The combination of Intel Xeon E31240 and Vulkan does not work.

[olegosidorini]

I compiled the application with the configuration
cmake .. -DSD_VULKAN=ON -DGGML_AVX2=OFF -DGGML_FMA=OFF
in order to run it on an Intel Xeon E31240, and it does not work.
As a result of running the test example from the website, I get a black or yellow rectangle, both on an Nvidia GTX 1070 and on an AMD Radeon Pro VII.

[olegosidorini]

[DEBUG] main.cpp:515 - version: stable-diffusion.cpp version master-525-d6dd6d7, commit d6dd6d7
[DEBUG] main.cpp:516 - System Info:
SSE3 = 1 | AVX = 1 | AVX2 = 0 | AVX512 = 0 | AVX512_VBMI = 0 | AVX512_VNNI = 0 | FMA = 0 | NEON = 0 | ARM_FMA = 0 | F16C = 0 | FP16_VA = 0 | WASM_SIMD = 0 | VSX = 0 |
[DEBUG] main.cpp:517 - SDCliParams {
mode: img_gen,
output_path: "/home/olego/Pictures/out.png",
verbose: true,
color: false,
canny_preprocess: false,
convert_name: false,
preview_method: none,
preview_interval: 1,
preview_path: "preview.png",
preview_fps: 16,
taesd_preview: false,
preview_noisy: false
}
[DEBUG] main.cpp:518 - SDContextParams {
n_threads: 4,
model_path: "",
clip_l_path: "",
clip_g_path: "",
clip_vision_path: "",
t5xxl_path: "",
llm_path: "/home/olego/sd_models/z-image/Qwen3-4B-Instruct-2507-Q4_K_M.gguf",
llm_vision_path: "",
diffusion_model_path: "/home/olego/sd_models/z-image/z-image-Q4_K_M.gguf",
high_noise_diffusion_model_path: "",
vae_path: "/home/olego/sd_models/z-image/ae.safetensors",
taesd_path: "",
esrgan_path: "",
control_net_path: "",
embedding_dir: "",
embeddings: {
}
wtype: NONE,
tensor_type_rules: "",
lora_model_dir: ".",
photo_maker_path: "",
rng_type: cuda,
sampler_rng_type: NONE,
offload_params_to_cpu: true,
enable_mmap: false,
control_net_cpu: false,
clip_on_cpu: false,
vae_on_cpu: false,
flash_attn: false,
diffusion_flash_attn: true,
diffusion_conv_direct: false,
vae_conv_direct: false,
circular: false,
circular_x: false,
circular_y: false,
chroma_use_dit_mask: true,
qwen_image_zero_cond_t: false,
chroma_use_t5_mask: false,
chroma_t5_mask_pad: 1,
prediction: NONE,
lora_apply_mode: auto,
vae_tiling_params: { 0, 0, 0, 0.5, 0, 0 },
force_sdxl_vae_conv_scale: false
}
[DEBUG] main.cpp:519 - SDGenerationParams {
loras: "{
}",
high_noise_loras: "{
}",
prompt: "A cinematic, melancholic photograph of a solitary hooded figure walking through a sprawling, rain-slicked metropolis at night. The city lights are a chaotic blur of neon orange and cool blue, reflecting on the wet asphalt. The scene evokes a sense of being a single component in a vast machine. Superimposed over the image in a sleek, modern, slightly glitched font is the philosophical quote: 'THE CITY IS A CIRCUIT BOARD, AND I AM A BROKEN TRANSISTOR.' -- moody, atmospheric, profound, dark academic",
negative_prompt: "",
clip_skip: -1,
width: 512,
height: 1024,
batch_count: 1,
init_image_path: "",
end_image_path: "",
mask_image_path: "",
control_image_path: "",
ref_image_paths: [],
control_video_path: "",
auto_resize_ref_image: true,
increase_ref_index: false,
pm_id_images_dir: "",
pm_id_embed_path: "",
pm_style_strength: 20,
skip_layers: [7, 8, 9],
sample_params: (txt_cfg: 5.00, img_cfg: 5.00, distilled_guidance: 3.50, slg.layer_count: 3, slg.layer_start: 0.01, slg.layer_end: 0.20, slg.scale: 0.00, scheduler: NONE, sample_method: NONE, sample_steps: 20, eta: 0.00, shifted_timestep: 0, flow_shift: inf),
high_noise_skip_layers: [7, 8, 9],
