AMD GPU Segmentation fault in Line 53

[ZenonEl]

Describe the bug
Segmentation fault when starting the Server

To Reproduce
Steps to reproduce the behavior:

start easy-diffusion
./start.sh

Expected behavior
Easy-Diffusion should start and work.

Screenshots
If applicable, add screenshots to help explain your problem.

Desktop (please complete the following information):

• OS: CachyOS (based on the Arch Linux)
• Browser: Brave
• Version: 1.76.74 Chromium: 134.0.6998.89

Hardware:
CPU: AMD Ryzen 5 3500X 6-Core Processor
RAM: 16GB
GPU: AMD Radeon RX 7600
VRAM: 8GB

Additional context
Add any other context about the problem here.

❯ ./start.sh
/usr/bin/curl
/usr/bin/tar
/usr/bin/bzip2
/usr/bin/git
git version 2.48.1
easy-diffusion/installer_files/env/bin/conda
conda 25.1.1


Easy Diffusion - v3

Easy Diffusion's git repository was already installed. Updating from main..
Нет локальных изменений для сохранения
Указатель HEAD сейчас на коммите 1b3d048b Use safetensors as the default model instead of ckpt
Уже на «main»
Эта ветка соответствует «origin/main».
Уже актуально.
easy-diffusion/installer_files/env/lib/python3.9/site-packages/conda/base/context.py:202: FutureWarning: Adding 'defaults' to channel list implicitly is deprecated and will be removed in 25.3.

To remove this warning, please choose a default channel explicitly with conda's regular configuration system, e.g. by adding 'defaults' to the list of channels:

  conda config --add channels defaults

For more information see https://docs.conda.io/projects/conda/en/stable/user-guide/configuration/use-condarc.html

  deprecated.topic(
torchruntime: 1.15.1
torch: 2.5.1+rocm6.2
torchvision: 0.20.1+rocm6.2
sdkit: 2.0.22.8
stable-diffusion-sdkit: 2.1.5
diffusers: 0.28.2
{}


Easy Diffusion installation complete, starting the server!


[INFO] Setting env variable HSA_OVERRIDE_GFX_VERSION=11.0.0
[INFO] Setting env variable HIP_VISIBLE_DEVICES=0
--- GPUs ---
0. GPU(vendor_id='1002', vendor_name='Advanced Micro Devices, Inc. [AMD/ATI]', device_id='7480', device_name='Navi 33 [Radeon RX 7600/7600 XT/7600M XT/7600S/7700S / PRO W7600]', is_discrete=True)

--- RECOMMENDED TORCH PLATFORM ---
rocm6.2

--- CONFIGURATION ---
[INFO] Setting env variable HSA_OVERRIDE_GFX_VERSION=11.0.0
[INFO] Setting env variable HIP_VISIBLE_DEVICES=0
PYTHONPATH=easy-diffusion/installer_files/env/lib/python3.8/site-packages
Python:  easy-diffusion/installer_files/env/bin/python
Version: 3.9.21

