Batch Mode + Maintain Aspect Ratio + Multi-GPU Random Seed + Fixed Multi-GPU CuSolver Error + Fixed 20-min Load Time + Video Input#31

Open

pftq wants to merge 125 commits intoSkyworkAI:mainfrom

Commits on Apr 21, 2025

add prompt travel
chaojie
committed

Commits on Apr 23, 2025

Commits on Apr 24, 2025

Fixed 20-min load time on multi-gpu due to contention.
pftq
authored
Fixed 20-min load time on multi-gpu due to contention.
pftq
authored
Fixed 20-min load time on multi-gpu due to contention.
pftq
authored
Fixed 20-min load time on multi-gpu due to contention.
pftq
authored
Simplified image loading
pftq
authored
Simplified image loading
pftq
authored
remove hard code reference video overlap_history
[email protected]
committed
Merge branch 'main' into main
chaojie
authored
Fixed 20-min load time on multi-gpu due to contention.
pftq
authored
Fixed 20-min load time on multi-gpu due to contention.
pftq
authored
Fixed 20-min load time on multi-gpu due to contention.
pftq
authored
Fixed 20-min load time on multi-gpu due to contention.
pftq
authored
Fixed 20-min load time on multi-gpu due to contention. Reduced an additional 5-minutes by broadcasting text encoder as well.
pftq
authored
Fixed 20-min load time on multi-gpu due to contention. Reduced an additional 5-minutes by broadcasting text encoder as well.
pftq
authored
Fixed 20-min load time on multi-gpu due to contention. Reduced an additional 5-minutes by broadcasting text encoder as well.
pftq
authored
Fixed 20-min load time on multi-gpu due to contention. Reduced an additional 5-minutes by broadcasting text encoder as well. Fixed VRAM spike from redundant GPU 0 load.
pftq
authored
Fixed 20-min load time on multi-gpu due to contention. Reduced an additional 5-minutes by broadcasting text encoder as well. Fixed VRAM spike from redundant GPU 0 load.
pftq
authored
Fixed 20-min load time on multi-gpu due to contention. Reduced an additional 5-minutes by broadcasting text encoder as well. Fixed VRAM spike from redundant GPU 0 load.
pftq
authored
Smoothed pipeline broadcasting to avoid VRAM spikes.
pftq
authored
Smoothed pipeline broadcasting to avoid VRAM spikes.
pftq
authored
Empty cache after loading to control VRAM use
pftq
authored
Empty cache after loading to control VRAM use
pftq
authored
Empty cache after loading to control VRAM use
pftq
authored

Commits on Apr 25, 2025

Commits on Apr 28, 2025

Commits on May 1, 2025

Commits on May 5, 2025

Shortened filename with less prompt text.
pftq
authored