320 Commits

Include renames

Author	SHA1	Message	Date
comfyanonymousandGitHub	e2a6e30d89	Fix black image on turing when using int4 models. (#14864)	2026-07-09 23:17:06 -04:00
comfyanonymousandGitHub	73e84d5ec8	Support convrot int4 models. (#14859) ... linear_dtype in comfy_quant metadata can be used to set if the int4 op does the matrix multiplication in int8 or int4, the default is int4 on GPUs that support it with fallback to int8 for GPUs that don't.	2026-07-09 18:57:09 -04:00
Daxiong (Lin)andGitHub	6cc814437f	Update workflow templates to v0.11.6 (#14834)	2026-07-08 14:04:57 -07:00
Daxiong (Lin)andGitHub	b08debceca	chore: update embedded docs to v0.5.7 (#14783)	2026-07-06 09:56:09 +08:00
Daxiong (Lin)andGitHub	a04ebe05c2	chore: update workflow templates to v0.11.2 (#14741)	2026-07-03 19:08:11 +08:00
Daxiong (Lin)andGitHub	6fca64780c	chore: update workflow templates to v0.11.1 (#14698)	2026-06-30 14:28:09 -07:00
comfyanonymousandGitHub	d395813bcd	Fix memory leak related to int8. (#14697)	2026-06-30 14:08:59 -07:00
Comfy Org PR BotandGitHub	510ed5c384	Bump comfyui-frontend-package to 1.45.20 (#14684)	2026-06-30 16:25:03 +08:00
comfyanonymousandGitHub	7851410511	Better and faster int8 lora applying. (#14685)	2026-06-29 21:52:08 -04:00
3 people	a58473fd9b	chore: update embedded docs to v0.5.6 (#14668) ... Co-authored-by: Alexis Rolland <alexisrolland@hotmail.com>	2026-06-29 17:08:06 +08:00
comfyanonymousandGitHub	a95e461916	int8 support on turing GPUs. (#14662)	2026-06-27 15:53:11 -07:00
pythongosssssandGitHub	603d891eaf	Update GLSL node to use ANGLE library (CORE-162) (#13195)	2026-06-27 08:40:31 +08:00
comfyanonymousandGitHub	470ac36a0a	Fix int8 loras causing lower quality requant with wrong settings. (#14650) ... * Update comfy-kitchen * Support requantizing with same settings as orig quant.	2026-06-26 16:41:29 -07:00
comfyanonymousandGitHub	7cb784e0f4	Faster int8. (#14641)	2026-06-25 15:25:47 -07:00
comfyanonymousandGitHub	1a510f0423	Support int8 models. (#14636)	2026-06-25 11:23:58 -07:00
Daxiong (Lin)andGitHub	639c8fa788	chore: update workflow templates to v0.10.7 (#14632)	2026-06-25 23:05:34 +08:00
Daxiong (Lin)andGitHub	12218db68a	Update the template to bring the HH1.1 templates back (#14613)	2026-06-24 21:01:25 +08:00
Daxiong (Lin)andGitHub	261bdb7cac	chore: update workflow templates to v0.10.3 (#14603)	2026-06-23 13:06:26 -04:00
Daxiong (Lin)andGitHub	d0b640fff7	chore: update workflow templates to v0.10.2 (#14600)	2026-06-23 23:35:21 +08:00
Daxiong (Lin)andGitHub	6978a466b8	chore: update embedded docs to v0.5.5 (#14585)	2026-06-22 18:29:02 +08:00
Comfy Org PR BotandGitHub	bc11e8a65a	Bump comfyui-frontend-package to 1.45.19 (#14559)	2026-06-19 16:01:34 -07:00
Daxiong (Lin)andGitHub	83a3f03218	chore: update workflow templates to v0.10.0 (#14482)	2026-06-15 08:06:15 -07:00
rattusandGitHub	ec4dec93d2	Comfy Aimdo 0.4.10 + Dynamic --reserve-vram + --vram-headroom (#14480) ... * main: implement --vram-headroom Implement --vram-headroom for dynamic vram as a hybrid debug/diagnostic option that can be used for people who still report shared VRAM spills. They can trial and error the setting to maintain a bit more headroom to avoid shared VRAM spills. * main: implement --reserve-vram Implement --reserve-vram as extra headroom on the simple method which is semantically as close as possible to the stated functionality and formet behaviour of non-dynamic VRAM.	2026-06-15 07:54:36 -07:00
Daxiong (Lin)andGitHub	7d4194d984	chore: update embedded docs to v0.5.4 (#14478)	2026-06-15 16:35:36 +08:00
Daxiong (Lin)andGitHub	38f750d80e	chore: update embedded docs to v0.5.3 (#14350)	2026-06-08 22:58:52 +08:00
rattusandGitHub	ec6aa979a6	aimdo 049 (#14300)	2026-06-05 08:40:03 -07:00
Comfy Org PR BotandGitHub	4e1f7cb1db	Bump comfyui-frontend-package to 1.45.15 (#14265)	2026-06-04 11:41:33 -07:00
Daxiong (Lin)andGitHub	6ecca5f468	chore: update workflow templates to v0.9.98 (#14284)	2026-06-04 09:40:44 -07:00
Daxiong (Lin)andGitHub	f0619af659	chore: update workflow templates to v0.9.94 (#14263)	2026-06-03 09:10:26 -07:00