Launching uvicorn

08:52:22.159 INFO MainThread started in                                                      server.py:31
easy-diffusion/stable-diffusion
08:52:22.162 INFO MainThread started at 03/17/25 08:52:22                                    server.py:32
easy-diffusion/installer_files/env/lib/python3.9/site-packages/pydantic/_internal/_config.py:345: UserWarning: Valid config keys have changed in V2:
* 'orm_mode' has been renamed to 'from_attributes'
  warnings.warn(message, UserWarning)
stable-diffusion model(s) found.
gfpgan model(s) found.
realesrgan model(s) found.
vae model(s) found.
08:52:22.351 INFO MainThread Start new Rendering Thread on device: cuda               task_manager.py:410
08:52:22.354 INFO cuda Device usage during initialization:                                  runtime.py:35
08:52:22.355 INFO cuda CPU utilization: 0.0%, System RAM used: 5.5 of 15.5 GiB, GPU    memory_utils.py:58
RAM used (cuda:0): 0.1 of 8.0 GiB (peak: 0.0 GiB)
08:52:22.356 INFO cuda Setting cuda as active, with precision: half                 device_manager.py:123
easy-diffusion/installer_files/env/lib/python3.9/site-packages/transformers/utils/generic.py:311: FutureWarning: `torch.utils._pytree._register_pytree_node` is deprecated. Please use `torch.utils._pytree.register_pytree_node` instead.
  torch.utils._pytree._register_pytree_node(
easy-diffusion/installer_files/env/lib/python3.9/site-packages/transformers/utils/generic.py:311: FutureWarning: `torch.utils._pytree._register_pytree_node` is deprecated. Please use `torch.utils._pytree.register_pytree_node` instead.
  torch.utils._pytree._register_pytree_node(
08:52:22.480 INFO cuda loading stable-diffusion model from                                 __init__.py:52
easy-diffusion/models/stable-diffusion/sd-v1-5.safetensors
to device: cuda
No module 'xformers'. Proceeding without it.
easy-diffusion/installer_files/env/lib/python3.9/site-packages/diffusers/models/transformers/transformer_2d.py:34: FutureWarning: `Transformer2DModelOutput` is deprecated and will be removed in version 1.0.0. Importing `Transformer2DModelOutput` from `diffusers.models.transformer_2d` is deprecated and this will be removed in a future version. Please use `from diffusers.models.modeling_outputs import Transformer2DModelOutput`, instead.
  deprecate("Transformer2DModelOutput", "1.0.0", deprecation_message)
08:52:23.436 INFO MainThread Opening browser..                                                 app.py:317
08:52:23.464 INFO cuda loading on diffusers                                               __init__.py:168
08:52:23.465 INFO cuda using config:                                                      __init__.py:170
easy-diffusion/installer_files/env/lib/python3.9/site-packa
ges/sdkit/models/models_db/configs/v1-inference.yaml
08:52:23.477 INFO cuda using attn_precision: fp16                                         __init__.py:186
╭─────────────────────────────────────── Easy Diffusion is ready ───────────────────────────────────────╮
│                                                                                                       │
│ Easy Diffusion is ready to serve requests.                                                            │
│                                                                                                       │
│ A new browser tab should have been opened by now.                                                     │
│ If not, please open your web browser and navigate to http://localhost:9000/                           │
│                                                                                                       │
╰───────────────────────────────────────────────────────────────────────────────────────────────────────╯
2025-03-17T08:52:23.582980386+04:00  INFO browsers: Starting Browsers
2025-03-17T08:52:23.583001806+04:00  INFO browsers: Logging to .local/state/software.Browsers/logs/browsers.log
2025-03-17T08:52:23.583025942+04:00  INFO browsers::communicate: No other instance of Browsers was running
2025-03-17T08:52:23.583029338+04:00  INFO browsers::communicate: Cleaning up previous local socket file
2025-03-17T08:52:23.583052973+04:00  INFO browsers::communicate: Started socket listener for new instances of Browser
2025-03-17T08:52:23.583122755+04:00  INFO basically_main: browsers::browser_repository: Repository: /usr/share/software.Browsers/resources/repository/application-repository.toml
2025-03-17T08:52:23.628621346+04:00  INFO basically_main: browsers::linux::linux_utils: Default for x-scheme-handler/https is 'software.Browsers.desktop'
2025-03-17T08:52:23.64838373+04:00  INFO basically_main: browsers::linux::linux_utils: Default for x-scheme-handler/http is 'software.Browsers.desktop'
2025-03-17T08:52:23.648425639+04:00  INFO basically_main: browsers::utils: Config: .config/software.Browsers/config.json
2025-03-17T08:52:23.648700158+04:00  INFO basically_main: browsers::utils: Config: .config/software.Browsers/config.json

(browsers:115649): GLib-GIO-CRITICAL **: 08:52:23.711: This application can not open files.
2025-03-17T08:52:25.712979261+04:00  INFO basically_main: browsers::gui::ui: Exiting Browsers
Окно или вкладка откроются в текущем сеансе браузера.
You have disabled the safety checker for <class 'diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline'> by passing `safety_checker=None`. Ensure that you abide to the conditions of the Stable Diffusion license and do not expose unfiltered results in services or applications open to the public. Both the diffusers team and Hugging Face strongly recommend to keep the safety filter enabled in all public facing circumstances, disabling it only for use-cases that involve analyzing network behavior or auditing its results. For more information, please have a look at https://github.com/huggingface/diffusers/pull/254 .
easy-diffusion/scripts/on_sd_start.sh: строка 53: 115419 Ошибка сегментирования                   (образ памяти сброшен на диск) python scripts/check_modules.py --launch-uvicorn
Press any key to continue

