pollock 2025-08-10 22:34:05 -05:00
• aa49ad1139 feat: Introduce DisTorch v2 with BlockSwap memory management John Pollock 2025-08-10 20:34:48 -05:00
• 898169fccf refactor: Move core logic into separate modules John Pollock 2025-08-10 09:53:33 -05:00
• 3e3190346b branch cleanup John Pollock 2025-08-10 02:13:10 -05:00
• fa6141911f feat: Standardize DisTorch UI for GGUF and SafeTensors John Pollock 2025-08-09 18:31:53 -05:00
• 996f298b4e feat: Implement robust block discovery and GGUF-style logging for SafeTensor DisTorch John Pollock 2025-08-09 17:51:15 -05:00
• 992b1c6587 docs: Update architecture document for DisTorch SafeTensor John Pollock 2025-08-09 10:46:05 -05:00
• aee0987779 feat: Refactor DisTorch SafeTensor wrapper and expand coverage John Pollock 2025-08-09 10:00:28 -05:00
• 4adcce61a2 Add extensive debug logging to DisTorch implementation John Pollock 2025-08-08 15:20:56 -05:00
• b59f9bd3a8 Rename DisTorchBlockSwap to DisTorch per user feedback John Pollock 2025-08-08 14:54:36 -05:00
• 897f785edf Initial block swap implementation v2 John Pollock 2025-08-08 14:50:26 -05:00
• 92a10cc6ec Minor cleanup John Pollock 2025-08-08 02:31:00 -05:00
• 79e9230f4c fix(gguf): correct missing type enum for CLIPLoaderGGUF and DualCLIPLoaderGGUF John Pollock 2025-08-08 01:36:59 -05:00
• 7a08dd97d5 feat: Experimental XPU support John Pollock 2025-08-07 16:22:53 -05:00
• 34a15594e2 Merge pull request #48 from ComfyNodePRs/update-publish-yaml John Pollock 2025-08-07 03:41:06 -05:00
• a6f13e5ff3 feat: update README and pyproject.toml for enhanced WanVideoWrapper integration and version bump to 1.8.0 John Pollock 2025-08-06 17:58:07 -05:00
• d4b930776e MultiGPU patches for WanVideoWrapper - took a slightly different approach on these nodes, as I my intention was always to play nice with Kijai's code. John Pollock 2025-08-06 17:33:18 -05:00
• 657fdac13a Fix WanVideo multi-GPU device mismatch issue John Pollock 2025-08-06 04:30:03 -05:00
• 582ca6a247 WanVideoWrapper MultiGPU integration - custom wrapper nodes John Pollock 2025-08-05 18:59:16 -05:00
• a05823ff0a feat: add CLIPVisionLoaderMultiGPU support and update version to 1.7.3 John Pollock 2025-04-17 18:43:01 -05:00
• 4ff9b80286 feat: add QuadrupleCLIPLoader / QuadrupleCLIPLoaderGGUF support and update version to 1.7.2 John Pollock 2025-04-17 17:00:20 -05:00
• b0159761e2 feat: add support for 'pixart' and 'wan' types in CLIPLoaderGGUF to match core class; update version to 1.7.1 John Pollock 2025-03-24 05:38:44 -05:00
• 2d81ef0a21 Support for kijai's ComfyUI-WanVideoWrapper John Pollock 2025-03-23 13:40:05 -05:00
• 1bf9333fc7 chore: update version to 1.6.2 and fix description formatting in pyproject.toml John Pollock 2025-02-12 12:18:13 -06:00
• a2093a4fc9 feat: add text encoder device handling, whereas CLIP can sometimes default to CPU, whereas using a DisTorch CLIP load you can load the layes on CPU buy use CUDA for processing. Especially helpful llava-llama John Pollock 2025-02-12 11:51:58 -06:00
• 6500ca9a47 docs: enhance README for clarity on DisTorch Virtual VRAM features and usage based on user success stories and improved ease of use John Pollock 2025-02-11 22:50:03 -06:00
• c431a7d870 Merge pull request #16 from eltociear/patch-1 John Pollock 2025-02-09 19:21:30 -06:00
• e8be359b35 docs: update README.md Ikko Eltociear Ashimine 2025-02-10 03:25:20 +09:00
• 62646d3ca3 Push Distorch 2.0 Virtual VRAM release out to Comfy Registry John Pollock 2025-02-07 22:00:48 -06:00
• 04882505cb Merge dev into main, taking dev version of __init__.py John Pollock 2025-02-07 21:52:58 -06:00
• c01ef265a8 Update README to enhance clarity on manual allocation strings and installation instructions John Pollock 2025-02-07 21:46:01 -06:00
• e36cec9fb7 Add new assets and update README for DisTorch 2.0 features John Pollock 2025-02-07 21:45:41 -06:00
• 375276a485 Add files via upload John Pollock 2025-02-07 21:26:07 -06:00
