Git commit
5114672
Operating System & Version
manjaro, semi updated, not fully, abou t don alst 2 weeks ago
GGML backends
Vulkan
Command-line arguments used
sd-cli -M vid_gen --diffusion-model /home/arp/sd.cpp-webui/models/unet/Wan2_2-TI2V-5B-Turbo-Q4_K_M.gguf --t5xxl /home/arp/sd.cpp-webui/models/text_encoders/umt5-xxl-encoder-Q3_K_M.gguf --taesd /home/arp/sd.cpp-webui/models/taesd/taew2_2.pth -p "" -i /mnt/d/projects/sd.cpp-jsui/user_data/uploads/1784892381262_12_7.png -W 720 -H 480 --steps 1 --sampling-method er_sde --scheduler discrete --cfg-scale 1 -s -1 --cache-mode easycache --video-frames 96 --fps 24 --params-backend "diffusion=cpu,te=cpu,vae=cpu" --eager-load --fa -o /mnt/d/projects/sd.cpp-jsui/outputs/any2video/output_001.mp4 --vae-tiling --max-vram 7.5
Steps to reproduce
run the command
What you expected to happen
it to not oom
What actually happened
it oom'ed asking 17gb vram
Logs / error messages / stack trace
ggml_vulkan: Found 2 Vulkan devices:
ggml_vulkan: 0 = AMD Radeon RX 580 2048SP (RADV POLARIS10) (radv) | uma: 0 | fp16: 0 | bf16: 0 | warp size: 64 | shared memory: 65536 | int dot: 0 | matrix cores: none
ggml_vulkan: 1 = Intel(R) UHD Graphics 630 (CML GT2) (Intel open-source Mesa driver) | uma: 1 | fp16: 1 | bf16: 0 | warp size: 32 | shared memory: 49152 | int dot: 0 | matrix cores: none
[INFO ] media_io.cpp:575 - crop input image from 2304x1792 to 2304x1536, image_path = /mnt/d/projects/sd.cpp-jsui/user_data/uploads/1784892381262_12_7.png
[INFO ] media_io.cpp:592 - resize input image from 2304x1536 to 720x480
[INFO ] stable-diffusion.cpp:717 - loading diffusion model from '/home/arp/sd.cpp-webui/models/unet/Wan2_2-TI2V-5B-Turbo-Q4_K_M.gguf'
[INFO ] model_loader.cpp:236 - load /home/arp/sd.cpp-webui/models/unet/Wan2_2-TI2V-5B-Turbo-Q4_K_M.gguf using gguf format
[ERROR] ggml_extend.hpp:72 - gguf_init_from_reader: tensor 'patch_embedding.weight' has invalid number of dimensions: 5 > 4
[ERROR] ggml_extend.hpp:72 - gguf_init_from_reader: failed to read tensor info
[INFO ] stable-diffusion.cpp:764 - loading t5xxl from '/home/arp/sd.cpp-webui/models/text_encoders/umt5-xxl-encoder-Q3_K_M.gguf'
[INFO ] model_loader.cpp:236 - load /home/arp/sd.cpp-webui/models/text_encoders/umt5-xxl-encoder-Q3_K_M.gguf using gguf format
[INFO ] stable-diffusion.cpp:801 - loading tae from '/home/arp/sd.cpp-webui/models/taesd/taew2_2.pth'
[INFO ] model_loader.cpp:245 - load /home/arp/sd.cpp-webui/models/taesd/taew2_2.pth using torch zip format
[INFO ] stable-diffusion.cpp:854 - Version: Wan 2.2 TI2V
[INFO ] stable-diffusion.cpp:909 - Weight type stat: f32: 592 | f16: 134 | q3_K: 120 | q4_K: 255 | q5_K: 3 | q6_K: 91
[INFO ] stable-diffusion.cpp:910 - Conditioner weight type stat: f32: 73 | q3_K: 120 | q4_K: 45 | q5_K: 3 | q6_K: 1
[INFO ] stable-diffusion.cpp:911 - Diffusion model weight type stat: f32: 519 | f16: 6 | q4_K: 210 | q6_K: 90
[INFO ] stable-diffusion.cpp:912 - VAE weight type stat:
[INFO ] wan.hpp:870 - Wan2.2-TI2V-5B
[INFO ] stable-diffusion.cpp:1391 - using TAE for encoding / decoding
[INFO ] stable-diffusion.cpp:1505 - Using flash attention
[INFO ] stable-diffusion.cpp:1519 - Using flash attention in the diffusion model
|################################### | 825/1195 - 178.10MB/s
|############################################# | 1067/1195 - 175.18MB/s
|##################################################| 1195/1195 - 174.81MB/s
[INFO ] model_loader.cpp:1279 - loading tensors completed, taking 35.51s (read: 173.82s, memcpy: 0.00s, convert: 2.37s, copy_to_backend: 0.00s)
[INFO ] stable-diffusion.cpp:1625 - total params memory size = 9391.23MB (VRAM 0.00MB, RAM 9391.23MB): text_encoders 6091.95MB(RAM), diffusion_model 3277.49MB(RAM), vae 21.79MB(RAM), controlnet 0.00MB(N/A), extensions 0.00MB(N/A)