~easy-diffusion 1m 39s
❯

I also tried this:
Run ./developer_console.sh
Run python -m pip show torch
Run torchruntime info
Run the torchruntime test

But also got a segmentation error when running torchruntime test

❯ ./developer_console.sh
Opening Stable Diffusion UI - Developer Console..

easy-diffusion/installer_files/env/lib/python3.9/site-packages/conda/base/context.py:202: FutureWarning: Adding 'defaults' to channel list implicitly is deprecated and will be removed in 25.3.

To remove this warning, please choose a default channel explicitly with conda's regular configuration system, e.g. by adding 'defaults' to the list of channels:

  conda config --add channels defaults

For more information see https://docs.conda.io/projects/conda/en/stable/user-guide/configuration/use-condarc.html

  deprecated.topic(
Environment Info:
/usr/bin/git
git version 2.48.1
easy-diffusion/installer_files/env/bin/conda
conda 25.1.1

easy-diffusion/installer_files/env/bin/python
Python 3.9.21
PYTHONPATH=easy-diffusion/installer_files/env/lib/python3.8/site-packages

(base) [xxx@cachyos-xxx easy-diffusion]$ python -m pip show torch
Name: torch
Version: 2.5.1+rocm6.2
Summary: Tensors and Dynamic neural networks in Python with strong GPU acceleration
Home-page: https://pytorch.org/
Author: PyTorch Team
Author-email: packages@pytorch.org
License: BSD-3-Clause
Location: easy-diffusion/installer_files/env/lib/python3.9/site-packages
Requires: filelock, fsspec, jinja2, networkx, pytorch-triton-rocm, sympy, typing-extensions
Required-by: accelerate, basicsr, clean-fid, clip-anytorch, compel, controlnet-aux, facexlib, gfpgan, invisible-watermark, k-diffusion, kornia, open-clip-torch, pytorch-lightning, realesrgan, test_tube, timm, torchdiffeq, torchmetrics, torchsde, torchvision
(base) [xxx@cachyos-xxx easy-diffusion]$ torchruntime info
--- GPUs ---
0. GPU(vendor_id='1002', vendor_name='Advanced Micro Devices, Inc. [AMD/ATI]', device_id='7480', device_name='Navi 33 [Radeon RX 7600/7600 XT/7600M XT/7600S/7700S / PRO W7600]', is_discrete=True)

--- RECOMMENDED TORCH PLATFORM ---
rocm6.2

--- CONFIGURATION ---
[INFO] Setting env variable HSA_OVERRIDE_GFX_VERSION=11.0.0
[INFO] Setting env variable HIP_VISIBLE_DEVICES=0
(base) [xxx@cachyos-xxx easy-diffusion]$ torchruntime test
[INFO] Setting env variable HSA_OVERRIDE_GFX_VERSION=11.0.0
[INFO] Setting env variable HIP_VISIBLE_DEVICES=0
--- IMPORT TEST ---
Torch version: 2.5.1+rocm6.2
--- / IMPORT TEST ---

--- DEVICE TEST ---
Installed torch platform: cuda
Device count: 1
Torch device (0): cuda:0
Device name (0): AMD Radeon RX 7600
--- / DEVICE TEST ---

--- MATH TEST ---
On torch device: cpu:0
   Simple math:
     x: tensor([0, 1, 2])
     x + 10: tensor([10, 11, 12])
   Norm:
     Size of x: 480.0 Mb on cpu
     Norm (11066.263671875), took 25.6 ms
On torch device: cuda:0
   Simple math:
Ошибка сегментирования (образ памяти сброшен на диск)
(base) [xxx@cachyos-xxx easy-diffusion]$

Rocm Info:

❯ rocminfo
ROCk module is loaded
=====================
HSA System Attributes
=====================
Runtime Version:         1.1
Runtime Ext Version:     1.6
System Timestamp Freq.:  1000.000000MHz
Sig. Max Wait Duration:  18446744073709551615 (0xFFFFFFFFFFFFFFFF) (timestamp count)
Machine Model:           LARGE
System Endianness:       LITTLE
Mwaitx:                  DISABLED
DMAbuf Support:          YES

