Repository navigation
[mirror #14] Add Linux ROCm and Vulkan legs to the prebuilt pipeline, so AMD hosts stop landing on the CPU build - #1
Open
LeoBorcherding wants to merge 204 commits into
Open
LeoBorcherding wants to merge 204 commits into
LeoBorcherding wants to merge 204 commits into
Conversation
Co-authored-by: leejet <leejet714@gmail.com>
Co-authored-by: leejet <leejet714@gmail.com>
Co-authored-by: leejet <leejet714@gmail.com>
…16 (unslothai#31) The dense collector read src1 as packed f32 and walked ne[1] rows only, so strided row views, batched ne[2]/ne[3] planes and F16/BF16 activations were summed from the wrong memory or skipped. Rows are now addressed through nb[1..3], every plane is accumulated into the weight's column vector, and F16/BF16 rows are converted with the ggml row helpers. Other types, or a strided nb[0], are refused with an error and counted instead of being read as floats. The device copy is sized by ggml_nbytes so a strided view no longer overruns it. The MUL_MAT_ID branch uses the same row reader. Saving the imatrix logs how many collected matmuls saw each layout.
…mpute headroom under --max-vram (unslothai#32) The partitioner filled the first device up to its weight capacity before using the next one, so a module whose weights fit the main device alone was never split, and the compute buffers then had no room under the budget. The 2 GiB compute headroom was also taken only from free memory, never from an explicit --max-vram budget, so a per-device budget was filled entirely with weights. Ranges are now sized in proportion to each device's capacity (the lower of free memory and its --max-vram budget, minus the headroom), as docs/backend.md already described, with the fill-first plan kept as a fallback when the proportional ranges do not fit. The planner is a header-only function with a standalone test (scripts/layer_split_ranges_test.cpp).
Brings in 17 upstream commits up to leejet/stable-diffusion.cpp 8cd731e (FLUX.3 Image Klein and Klein Base, leejet#2143).
…e flash attention Upstream leejet#2132 maps --diffusion-fa to --flash-attn off,diffusion=on and no longer sets the legacy diffusion_flash_attn flag, so the H3 VAE flash-attention rule (CUDA/ROCm by default) never fired after the sync. Key it on the resolved per-module state instead: diffusion FA on, VAE FA off, and no explicit vae= assignment.
Sync with upstream master-968-8cd731e (FLUX.3 Image Klein)
…ubmodule) Upstream master-982 moves the ggml submodule from d25b121 to dc99877 (leejet#2147), so 0001-0007 and 0009 no longer applied. Ported by cherry-picking each patch onto the new submodule commit; conflicts were adjacent additions plus four API moves: load_tile lost its swz template argument, launch_fattn gained async_kv_preload before the occupancy arguments, rms_norm_f32 now takes nchannels/nsamples and a do_scale/scale_out pair, and ggml_cuda_should_use_mmvf takes warp_size.
Merge upstream master-982 (MOSS-Image) and port the ggml patches to dc99877
…ive-branch skip moss_uncond_skip=false turns the built-in uncond skip off; moss_uncond_skip_start / moss_uncond_skip_end bound it in sigma-percent of the shifted flow schedule (ComfyUI percent_to_sigma, inclusive). Without these keys every step skips block 22 as before. Logs the number of skipped negative passes per sampling run.
…p set changes EasyCache / UCache / CacheDIT key a condition's cached output by its pointer only, so at the moss_uncond_skip window edge a reused &uncond diff could come from the other skip configuration. Track the effective negative-branch skip per pass and invalidate &uncond in the step cache when it flips. Only runs with a window set; the default path is unchanged. Verbose logs say whether each windowed MOSS uncond pass was computed or served from cache.
MOSS-Image: extra sample args to window or disable the block-22 negative-branch skip
# Conflicts: # .github/workflows/unsloth-sd-prebuilt.yml # docs/backend.md # scripts/unsloth/package_bundle.py # src/conditioning/conditioner.hpp # src/model_manager.cpp
This branch has not been deployed
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Mirror of unslothai#14 at
d2ae8f8for Codex and the dispatch-only GPU legs. Nothing here goes back to the source branch.