Running VAE Tiling with small TAESD buffer will crash

[rmatif]

Just noticed also that enabling VAE tiling while using TAESD will cause a crash for small TAESD buffer. Maybe in this case just ignore the tiling if the buffer is < a certain size

rmatif ➜ /workspaces/stable-diffusion.cpp (master) $ ./build/bin/sd -m models/Realistic_Vision_V6.0_NV_B1_fp16.safetensors --taesd models/taesd.safetensors -p "cute cat" -v -H 256 -W 256 --vae-tiling --cfg-scale 1 --steps 1
Option: 
    n_threads:         4
    mode:              txt2img
    model_path:        models/Realistic_Vision_V6.0_NV_B1_fp16.safetensors
    wtype:             unspecified
    clip_l_path:       
    clip_g_path:       
    t5xxl_path:        
    diffusion_model_path:   
    vae_path:          
    taesd_path:        models/taesd.safetensors
    esrgan_path:       
    controlnet_path:   
    embeddings_path:   
    stacked_id_embeddings_path:   
    input_id_images_path:   
    style ratio:       20.00
    normalize input image :  false
    output_path:       output.png
    init_img:          
    mask_img:          
    control_image:     
    clip on cpu:       false
    controlnet cpu:    false
    vae decoder on cpu:false
    diffusion flash attention:false
    strength(control): 0.90
    prompt:            cute cat
    negative_prompt:   
    min_cfg:           1.00
    cfg_scale:         1.00
    slg_scale:         0.00
    guidance:          3.50
    eta:               0.00
    clip_skip:         -1
    width:             256
    height:            256
    sample_method:     euler_a
    schedule:          default
    sample_steps:      1
    strength(img2img): 0.75
    rng:               cuda
    seed:              42
    batch_count:       1
    vae_tiling:        true
    upscale_repeats:   1
System Info: 
    SSE3 = 1
    AVX = 1
    AVX2 = 1
    AVX512 = 0
    AVX512_VBMI = 0
    AVX512_VNNI = 0
    FMA = 1
    NEON = 0
    ARM_FMA = 0
    F16C = 1
    FP16_VA = 0
    WASM_SIMD = 0
    VSX = 0
[DEBUG] stable-diffusion.cpp:188  - Using CPU backend
[INFO ] stable-diffusion.cpp:197  - loading model from 'models/Realistic_Vision_V6.0_NV_B1_fp16.safetensors'
[INFO ] model.cpp:908  - load models/Realistic_Vision_V6.0_NV_B1_fp16.safetensors using safetensors format
[DEBUG] model.cpp:979  - init from 'models/Realistic_Vision_V6.0_NV_B1_fp16.safetensors'
[INFO ] stable-diffusion.cpp:244  - Version: SD 1.x 
[INFO ] stable-diffusion.cpp:277  - Weight type:                 f16
[INFO ] stable-diffusion.cpp:278  - Conditioner weight type:     f16
[INFO ] stable-diffusion.cpp:279  - Diffusion model weight type: f16
[INFO ] stable-diffusion.cpp:280  - VAE weight type:             f16
[DEBUG] stable-diffusion.cpp:282  - ggml tensor size = 400 bytes
[DEBUG] clip.hpp:171  - vocab size: 49408
[DEBUG] clip.hpp:182  -  trigger word img already in vocab
[DEBUG] ggml_extend.hpp:1174 - clip params backend buffer size =  307.44 MB(RAM) (196 tensors)
[DEBUG] ggml_extend.hpp:1174 - unet params backend buffer size =  1640.25 MB(RAM) (686 tensors)
[DEBUG] stable-diffusion.cpp:419  - loading weights
[DEBUG] model.cpp:1727 - loading tensors from models/Realistic_Vision_V6.0_NV_B1_fp16.safetensors
  |==================================================| 1130/1130 - 500.00it/s
[INFO ] tae.hpp:214  - loading taesd from 'models/taesd.safetensors', decode_only = true
[DEBUG] ggml_extend.hpp:1174 - taesd params backend buffer size =   2.34 MB(RAM) (67 tensors)
[INFO ] model.cpp:908  - load models/taesd.safetensors using safetensors format
[DEBUG] model.cpp:979  - init from 'models/taesd.safetensors'
[DEBUG] model.cpp:1727 - loading tensors from models/taesd.safetensors
  |=========================>                        | 67/134 - 1000.00it/s[INFO ] tae.hpp:236  - taesd model loaded
[INFO ] stable-diffusion.cpp:503  - total params memory size = 1950.03MB (VRAM 2.34MB, RAM 1947.69MB): clip 307.44MB(RAM), unet 1640.25MB(RAM), vae 2.34MB(VRAM), controlnet 0.00MB(VRAM), pmid 0.00MB(RAM)
[INFO ] stable-diffusion.cpp:522  - loading model from 'models/Realistic_Vision_V6.0_NV_B1_fp16.safetensors' completed, taking 0.65s
[INFO ] stable-diffusion.cpp:556  - running in eps-prediction mode
[DEBUG] stable-diffusion.cpp:600  - finished loaded file
[DEBUG] stable-diffusion.cpp:1548 - txt2img 256x256
[DEBUG] stable-diffusion.cpp:1241 - prompt after extract and remove lora: "cute cat"
[INFO ] stable-diffusion.cpp:690  - Attempting to apply 0 LoRAs
[INFO ] stable-diffusion.cpp:1246 - apply_loras completed, taking 0.00s
[DEBUG] conditioner.hpp:357  - parse 'cute cat' to [['cute cat', 1], ]
[DEBUG] clip.hpp:311  - token length: 77
[DEBUG] ggml_extend.hpp:1126 - clip compute buffer size: 1.40 MB(RAM)
[DEBUG] conditioner.hpp:485  - computing condition graph completed, taking 178 ms
[INFO ] stable-diffusion.cpp:1379 - get_learned_condition completed, taking 178 ms
[INFO ] stable-diffusion.cpp:1402 - sampling using Euler A method
[INFO ] stable-diffusion.cpp:1439 - generating image: 1/1 - seed 42
[DEBUG] stable-diffusion.cpp:808  - Sample
[DEBUG] ggml_extend.hpp:1126 - unet compute buffer size: 49.43 MB(RAM)
  |==================================================| 1/1 - 2.69s/it
[INFO ] stable-diffusion.cpp:1478 - sampling completed, taking 2.69s
[INFO ] stable-diffusion.cpp:1486 - generating 1 latent images completed, taking 2.75s
[INFO ] stable-diffusion.cpp:1489 - decoding 1 latents
[DEBUG] ggml_extend.hpp:616  - tile work buffer size: 3.06 MB
[DEBUG] ggml_extend.hpp:1126 - taesd compute buffer size: 480.00 MB(RAM)
[INFO ] ggml_extend.hpp:630  - processing 1 tiles
  |==================================================| 1/1 - 0.00it/s
Segmentation fault (core dumped)