==========
HSA Agents
==========
*******
Agent 1
*******
  Name:                    AMD Ryzen 5 3500X 6-Core Processor
  Uuid:                    CPU-XX
  Marketing Name:          AMD Ryzen 5 3500X 6-Core Processor
  Vendor Name:             CPU
  Feature:                 None specified
  Profile:                 FULL_PROFILE
  Float Round Mode:        NEAR
  Max Queue Number:        0(0x0)
  Queue Min Size:          0(0x0)
  Queue Max Size:          0(0x0)
  Queue Type:              MULTI
  Node:                    0
  Device Type:             CPU
  Cache Info:
    L1:                      32768(0x8000) KB
  Chip ID:                 0(0x0)
  ASIC Revision:           0(0x0)
  Cacheline Size:          64(0x40)
  Max Clock Freq. (MHz):   3600
  BDFID:                   0
  Internal Node ID:        0
  Compute Unit:            6
  SIMDs per CU:            0
  Shader Engines:          0
  Shader Arrs. per Eng.:   0
  WatchPts on Addr. Ranges:1
  Memory Properties:
  Features:                None
  Pool Info:
    Pool 1
      Segment:                 GLOBAL; FLAGS: FINE GRAINED
      Size:                    16300016(0xf8b7f0) KB
      Allocatable:             TRUE
      Alloc Granule:           4KB
      Alloc Recommended Granule:4KB
      Alloc Alignment:         4KB
      Accessible by all:       TRUE
    Pool 2
      Segment:                 GLOBAL; FLAGS: EXTENDED FINE GRAINED
      Size:                    16300016(0xf8b7f0) KB
      Allocatable:             TRUE
      Alloc Granule:           4KB
      Alloc Recommended Granule:4KB
      Alloc Alignment:         4KB
      Accessible by all:       TRUE
    Pool 3
      Segment:                 GLOBAL; FLAGS: KERNARG, FINE GRAINED
      Size:                    16300016(0xf8b7f0) KB
      Allocatable:             TRUE
      Alloc Granule:           4KB
      Alloc Recommended Granule:4KB
      Alloc Alignment:         4KB
      Accessible by all:       TRUE
    Pool 4
      Segment:                 GLOBAL; FLAGS: COARSE GRAINED
      Size:                    16300016(0xf8b7f0) KB
      Allocatable:             TRUE
      Alloc Granule:           4KB
      Alloc Recommended Granule:4KB
      Alloc Alignment:         4KB
      Accessible by all:       TRUE
  ISA Info:
*******
Agent 2
*******
  Name:                    gfx1102
  Uuid:                    GPU-XX
  Marketing Name:          AMD Radeon RX 7600
  Vendor Name:             AMD
  Feature:                 KERNEL_DISPATCH
  Profile:                 BASE_PROFILE
  Float Round Mode:        NEAR
  Max Queue Number:        128(0x80)
  Queue Min Size:          64(0x40)
  Queue Max Size:          131072(0x20000)
  Queue Type:              MULTI
  Node:                    1
  Device Type:             GPU
  Cache Info:
    L1:                      32(0x20) KB
    L2:                      2048(0x800) KB
  Chip ID:                 29824(0x7480)
  ASIC Revision:           0(0x0)
  Cacheline Size:          128(0x80)
  Max Clock Freq. (MHz):   2356
  BDFID:                   2048
  Internal Node ID:        1
  Compute Unit:            32
  SIMDs per CU:            2
  Shader Engines:          2
  Shader Arrs. per Eng.:   2
  WatchPts on Addr. Ranges:4
  Coherent Host Access:    FALSE
  Memory Properties:
  Features:                KERNEL_DISPATCH
  Fast F16 Operation:      TRUE
  Wavefront Size:          32(0x20)
  Workgroup Max Size:      1024(0x400)
  Workgroup Max Size per Dimension:
    x                        1024(0x400)
    y                        1024(0x400)
    z                        1024(0x400)
  Max Waves Per CU:        32(0x20)
  Max Work-item Per CU:    1024(0x400)
  Grid Max Size:           4294967295(0xffffffff)
  Grid Max Size per Dimension:
    x                        4294967295(0xffffffff)
    y                        4294967295(0xffffffff)
    z                        4294967295(0xffffffff)
  Max fbarriers/Workgrp:   32
  Packet Processor uCode:: 542
  SDMA engine uCode::      21
  IOMMU Support::          None
  Pool Info:
    Pool 1
      Segment:                 GLOBAL; FLAGS: COARSE GRAINED
      Size:                    8372224(0x7fc000) KB
      Allocatable:             TRUE
      Alloc Granule:           4KB
      Alloc Recommended Granule:2048KB
      Alloc Alignment:         4KB
      Accessible by all:       FALSE
    Pool 2
      Segment:                 GLOBAL; FLAGS: EXTENDED FINE GRAINED
      Size:                    8372224(0x7fc000) KB
      Allocatable:             TRUE
      Alloc Granule:           4KB
      Alloc Recommended Granule:2048KB
      Alloc Alignment:         4KB
      Accessible by all:       FALSE
    Pool 3
      Segment:                 GROUP
      Size:                    64(0x40) KB
      Allocatable:             FALSE
      Alloc Granule:           0KB
      Alloc Recommended Granule:0KB
      Alloc Alignment:         0KB
      Accessible by all:       FALSE
  ISA Info:
    ISA 1
      Name:                    amdgcn-amd-amdhsa--gfx1102
      Machine Models:          HSA_MACHINE_MODEL_LARGE
      Profiles:                HSA_PROFILE_BASE
      Default Rounding Mode:   NEAR
      Default Rounding Mode:   NEAR
      Fast f16:                TRUE
      Workgroup Max Size:      1024(0x400)
      Workgroup Max Size per Dimension:
        x                        1024(0x400)
        y                        1024(0x400)
        z                        1024(0x400)
      Grid Max Size:           4294967295(0xffffffff)
      Grid Max Size per Dimension:
        x                        4294967295(0xffffffff)
        y                        4294967295(0xffffffff)
        z                        4294967295(0xffffffff)
      FBarrier Max Size:       32
*** Done ***

