All issues

Search results

Mage-Flow-Edit-Turbo ~1.6x slower on ROCm after graph-plan caching change (#1963)

Status: Open.

#2014 In leejet/stable-diffusion.cpp;

· basementops opened on Sep 21, 2026

[Bug] --vae-conv-direct + 1536x2048 decode hits GGML_ASSERT(s02 <= uint32 max) in ggml-cuda binbcast.cu (ROCm; same code on CUDA) — works without the flag and on Vulkan

Status: Open.

#2013 In leejet/stable-diffusion.cpp;

· ShadowLayer90 opened on Sep 21, 2026

[Bug] Windows ROCm (gfx1201): Qwen-Image-2.1 aborts at first DiT MUL_MAT with "invalid argument" — stale hipErrorInvalidValue from rocBLAS querying hipDeviceAttributeExpertSchedMode; GGML_CUDA_FORCE_CUBLAS=1 works around it

Status: Open.

#2012 In leejet/stable-diffusion.cpp;

· ShadowLayer90 opened on Sep 21, 2026

SIGFPE (division by zero) in ggml_conv_3d during Wan2.2 TI2V-5B image-to-video generation

Status: Open.

#2009 In leejet/stable-diffusion.cpp;

· basementops opened on Sep 20, 2026

[Bug] ROCm/gfx1100: qwen_image models fail with "unspecified launch failure" at first compute after #1994 (works on 6b3edaa)

Status: Open.

#2008 In leejet/stable-diffusion.cpp;

· Radu0120 opened on Sep 20, 2026

[Bug] sd-server decodes ref_images at the request's output size, while sd-cli keeps the reference's own size — models conditioned on a low-resolution reference cannot be served

bugSomething isn't workingSomething isn't working

Status: Open.

#2004 In leejet/stable-diffusion.cpp;

· larochef opened on Sep 20, 2026

[Bug] PiD 1.5: --diffusion-fa zeroes whole 8×8 patch tokens above 2560², giving a black frame from the second step on (ROCm / gfx1151)

bugSomething isn't workingSomething isn't working

Status: Open.

#2003 In leejet/stable-diffusion.cpp;

· larochef opened on Sep 20, 2026

[Bug (solved)] mention of AMD linux/driver defaults

bugSomething isn't workingSomething isn't working

Status: Open.

#2000 In leejet/stable-diffusion.cpp;

· Sx1Fl0NdqlkM6N4s2pswXKUI0k9FKTaOckLbNXu opened on Sep 19, 2026

[Bug] vulkan: SIGABRT when matmul grid exceeds maxComputeWorkGroupCount (krea2 / Qwen3VL edit, 2 reference images, n=2703360)

bugSomething isn't workingSomething isn't working

Status: Open.

#1991 In leejet/stable-diffusion.cpp;

· akhenakh opened on Sep 18, 2026

[Apple M5 Max] Z-Image-Turbo (Metal): tensor-API path silently outputs blank white PNGs (exit 0) — GGML_METAL_TENSOR_DISABLE=1 restores correct output

Status: Open.

#1990 In leejet/stable-diffusion.cpp;

· ziyu4huang opened on Sep 18, 2026

CUDA: model manager reports "available 0.00 MB device" on discrete GPUs — klein/z-image fail at weight prep

Status: Open.

#1989 In leejet/stable-diffusion.cpp;

· damenordev opened on Sep 18, 2026

[Feature] API Key support for sd-server

enhancementNew feature or requestNew feature or request

Status: Open.

#1988 In leejet/stable-diffusion.cpp;

· lemonzest79 opened on Sep 17, 2026