• 1af932be16 Add files via upload John Pollock 2025-02-07 20:34:50 -06:00
• 8642f75a12 Add files via upload John Pollock 2025-02-07 20:17:37 -06:00
• c8c4e74692 cleaning up experimental workflows John Pollock 2025-02-07 19:17:06 -06:00
• 9bd984b420 Update default value for virtual VRAM GB to 4.0 in override_class_with_distorch John Pollock 2025-02-07 18:27:14 -06:00
• de2219c974 Refactor virtual VRAM allocation logic and improve logging format John Pollock 2025-02-07 18:24:28 -06:00
• c98a535435 Refactor logging in DisTorch analysis and update allocation handling for virtual VRAM John Pollock 2025-02-07 16:09:35 -06:00
• 3a4c6d50c8 Virtual VRAM "automatic" mode for DisTorch, WIP but working John Pollock 2025-02-07 15:05:08 -06:00
• 5a403e638c MergeFluxLoRAsQuantizeAndLoad, WIP John Pollock 2025-02-07 04:43:45 -06:00
• 1ffa88f271 Benchmarking JSON for Reddit user John Pollock 2025-02-06 14:29:26 -06:00
• 08e3bbfcb0 add example for reddit user John Pollock 2025-02-06 11:53:05 -06:00
• d0d33a69ac Corrected Florence2 nodes and removed DiffSynth to remain consistent with emergency release to main John Pollock 2025-02-06 03:09:00 -06:00
• 3679c268ac 🔧 BREAKING CHANGE: DiffSynth functionality has been temporarily removed while investigating device management issue (Having two GPUs with different amounts of VRAM leads to OOMs pollockjj/ComfyUI-MultiGPU#13). John Pollock 2025-02-06 00:32:21 -06:00
• 4182275f85 Multiple LoRA DisTorch example John Pollock 2025-02-05 10:55:17 -06:00
• 4a8d70a0d4 refactored to move stable wrapper nodes into nodes.py and remainder in init.py John Pollock 2025-02-03 09:15:05 -06:00
• 262ceea716 Add precompiled llama-quantize binaries for flux Just-in-time GGUF quantization John Pollock 2025-02-03 00:15:18 -06:00
• df779bf820 Add new example JSON for Maleficent_End8551 workflow John Pollock 2025-02-01 12:35:49 -06:00
• f5038d73bd bitsandbytes_NF4 example workflows John Pollock 2025-02-01 08:00:00 -06:00
• 11d41b5192 Example of taking lowvram flow even lower John Pollock 2025-02-01 06:19:40 -06:00
• 89ded6cc48 Added new ltxvideo example for lowvram, high dram - run ~5 minutes for a 5 second video on my 3090 John Pollock 2025-01-31 11:42:49 -06:00
• 3e130e3dfb Remove log_comfy_states function - no longer needed John Pollock 2025-01-31 06:45:43 -06:00
• 005b5b1882 This release includes an embeddings adapter for the IP2V part of kijai's CLIP loader for HunyuanVideo. See examples. Bump version to 1.4.3 and update category for HunyuanVideoEmbeddingsAdapter to multigpu; enhance README with new workflow examples for HunyuanVideo GGUF-quantized models. John Pollock 2025-01-29 11:55:16 -06:00
• 3260b7e38e Add HunyuanVideoEmbeddingsAdapter class for using kijai's IP2V conditioning video embeddings in the standard sampler, allowing it to be used with GGUF/DisTorch methods. John Pollock 2025-01-29 09:25:19 -06:00
• 185ecc2ab5 Push fix to check_module_exists() to Comfy Registry, thanks @3dluvr John Pollock 2025-01-29 05:16:45 -06:00
• fc4e2b2f12 Merge pull request #12 from 3dluvr/main John Pollock 2025-01-29 05:12:49 -06:00
• 379ecce687 Fix check_module_exists() to use folder_paths 3dluvr 2025-01-28 19:59:44 -05:00
• 41e974f086 1.40 -->1.4.1 Pop 1.4 (DisTorch) to Comfy Registry John Pollock 2025-01-28 15:41:13 -06:00
• 369255ff8d Merge branch 'main' of https://github.com/pollockjj/ComfyUI-MultiGPU into dev John Pollock 2025-01-28 06:26:25 -06:00
• 5ff6a90b17 preparing DisTorch for release to :main: John Pollock 2025-01-28 06:06:53 -06:00
• 8f3ff06fa3 rename files for consistency, new DisTorch examples John Pollock 2025-01-28 05:19:50 -06:00
• c07a345c45 Fix case sensitivity in module check for HunyuanVideoWrapper in __init__.py pollock 2025-01-27 18:32:20 -05:00
• 7ccea97c52 Update device allocation format and enhance module check for case insensitivity in __init__.py John Pollock 2025-01-27 17:24:15 -06:00