❯

[cmdr2]

Hi @ZenonEl Sorry for the extremely late reply. Can you please try these commands:

1. Run ./dev_console.sh
2. Run python -m pip uninstall torch torchvision
3. Run python -m pip install --upgrade --force torch torchvision --index-url https://download.pytorch.org/whl/rocm6.4
4. Run torchruntime info
5. Run torchruntime test

Does that still fail? This basically upgrades torch to rocm 6.4. Thanks!

[ZenonEl]

Hi, absolutely nothing to worry about! The most important thing is that you responded, so thank you anyway :3
I did what you wrote. The error disappeared, and I am even pleasantly surprised that a solution to this problem has finally been found. For the first time, I was able to start generating an image (standard prompt and standard SD model 1.5), but then another problem appeared HSA_STATUS_ERROR_INVALID_ISA:

New Error

17:28:47.014 INFO cuda Running on diffusers: {'guidance_scale': 7.5, 'generator': <torch._C.Generator object   image_generator.py:470
at 0x7f023806e550>, 'width': 512, 'height': 512, 'num_inference_steps': 25, 'num_images_per_prompt': 1,                              
'callback': <function make_with_diffusers.<locals>.<lambda> at 0x7f023808eaf0>, 'callback_steps': 1,                                 
'prompt_embeds': tensor([[[-0.3884,  0.0229, -0.0523,  ..., -0.4902, -0.3066,  0.0674],                                              
         [ 0.0283, -1.3262,  0.3079,  ..., -0.5259,  0.9785,  0.6660],                                                               
         [ 0.4600,  0.5586,  1.6650,  ..., -1.9502, -1.2285,  0.0107],                                                               
         ...,                                                                                                                        
         [-3.0410, -0.0674, -0.1814,  ...,  0.3965, -0.0195,  0.7656],                                                               
         [-3.0566, -0.1060, -0.1948,  ...,  0.4248, -0.0186,  0.7568],                                                               
         [-2.9844, -0.0840, -0.1721,  ...,  0.4365,  0.0088,  0.7480]]],                                                             
       device='cuda:0', dtype=torch.float16), 'negative_prompt_embeds': tensor([[[-0.3884,  0.0229, -0.0523,                         
..., -0.4902, -0.3066,  0.0674],                                                                                                     
         [-0.3718, -1.4502, -0.3406,  ...,  0.9468,  0.1860, -1.1035],                                                               
         [-0.5107, -1.4629, -0.2922,  ...,  1.0420,  0.0695, -1.0283],                                                               
         ...,                                                                                                                        
         [ 0.5005, -0.9561, -0.6626,  ...,  1.5996, -1.0635, -0.2195],                                                               
         [ 0.4983, -0.9468, -0.6680,  ...,  1.6455, -1.0869, -0.2085],                                                               
         [ 0.4915, -0.8140, -0.4929,  ...,  1.6084, -1.0186, -0.2478]]],                                                             
       device='cuda:0', dtype=torch.float16)}                                                                                        
Soft/easy-diffusion/installer_files/env/lib/python3.9/site-packages/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion.py:874: FutureWarning: `callback` is deprecated and will be removed in version 1.0.0. Passing `callback` as an input argument to `__call__` is deprecated, consider using `callback_on_step_end`