rattusandGitHub	bd7da053ae	comfy-aimdo: 0.4.8 (#14244) ... Aimdo 0.4.8 fixes a crash in multi-gpu due to contention on the singleton bounce buffer.	2026-06-02 18:57:16 -07:00
Daxiong (Lin)andGitHub	412d9ac33a	chore: update workflow templates to v0.9.92 (#14212)	2026-06-01 07:41:00 -07:00
Daxiong (Lin)andGitHub	ea73d3b2ea	chore: update embedded docs to v0.5.2 (#14193)	2026-05-31 07:49:59 +08:00
rattusandGitHub	e154da83b1	Threaded Loader performance fixes / improvements (+ Aimdo 0.4.6) (#14116) ... * memory_management: Add direct to read GPU mode Make destination optional (or make it optionally GPU) and use aimdo to file_read direct to GPU. * ops: Remove stream pin buffers and use aimdo reads This consumed too much RAM and its better to just take the hit on the CPU syncing back the stream on a short ring buffer. Aimdo implements this so just rip the stream pin buffer from comfy. * model_management: all active pin registration movement Its better to just let the active model load past the pin limit as pins and let the pins move around. The saves the HDD and SATA people disk traffic while only costing a few GPU syncs. * utils: use aimdo file handle This opens on windows with more favourable flags * mp: only count the model proper for loaded_ram and vram Exclude live loras from the numbers to avoid the case where the reported loaded memory exceeds the size of the model. This causes me confusion in the Kijai visualizer when it looked fully loaded but was hitting disk due to this accounding disrepency. * utils: add bit reverse utility useful for max scattering something ordered. * pinned_memory: Implement offload balancing Use a max scatter alogorithm to prioritize pins of the same size such that when doing a little bit of offloading it gets scattered, allowing the prefetcher to more evenly swollow the offload. * comfy-aimdo 0.4.7 Aimdo 0.4.7 implement VRAM buffer exhaustion predection to avoid early speculative load of weights that definately wont fix once the inference gets further in. * model-prefetch: consolidate pin ensures on the sync point This could happen mid prefetch block, cause a sync of the entire block and lose overlap. Get ahead of the problem with a free down at the natural compute stream sync point. * mm: Put a 2GB min on the pin ceiling This is reasonably bad if it starts causing swap pressure, moreso than during normal ram-cache proceedings. Clamp it. * add --fast-disk	2026-05-30 15:20:04 -04:00
comfyanonymousandGitHub	0b04660ba3	Speed up anima a bit on nvidia. (#14181)	2026-05-29 22:47:10 -07:00
Daxiong (Lin)andGitHub	b10a61615c	chore: update workflow templates to v0.9.91 (#14163)	2026-05-28 22:42:17 -07:00
comfyanonymousandGitHub	ade4dfd96a	Update and pin comfy-kitchen version to 0.2.9 (#14161)	2026-05-28 19:23:17 -07:00
comfyanonymousandGitHub	bcf805aaea	Bump pyav package to fix some image loading issues. (#14160)	2026-05-28 17:38:01 -07:00
Daxiong (Lin)andGitHub	53eba227f5	chore: update workflow templates to v0.9.85 (#14134)	2026-05-27 05:32:58 -07:00
Daxiong (Lin)andGitHub	a4141a0f5a	chore: update embedded docs to v0.5.1 (#14101)	2026-05-26 01:57:18 +08:00
rattusandGitHub	39f963b4b0	mark loads to pins as cold immediately (#14088) ... This does the posix_fadvise to kick pins out of the disk cache (to avoid a double copy in RAM).	2026-05-24 15:25:59 -07:00
Comfy Org PR BotandGitHub	0af123022d	Bump comfyui-frontend-package to 1.44.19 (#14074)	2026-05-24 08:27:52 +08:00
Daxiong (Lin)andGitHub	2ca1480f91	chore: update workflow templates to v0.9.82 (#14034)	2026-05-21 11:48:20 -07:00