high_noise_sample_params: (txt_cfg: 7.00, img_cfg: 7.00, distilled_guidance: 3.50, slg.layer_count: 3, slg.layer_start: 0.01, slg.layer_end: 0.20, slg.scale: 0.00, scheduler: NONE, sample_method: NONE, sample_steps: 20, eta: 0.00, shifted_timestep: 0, flow_shift: inf),
custom_sigmas: [],
cache_mode: "",
cache_option: "",
cache: disabled (threshold=1, start=0.15, end=0.95),
moe_boundary: 0.875,
video_frames: 1,
fps: 16,
vace_strength: 1,
strength: 0.75,
control_strength: 0.9,
seed: 42,
upscale_repeats: 1,
upscale_tile_size: 128,
}
[DEBUG] stable-diffusion.cpp:181 - Using Vulkan backend
[DEBUG] ggml_extend.hpp:75 - ggml_vulkan: Found 3 Vulkan devices:
[DEBUG] ggml_extend.hpp:75 - ggml_vulkan: 0 = AMD Radeon Pro VII (RADV VEGA20) (radv) | uma: 0 | fp16: 1 | bf16: 0 | warp size: 64 | shared memory: 65536 | int dot: 1 | matrix cores: none
[DEBUG] ggml_extend.hpp:75 - ggml_vulkan: 1 = NVIDIA GeForce GTX 1070 (NVIDIA) | uma: 0 | fp16: 0 | bf16: 0 | warp size: 32 | shared memory: 49152 | int dot: 1 | matrix cores: none
[DEBUG] ggml_extend.hpp:75 - ggml_vulkan: 2 = NVIDIA P104-100 (NVIDIA) | uma: 0 | fp16: 0 | bf16: 0 | warp size: 32 | shared memory: 49152 | int dot: 1 | matrix cores: none
[INFO ] stable-diffusion.cpp:202 - Vulkan: Using device 0
[INFO ] stable-diffusion.cpp:267 - loading diffusion model from '/home/olego/sd_models/z-image/z-image-Q4_K_M.gguf'
[INFO ] model.cpp:366 - load /home/olego/sd_models/z-image/z-image-Q4_K_M.gguf using gguf format
[DEBUG] model.cpp:412 - init from '/home/olego/sd_models/z-image/z-image-Q4_K_M.gguf'
[INFO ] stable-diffusion.cpp:314 - loading llm from '/home/olego/sd_models/z-image/Qwen3-4B-Instruct-2507-Q4_K_M.gguf'
[INFO ] model.cpp:366 - load /home/olego/sd_models/z-image/Qwen3-4B-Instruct-2507-Q4_K_M.gguf using gguf format
[DEBUG] model.cpp:412 - init from '/home/olego/sd_models/z-image/Qwen3-4B-Instruct-2507-Q4_K_M.gguf'
[INFO ] stable-diffusion.cpp:328 - loading vae from '/home/olego/sd_models/z-image/ae.safetensors'
[INFO ] model.cpp:369 - load /home/olego/sd_models/z-image/ae.safetensors using safetensors format
[DEBUG] model.cpp:503 - init from '/home/olego/sd_models/z-image/ae.safetensors', prefix = 'vae.'
[INFO ] stable-diffusion.cpp:345 - Version: Z-Image
[INFO ] stable-diffusion.cpp:373 - Weight type stat: f32: 634 | q4_K: 303 | q5_K: 58 | q6_K: 72 | bf16: 28
[INFO ] stable-diffusion.cpp:374 - Conditioner weight type stat: f32: 145 | q4_K: 216 | q6_K: 37
[INFO ] stable-diffusion.cpp:375 - Diffusion model weight type stat: f32: 245 | q4_K: 87 | q5_K: 58 | q6_K: 35 | bf16: 28
[INFO ] stable-diffusion.cpp:376 - VAE weight type stat: f32: 244
[DEBUG] stable-diffusion.cpp:378 - ggml tensor size = 400 bytes
[DEBUG] llm.hpp:286 - merges size 151387
[DEBUG] llm.hpp:318 - vocab size: 151669
[DEBUG] llm.hpp:1136 - llm: num_layers = 36, vocab_size = 151936, hidden_size = 2560, intermediate_size = 9728
[DEBUG] ggml_extend.hpp:1990 - qwen3 params backend buffer size = 3555.38 MB(RAM) (398 tensors)
[DEBUG] ggml_extend.hpp:1990 - z_image params backend buffer size = 4833.70 MB(RAM) (453 tensors)
[DEBUG] ggml_extend.hpp:1990 - vae params backend buffer size = 94.57 MB(RAM) (138 tensors)
[INFO ] stable-diffusion.cpp:752 - Using flash attention in the diffusion model
[DEBUG] stable-diffusion.cpp:781 - loading weights
[DEBUG] model.cpp:1350 - using 4 threads for model loading