  deprecate(
Soft/easy-diffusion/installer_files/env/lib/python3.9/site-packages/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion.py:880: FutureWarning: `callback_steps` is deprecated and will be removed in version 1.0.0. Passing `callback_steps` as an input argument to `__call__` is deprecated, consider using `callback_on_step_end`
  deprecate(
  0%|                                                                                                         | 0/25 [00:00<?, ?it/s]:0:rocdevice.cpp            :2993: 6661042709 us:  Callback: Queue 0x7f0200400000 aborting with error : HSA_STATUS_ERROR_INVALID_ISA: The instruction set architecture is invalid. code: 0x100f
Soft/easy-diffusion/scripts/on_sd_start.sh: строка 54: 117986 Аварийный останов      (образ памяти сброшен на диск) python scripts/check_modules.py --launch-uvicorn
Press any key to continue

Commands output:

• Run ./dev_console.sh (developer_console.sh)
• Run python -m pip uninstall torch torchvision
• Run python -m pip install --upgrade --force torch torchvision --index-url https://download.pytorch.org/whl/rocm6.4
• Run torchruntime info
• Run torchruntime test

(base) [zenonel@cachyos-zenonel easy-diffusion]$ python -m pip uninstall torch torchvision
Found existing installation: torch 2.5.1+rocm6.2
Uninstalling torch-2.5.1+rocm6.2:
  Would remove:
    Soft/easy-diffusion/installer_files/env/bin/convert-caffe2-to-onnx
    Soft/easy-diffusion/installer_files/env/bin/convert-onnx-to-caffe2
    Soft/easy-diffusion/installer_files/env/bin/torchfrtrace
    Soft/easy-diffusion/installer_files/env/bin/torchrun
    Soft/easy-diffusion/installer_files/env/lib/python3.9/site-packages/functorch/*
    Soft/easy-diffusion/installer_files/env/lib/python3.9/site-packages/torch-2.5.1+rocm6.2.dist-info/*
    Soft/easy-diffusion/installer_files/env/lib/python3.9/site-packages/torch/*
    Soft/easy-diffusion/installer_files/env/lib/python3.9/site-packages/torchgen/*
Proceed (Y/n)? 
  Successfully uninstalled torch-2.5.1+rocm6.2
Found existing installation: torchvision 0.20.1+rocm6.2
Uninstalling torchvision-0.20.1+rocm6.2:
  Would remove:
    Soft/easy-diffusion/installer_files/env/lib/python3.9/site-packages/torchvision-0.20.1+rocm6.2.dist-info/*
    Soft/easy-diffusion/installer_files/env/lib/python3.9/site-packages/torchvision.libs/libjpeg.ceea7512.so.62
    Soft/easy-diffusion/installer_files/env/lib/python3.9/site-packages/torchvision.libs/libpng16.7f72a3c5.so.16
    easy-diffusion/installer_files/env/lib/python3.9/site-packages/torchvision.libs/libwebp.4a54d2c8.so.4
    Soft/easy-diffusion/installer_files/env/lib/python3.9/site-packages/torchvision.libs/libz.5f199d92.so.1
    Soft/easy-diffusion/installer_files/env/lib/python3.9/site-packages/torchvision/*
Proceed (Y/n)? 
  Successfully uninstalled torchvision-0.20.1+rocm6.2
(base) [zenonel@cachyos-zenonel easy-diffusion]$ python -m pip install --upgrade --force torch torchvision --index-url https://download.pytorch.org/whl/rocm6.4
Looking in indexes: https://download.pytorch.org/whl/rocm6.4
Collecting torch
. . .
Successfully installed MarkupSafe-2.1.5 filelock-3.13.1 fsspec-2024.6.1 importlib-metadata-7.1.0 jinja2-3.1.4 mpmath-1.3.0 networkx-3.2.1 numpy-1.26.3 pillow-11.0.0 pytorch-triton-rocm-3.4.0 setuptools-70.2.0 sympy-1.13.3 torch-2.8.0+rocm6.4 torchvision-0.23.0+rocm6.4 typing-extensions-4.12.2 zipp-3.19.2
(base) [zenonel@cachyos-zenonel easy-diffusion]$ torchruntime info
--- GPUs ---
0. GPU(vendor_id='1002', vendor_name='Advanced Micro Devices, Inc. [AMD/ATI]', device_id='7480', device_name='Navi 33 [Radeon RX 7600/7600 XT/7600M XT/7600S/7700S / PRO W7600]', is_discrete=True)

