Add install.sh script for automated setup #53
+503
−4
Conversation
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
- Initializes CUTLASS git submodule - Validates CUDA toolkit and PyTorch environment - Builds CUDA extensions for sm_80, sm_89, sm_90, sm_120a (Blackwell) - Includes --clean option for fresh builds - Verifies installation on completion 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Open
- Check for Miniconda, offer to install if missing - Create conda environment with Python 3.12 - Install PyTorch nightly with CUDA 13.0 (for RTX 5090/Blackwell) - Install psutil dependency - Initialize CUTLASS git submodule - Build TurboDiffusion CUDA extensions - Install SpargeAttn for sparse attention optimization - Add GPU info verification at end Target: media-msi.covershot.app (RTX 5090) 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
scripts/comfyui-turbo.sh: - Configurable paths for conda, ComfyUI, CUDA - Start/stop/status commands - Logging to /tmp/comfyui_turbo.log - Ready for cron @reboot setup Usage: ./comfyui-turbo.sh # Start ./comfyui-turbo.sh --stop # Stop ./comfyui-turbo.sh --status # Check status 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Add scripts/cache_t5.py to pre-cache T5 embeddings (saves ~11GB VRAM) - Add --cached_embedding and --skip_t5 args to wan2.2_i2v_infer.py - Update install.sh with module symlinks for rcm/imaginaire/ops/SLA - Fix spas_sage_attn import name in install verification This enables 2-pass inference: cache embeddings first, then run inference without loading the 11GB T5 model. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Document memory optimization with pre-cached T5 embeddings - Add memory comparison table (30GB+ vs ~18GB peak VRAM) - Include step-by-step instructions for cache_t5.py usage - Note: cached embedding is ~4MB vs 11GB T5 model 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Auto-detect GPU compute capability - Patch SpargeAttn setup.py to add sm_120 support for RTX 5090 - Build with correct TORCH_CUDA_ARCH_LIST 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Author
|
Offloads DiT models before VAE decode to free VRAM. Enables 720p 81-frame generation on 32GB GPUs. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Collaborator
|
I think the UMT5 model is already released after the text embeddings are computed, and we did observe a decrease in memory usage during our tests. TurboDiffusion/turbodiffusion/inference/wan2.2_i2v_infer.py Lines 85 to 88 in c960373 |
Open
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
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.
🤖 Generated with Claude Code