[INFO ] stable-diffusion.cpp:1741 - running in FLOW mode
[WARN ] stable-diffusion.cpp:3947 - align generation request up 720x480 to 736x480 (multiple=32)
[WARN ] stable-diffusion.cpp:3923 - align video frames from 156 to 153 for Wan 2.2 TI2V
[INFO ] denoiser.hpp:1026 - get_sigmas with discrete scheduler
[INFO ] stable-diffusion.cpp:4171 - sampling using ER-SDE method
[INFO ] stable-diffusion.cpp:5833 - IMG2VID
|==================================================| 2/2 - 15.15it/s
[INFO ] stable-diffusion.cpp:5861 - encode_first_stage completed, taking 154 ms
[INFO ] stable-diffusion.cpp:6005 - get_learned_condition completed, taking 24.54s
[INFO ] stable-diffusion.cpp:6367 - generate_video 736x480x153
[INFO ] sample-cache.cpp:63 - EasyCache enabled - threshold: 0.200, start: 0.15, end: 0.95
|==================================================| 1/1 - 130.25s/it
[INFO ] sample-cache.cpp:309 - EasyCache completed without skipping steps
[INFO ] stable-diffusion.cpp:6465 - sampling completed, taking 130.29s
[INFO ] stable-diffusion.cpp:6608 - generating latent video completed, taking 130.88s
ggml_vulkan: Device memory allocation of size 11041505280 failed.
ggml_vulkan: Requested buffer size exceeds device buffer size limit: ErrorOutOfDeviceMemory
[ERROR] ggml_extend.hpp:72 - ggml_gallocr_reserve_n_impl: failed to allocate Vulkan0 buffer of size 17872957448
[ERROR] ggml_extend.hpp:2166 - taehv: failed to allocate the compute buffer
[ERROR] ggml_extend.hpp:2732 - taehv alloc compute buffer failed
[ERROR] vae.hpp:48 - vae decode compute failed while processing a tile
[ERROR] vae.hpp:212 - vae decode compute failed
[INFO ] stable-diffusion.cpp:6036 - decode_first_stage completed, taking 0.48s
[ERROR] stable-diffusion.cpp:6038 - decode_first_stage failed for video
[ERROR] main.cpp:899 - generate failed
Additional context / environment details
rx 580 2048sp, 8gb vram
Git commit
5114672
Operating System & Version
manjaro, semi updated, not fully, abou t don alst 2 weeks ago
GGML backends
Vulkan
Command-line arguments used
sd-cli -M vid_gen --diffusion-model /home/arp/sd.cpp-webui/models/unet/Wan2_2-TI2V-5B-Turbo-Q4_K_M.gguf --t5xxl /home/arp/sd.cpp-webui/models/text_encoders/umt5-xxl-encoder-Q3_K_M.gguf --taesd /home/arp/sd.cpp-webui/models/taesd/taew2_2.pth -p "" -i /mnt/d/projects/sd.cpp-jsui/user_data/uploads/1784892381262_12_7.png -W 720 -H 480 --steps 1 --sampling-method er_sde --scheduler discrete --cfg-scale 1 -s -1 --cache-mode easycache --video-frames 96 --fps 24 --params-backend "diffusion=cpu,te=cpu,vae=cpu" --eager-load --fa -o /mnt/d/projects/sd.cpp-jsui/outputs/any2video/output_001.mp4 --vae-tiling --max-vram 7.5
Steps to reproduce
run the command
What you expected to happen
it to not oom
What actually happened
it oom'ed asking 17gb vram
Logs / error messages / stack trace
ggml_vulkan: Found 2 Vulkan devices:
ggml_vulkan: 0 = AMD Radeon RX 580 2048SP (RADV POLARIS10) (radv) | uma: 0 | fp16: 0 | bf16: 0 | warp size: 64 | shared memory: 65536 | int dot: 0 | matrix cores: none
ggml_vulkan: 1 = Intel(R) UHD Graphics 630 (CML GT2) (Intel open-source Mesa driver) | uma: 1 | fp16: 1 | bf16: 0 | warp size: 32 | shared memory: 49152 | int dot: 0 | matrix cores: none
[INFO ] media_io.cpp:575 - crop input image from 2304x1792 to 2304x1536, image_path = /mnt/d/projects/sd.cpp-jsui/user_data/uploads/1784892381262_12_7.png
[INFO ] media_io.cpp:592 - resize input image from 2304x1536 to 720x480
[INFO ] stable-diffusion.cpp:717 - loading diffusion model from '/home/arp/sd.cpp-webui/models/unet/Wan2_2-TI2V-5B-Turbo-Q4_K_M.gguf'
[INFO ] model_loader.cpp:236 - load /home/arp/sd.cpp-webui/models/unet/Wan2_2-TI2V-5B-Turbo-Q4_K_M.gguf using gguf format
[ERROR] ggml_extend.hpp:72 - gguf_init_from_reader: tensor 'patch_embedding.weight' has invalid number of dimensions: 5 > 4