[DEBUG] model.cpp:1372 - loading tensors from /home/olego/sd_models/z-image/z-image-Q4_K_M.gguf
|====================> | 453/1095 - 37.72it/s
[DEBUG] model.cpp:1372 - loading tensors from /home/olego/sd_models/z-image/Qwen3-4B-Instruct-2507-Q4_K_M.gguf
|======================================> | 851/1095 - 47.60it/s
[DEBUG] model.cpp:1372 - loading tensors from /home/olego/sd_models/z-image/ae.safetensors
|==================================================| 1095/1095 - 59.25it/s
[INFO ] model.cpp:1592 - loading tensors completed, taking 18.48s (process: 0.00s, read: 17.74s, memcpy: 0.00s, convert: 0.40s, copy_to_backend: 0.00s)
[DEBUG] stable-diffusion.cpp:816 - finished loaded file
[INFO ] stable-diffusion.cpp:874 - total params memory size = 8483.65MB (VRAM 8483.65MB, RAM 0.00MB): text_encoders 3555.38MB(VRAM), diffusion_model 4833.70MB(VRAM), vae 94.57MB(VRAM), controlnet 0.00MB(VRAM), pmid 0.00MB(VRAM)
[INFO ] stable-diffusion.cpp:955 - running in FLOW mode
[DEBUG] stable-diffusion.cpp:3639 - generate_image 512x1024
[INFO ] stable-diffusion.cpp:3675 - sampling using Euler method
[INFO ] denoiser.hpp:494 - get_sigmas with discrete scheduler
[INFO ] stable-diffusion.cpp:3802 - TXT2IMG
[DEBUG] conditioner.hpp:1864 - parse '<|im_start|>user
A cinematic, melancholic photograph of a solitary hooded figure walking through a sprawling, rain-slicked metropolis at night. The city lights are a chaotic blur of neon orange and cool blue, reflecting on the wet asphalt. The scene evokes a sense of being a single component in a vast machine. Superimposed over the image in a sleek, modern, slightly glitched font is the philosophical quote: 'THE CITY IS A CIRCUIT BOARD, AND I AM A BROKEN TRANSISTOR.' -- moody, atmospheric, profound, dark academic<|im_end|>
<|im_start|>assistant
' to [['<|im_start|>user
', 1], ['A cinematic, melancholic photograph of a solitary hooded figure walking through a sprawling, rain-slicked metropolis at night. The city lights are a chaotic blur of neon orange and cool blue, reflecting on the wet asphalt. The scene evokes a sense of being a single component in a vast machine. Superimposed over the image in a sleek, modern, slightly glitched font is the philosophical quote: 'THE CITY IS A CIRCUIT BOARD, AND I AM A BROKEN TRANSISTOR.' -- moody, atmospheric, profound, dark academic', 1], ['<|im_end|>
<|im_start|>assistant
', 1], ]
[DEBUG] llm.hpp:260 - split prompt "<|im_start|>user
" to tokens ["<|im_start|>", "user", "Ċ", ]
[DEBUG] llm.hpp:260 - split prompt "A cinematic, melancholic photograph of a solitary hooded figure walking through a sprawling, rain-slicked metropolis at night. The city lights are a chaotic blur of neon orange and cool blue, reflecting on the wet asphalt. The scene evokes a sense of being a single component in a vast machine. Superimposed over the image in a sleek, modern, slightly glitched font is the philosophical quote: 'THE CITY IS A CIRCUIT BOARD, AND I AM A BROKEN TRANSISTOR.' -- moody, atmospheric, profound, dark academic" to tokens ["A", "Ġcinematic", ",", "Ġmelanch", "olic", "Ġphotograph", "Ġof", "Ġa", "Ġsolitary", "Ġhood", "ed", "Ġfigure", "Ġwalking", "Ġthrough", "Ġa", "Ġsprawling", ",", "Ġrain", "-s", "lick", "ed", "Ġmet", "ropolis", "Ġat", "Ġnight", ".", "ĠThe", "Ġcity", "Ġlights", "Ġare", "Ġa", "Ġchaotic", "Ġblur", "Ġof", "Ġneon", "Ġorange", "Ġand", "Ġcool", "Ġblue", ",", "Ġreflecting", "Ġon", "Ġthe", "Ġwet", "Ġasphalt", ".", "ĠThe", "Ġscene", "Ġev", "okes", "Ġa", "Ġsense", "Ġof", "Ġbeing", "Ġa", "Ġsingle", "Ġcomponent", "Ġin", "Ġa", "Ġvast", "Ġmachine", ".", "ĠSuper", "im", "posed", "Ġover", "Ġthe", "Ġimage", "Ġin", "Ġa", "Ġsleek", ",", "Ġmodern", ",", "Ġslightly", "Ġglitch", "ed", "Ġfont", "Ġis", "Ġthe", "Ġphilosophical", "Ġquote", ":", "Ġ'", "THE", "ĠCITY", "ĠIS", "ĠA", "ĠC", "IR", "CU", "IT", "ĠBOARD", ",", "ĠAND", "ĠI", "ĠAM", "ĠA", "ĠBRO", "KEN", "ĠTRANS", "IST", "OR", ".'", "Ġ--", "Ġmo", "ody", ",", "Ġatmospheric", ",", "Ġprofound", ",", "Ġdark", "Ġacademic", ]
[DEBUG] llm.hpp:260 - split prompt "<|im_end|>
<|im_start|>assistant
" to tokens ["<|im_end|>", "Ċ", "<|im_start|>", "assistant", "Ċ", ]
[INFO ] ggml_extend.hpp:1901 - qwen3 offload params (3555.38 MB, 398 tensors) to runtime backend (Vulkan0), taking 12.32s
[DEBUG] ggml_extend.hpp:1803 - qwen3 compute buffer size: 13.40 MB(VRAM)
[DEBUG] conditioner.hpp:2149 - computing condition graph completed, taking 21159 ms
[DEBUG] conditioner.hpp:1864 - parse '<|im_start|>user
<|im_end|>
<|im_start|>assistant
' to [['<|im_start|>user
', 1], ['<|im_end|>
<|im_start|>assistant
', 1], ]
[DEBUG] llm.hpp:260 - split prompt "<|im_start|>user
" to tokens ["<|im_start|>", "user", "Ċ", ]
[DEBUG] llm.hpp:260 - split prompt "<|im_end|>
<|im_start|>assistant
" to tokens ["<|im_end|>", "Ċ", "<|im_start|>", "assistant", "Ċ", ]
[INFO ] ggml_extend.hpp:1901 - qwen3 offload params (3555.38 MB, 398 tensors) to runtime backend (Vulkan0), taking 12.32s
[DEBUG] ggml_extend.hpp:1803 - qwen3 compute buffer size: 0.88 MB(VRAM)
[DEBUG] conditioner.hpp:2149 - computing condition graph completed, taking 12378 ms
[INFO ] stable-diffusion.cpp:3381 - get_learned_condition completed, taking 33537 ms
[INFO ] stable-diffusion.cpp:3492 - generating image: 1/1 - seed 42
[INFO ] ggml_extend.hpp:1901 - z_image offload params (4833.70 MB, 453 tensors) to runtime backend (Vulkan0), taking 15.39s
[DEBUG] ggml_extend.hpp:1803 - z_image compute buffer size: 301.23 MB(VRAM)
|==================================================| 20/20 - 9.15s/it
[INFO ] stable-diffusion.cpp:3534 - sampling completed, taking 182.96s
[INFO ] stable-diffusion.cpp:3545 - generating 1 latent images completed, taking 183.41s
[INFO ] stable-diffusion.cpp:3548 - decoding 1 latents
[INFO ] ggml_extend.hpp:1901 - vae offload params ( 94.57 MB, 138 tensors) to runtime backend (Vulkan0), taking 0.27s
[DEBUG] ggml_extend.hpp:1803 - vae compute buffer size: 5024.50 MB(VRAM)
[DEBUG] stable-diffusion.cpp:2791 - computing vae decode graph completed, taking 4.09s
[INFO ] stable-diffusion.cpp:3558 - latent 1 decoded, taking 4.09s
[INFO ] stable-diffusion.cpp:3562 - decode_first_stage completed, taking 4.09s
[INFO ] stable-diffusion.cpp:3914 - generate_image completed in 221.05s
[INFO ] main.cpp:424 - save result image 0 to '/home/olego/Pictures/out.png' (success)
[INFO ] main.cpp:473 - 1/1 images saved

[wbruna]

[DEBUG] ggml_extend.hpp:1803 - vae compute buffer size: 5024.50 MB(VRAM)

This could be caused by Vullkan memory issues. In my setup, I consistently get a uniform-colored image like this when I mistakenly run the generation on my iGPU with parameters appropriate for my video card.

Are you able to generate smaller images, e.g. 512x512? If so, using --vae-conv-direct and/or --vae-tiling may help.