• c74f66b9ad icon John Pollock 2025-01-27 23:06:56 -06:00
• 3dbfcc7135 Chasing down bug causing incorrect patched device with distorch code. Re-integrated distorch into __init__.py as one of the consequences. John Pollock 2025-01-25 17:49:12 -06:00
• 667115eb19 chore(publish): update GitHub Actions workflow for node publishing snomiao 2025-01-25 17:27:43 +00:00
• d67b18474a Merge branch 'dev' of https://github.com/pollockjj/ComfyUI-MultiGPU into dev John Pollock 2025-01-24 21:06:10 -06:00
• b99e152686 Refactor GGUF model patcher to enhance device handling and introduce model hashing for allocation tracking John Pollock 2025-01-24 21:06:01 -06:00
• 101fb49848 Refactor model hash creation and logging for clarity; remove unnecessary logging statements in distorch.py pollock 2025-01-24 11:00:38 -05:00
• ef614f840e Move distorch parameter storage from model attachment (non-working) to global table with hash. pollock 2025-01-24 09:22:43 -05:00
• 2ac057d338 Implement model hash creation and allocation storage for MultiGPU support in GGUF model patcher pollock 2025-01-24 05:37:17 -05:00
• 624c893942 Refactored into init.py and distorch.py John Pollock 2025-01-23 13:16:18 -06:00
• 43d7d1582d Refactor UnetLoaderGGUF registration to support MultiGPU and DisTorch versions pollock 2025-01-23 12:00:24 -05:00
• b8f314921c Refactor imports and logging messages for clarity and consistency pollock 2025-01-23 08:50:48 -05:00
• d9899d4df1 Refactor GGUF model patcher and analysis functions to improve device handling and logging John Pollock 2025-01-21 06:50:10 -06:00
• 5eb03a220a Add .vscode/settings.json to .gitignore to exclude IDE-specific settings, DisTorch work in progress. Clip also loaded and distributed. Much WIP. John Pollock 2025-01-21 02:50:10 -06:00
• d8d122f397 Refactor MultiGPU module registration and improve device handling logic towards releasing on :main: John Pollock 2025-01-20 12:11:12 -06:00
• 1a3cc9d151 Refactor MultiGPU device handling and improve logging for better traceability John Pollock 2025-01-20 11:34:14 -06:00
• 455a4ef3f8 Refactor code structure for improved readability and maintainability John Pollock 2025-01-20 07:40:33 -06:00
• a7c424f238 Continued clean-up of DisTorch code. John Pollock 2025-01-19 21:04:16 -06:00
• 018aec49b0 Add detailed logging for MultiGPU device memory allocation analysis John Pollock 2025-01-19 17:00:26 -06:00
• f119787bee ggml learning John Pollock 2025-01-19 14:24:11 -06:00
• bb2b08efc8 docs: Update README to clarify device selection and model offloading features for ComfyUI John Pollock 2025-01-15 09:35:32 -06:00
• ff7329298c feat: Releasing Hunyuan Video UNet device split support. Update project description and bump version to 1.4.0 John Pollock 2025-01-15 09:26:42 -06:00
• e73c8bd5e3 Add experimental DiffSynth block-swapping support via new GPU offload device John Pollock 2025-01-15 06:53:59 -06:00
• 3ade69a38e Merge branch 'main' into dev give new dev branch proper start John Pollock 2025-01-14 22:12:25 -06:00
• 54157b59b6 adding animated icon John Pollock 2025-01-14 15:32:12 -06:00
• a19bf2755b modified aspect ratio of icon John Pollock 2025-01-12 13:24:38 -06:00
• 20ba9d099c fix: Bump version to 1.3.1 and update icon URL to raw GitHub link John Pollock 2025-01-12 11:17:47 -06:00
• 92652cf816 feat: Update project description to include CPU device selection and bump version to 1.3.0; add icon URL for ComfyUI-MultiGPU John Pollock 2025-01-12 11:08:41 -06:00
• 4e253c5d9a docs: Update README to reflect hunyuanvideowrapper examples and additional example for DeviceSelectorMultiGPU usage John Pollock 2025-01-12 11:00:18 -06:00
• 5c6b55c61c added icon for comfy registry John Pollock 2025-01-12 10:36:14 -06:00
• 2447586cf4 Added examples of device_selector node, include use with kijai's hunyuanvideowrapper custom_node showing both methods of usage John Pollock 2025-01-11 17:58:15 -06:00