[ERROR] ggml_extend.hpp:72 - gguf_init_from_reader: failed to read tensor info
[INFO ] stable-diffusion.cpp:764 - loading t5xxl from '/home/arp/sd.cpp-webui/models/text_encoders/umt5-xxl-encoder-Q3_K_M.gguf'
[INFO ] model_loader.cpp:236 - load /home/arp/sd.cpp-webui/models/text_encoders/umt5-xxl-encoder-Q3_K_M.gguf using gguf format
[INFO ] stable-diffusion.cpp:801 - loading tae from '/home/arp/sd.cpp-webui/models/taesd/taew2_2.pth'
[INFO ] model_loader.cpp:245 - load /home/arp/sd.cpp-webui/models/taesd/taew2_2.pth using torch zip format
[INFO ] stable-diffusion.cpp:854 - Version: Wan 2.2 TI2V
[INFO ] stable-diffusion.cpp:909 - Weight type stat: f32: 592 | f16: 134 | q3_K: 120 | q4_K: 255 | q5_K: 3 | q6_K: 91
[INFO ] stable-diffusion.cpp:910 - Conditioner weight type stat: f32: 73 | q3_K: 120 | q4_K: 45 | q5_K: 3 | q6_K: 1
[INFO ] stable-diffusion.cpp:911 - Diffusion model weight type stat: f32: 519 | f16: 6 | q4_K: 210 | q6_K: 90
[INFO ] stable-diffusion.cpp:912 - VAE weight type stat:
[INFO ] wan.hpp:870 - Wan2.2-TI2V-5B
[INFO ] stable-diffusion.cpp:1391 - using TAE for encoding / decoding
[INFO ] stable-diffusion.cpp:1505 - Using flash attention
[INFO ] stable-diffusion.cpp:1519 - Using flash attention in the diffusion model
|################################### | 825/1195 - 178.10MB/s
|############################################# | 1067/1195 - 175.18MB/s
|##################################################| 1195/1195 - 174.81MB/s
[INFO ] model_loader.cpp:1279 - loading tensors completed, taking 35.51s (read: 173.82s, memcpy: 0.00s, convert: 2.37s, copy_to_backend: 0.00s)
[INFO ] stable-diffusion.cpp:1625 - total params memory size = 9391.23MB (VRAM 0.00MB, RAM 9391.23MB): text_encoders 6091.95MB(RAM), diffusion_model 3277.49MB(RAM), vae 21.79MB(RAM), controlnet 0.00MB(N/A), extensions 0.00MB(N/A)
[INFO ] stable-diffusion.cpp:1741 - running in FLOW mode
[WARN ] stable-diffusion.cpp:3947 - align generation request up 720x480 to 736x480 (multiple=32)
[WARN ] stable-diffusion.cpp:3923 - align video frames from 156 to 153 for Wan 2.2 TI2V
[INFO ] denoiser.hpp:1026 - get_sigmas with discrete scheduler
[INFO ] stable-diffusion.cpp:4171 - sampling using ER-SDE method
[INFO ] stable-diffusion.cpp:5833 - IMG2VID
|==================================================| 2/2 - 15.15it/s
[INFO ] stable-diffusion.cpp:5861 - encode_first_stage completed, taking 154 ms
[INFO ] stable-diffusion.cpp:6005 - get_learned_condition completed, taking 24.54s
[INFO ] stable-diffusion.cpp:6367 - generate_video 736x480x153
[INFO ] sample-cache.cpp:63 - EasyCache enabled - threshold: 0.200, start: 0.15, end: 0.95
|==================================================| 1/1 - 130.25s/it
[INFO ] sample-cache.cpp:309 - EasyCache completed without skipping steps
[INFO ] stable-diffusion.cpp:6465 - sampling completed, taking 130.29s
[INFO ] stable-diffusion.cpp:6608 - generating latent video completed, taking 130.88s
ggml_vulkan: Device memory allocation of size 11041505280 failed.
ggml_vulkan: Requested buffer size exceeds device buffer size limit: ErrorOutOfDeviceMemory
[ERROR] ggml_extend.hpp:72 - ggml_gallocr_reserve_n_impl: failed to allocate Vulkan0 buffer of size 17872957448
[ERROR] ggml_extend.hpp:2166 - taehv: failed to allocate the compute buffer
[ERROR] ggml_extend.hpp:2732 - taehv alloc compute buffer failed
[ERROR] vae.hpp:48 - vae decode compute failed while processing a tile
[ERROR] vae.hpp:212 - vae decode compute failed
[INFO ] stable-diffusion.cpp:6036 - decode_first_stage completed, taking 0.48s
[ERROR] stable-diffusion.cpp:6038 - decode_first_stage failed for video
[ERROR] main.cpp:899 - generate failed
Additional context / environment details
rx 580 2048sp, 8gb vram