rattusandGitHub	5aa5ccc9e0	Multi-threaded load of models from disk (big load time speedups & Offload to disk) (CORE-43,CORE-152,CORE-164,CORE-165,CORE-117) (#13802) ... * model_management: disable non-dynamic smart memory Disable smart memory outright for non dynamic models. This is a minor step towards deprecation of --disable-dynamic-vram and the legacy ModelPatcher. This is needed for estimate-free model development, where new models can opt-out of supplying a memory estimate and not have to worry about hard VRAM allocations due to legacy non-dynamic model patchers This is also a general stability increase for a lot of stray use cases where estimates may still be off and going forward we are not going to accurately maintain such estimates. * pinned_memory: implement with aimdo growable buffer Use a single growable buffer so we can do threaded pre-warming on pinned memory. * mm: use aimdo to do transfer from disk to pin Aimdo implements a faster threaded loader. * Add stream host pin buffer for AIMDO casts Introduce per-offload-stream HostBuffer reuse for pinned staging, include it in cast buffer reset synchronization. Defer actual casts that go via this pin path to a separate pass such that the buffer can be allocated monolithically (to avoid cudaHostRegister thrash). * remove old pin path * Implement JIT pinned memory pressure Replace the predictive pin pressure mechanism with JIT PIN memory pressure. * LowVRAMPatch: change to two-phase visit * lora: re-implement as inplace swiss-army-knife operation * prepare for multiple pin sets * implement pinned loras * requirements: comfy-aimdo 0.4.0 * ops: remove unused arg This was defeatured in aimdo iteration * ops: sync the CPU with only the offload stream activity This was syncing with the offload stream which itself is synced with the compute stream, so this was syncing CPU with compute transitively. Define the event to sync it more gently. * pins: implement freeing intermediate for pinned memory Pinning is more important than inactive intermediates and the stream pin buffer is more important than even active intermediates. * execution: implement pin eviction on RAM presure Add back proper pin freeing on RAM pressure * implement pin registration swaps Uncap the windows pins from 50% by extending the pool and have a pressure mechanism to move the pin reservations om demand. This unfortunately implies a GPU sync to do the freeing so significant hysterisis needs to be added to consolidate these pressure events. * cli_args/execution: Implement lower background cache-ram threshold Limit the amount of RAM background intermediates can use, so that switching workflows doesn't degrade performance too much. * make default * bump aimdo * model-patcher: force-cast tiny weights Flux 2 gets crazy stalls due to a mix of tiny and giant weights creating lopsided steam buffer rotations which creates stalls. * ops: refactor in prep for chunking * mm: delegate pin-on-the-way to aimdo Aimdo is able to chunk and slice this on the way for better CPU->GPU overlap. The main advantage is the ability to shorten the bus contention window between previous weight transfer and the next weights vbar fault. * bump aimdo * pinning updates * specify hostbuf max allocation size There a signs of virtual memory exhaustion on some linux systems when throwing 128GB for every little piece. Pass the actual to save aimdo from over-estimates * tests: update execution tests for caching The default caching changed to ram-cache so update these tests accordingly. Remove the LRU 0 test as this also falls through to RAM cache.	2026-05-20 17:03:58 -07:00
Daxiong (Lin)andGitHub	4efe1ddb5c	chore: update workflow templates to v0.9.79 (#14011)	2026-05-20 23:46:20 +08:00
Daxiong (Lin)andGitHub	3d870ff51f	chore: update workflow templates to v0.9.77 (#13895)	2026-05-15 01:25:18 +08:00
3 people	afb4fa15d5	chore: update workflow templates to v0.9.75 (#13877) ... Co-authored-by: Jedrzej Kosinski <kosinkadink1@gmail.com>	2026-05-13 12:33:12 -07:00
Daxiong (Lin)andGitHub	240363f11e	chore: update embedded docs to v0.5.0 (#13865)	2026-05-13 13:33:29 +08:00
Daxiong (Lin)andGitHub	aa9d2fc713	chore: update workflow templates to v0.9.73 (#13822)	2026-05-10 19:10:13 +08:00
3 people	a4b7e3beed	Bump comfyui-frontend-package to 1.43.18 (#13809) ... Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>	2026-05-09 07:53:10 -07:00
3 people	25757a53c9	chore: update workflow templates to v0.9.72 (#13732) ... Co-authored-by: Jedrzej Kosinski <kosinkadink1@gmail.com>	2026-05-07 00:28:18 -07:00