--- RECOMMENDED TORCH PLATFORM ---
rocm6.2

--- CONFIGURATION ---
[INFO] Setting env variable HSA_OVERRIDE_GFX_VERSION=11.0.0
[INFO] Setting env variable HIP_VISIBLE_DEVICES=0
(base) [zentfany@cachyos-zentfany easy-diffusion]$ torchruntime test
[INFO] Setting env variable HSA_OVERRIDE_GFX_VERSION=11.0.0
[INFO] Setting env variable HIP_VISIBLE_DEVICES=0
--- IMPORT TEST ---
Torch version: 2.8.0+rocm6.4
--- / IMPORT TEST ---

--- DEVICE TEST ---
Installed torch platform: cuda
Device count: 1
Torch device (0): cuda:0
Device name (0): AMD Radeon RX 7600
--- / DEVICE TEST ---

--- MATH TEST ---
On torch device: cpu:0
   Simple math:
     x: tensor([0, 1, 2])
     x + 10: tensor([10, 11, 12])
   Norm:
     Size of x: 480.0 Mb on cpu
     Norm (11065.63671875), took 27.4 ms
On torch device: cuda:0
   Simple math:
     x: tensor([0, 1, 2], device='cuda:0')
     x + 10: tensor([10, 11, 12], device='cuda:0')
   Norm:
     Size of x: 480.0 Mb on cuda:0
     Norm (11217.1513671875), took 2.1 ms
--- / MATH TEST ---

--- FUNCTIONAL TEST ---
On torch device: cuda:0
   Tensor operations: PASSED
   Random operations: PASSED
   Reductions: PASSED
   Broadcasting: PASSED
   Functional ReLU: PASSED
   Functional Softmax: PASSED
   Functional Cross Entropy: PASSED
   Functional Sigmoid: PASSED
   Functional Tanh: PASSED
   Functional Dropout: PASSED
   Linear module: PASSED
   Convolution module: PASSED
   BatchNorm module: PASSED
   Multihead Attention: PASSED
--- / FUNCTIONAL TEST ---

(base) [zenonel@cachyos-zenonel easy-diffusion]$ 

Additional data from generation start

Maybe this will help you and there is something to hold on to here. Otherwise, it seems that I did not see any errors or warnings further in the logs after the start of the generation.
Only notifications about the generation parameters. After which the new error, mentioned earlier, appears:

17:27:58.667 INFO cuda using attn_precision: fp16 __init__.py:186
[2] Sandbox: CanCreateUserNamespace() clone() failure: EPERM
Gtk-Message: 17:28:00.185: Failed to load module "xapp-gtk3-module"
Gtk-Message: 17:28:00.185: Failed to load module "canberra-gtk-module"
17:28:06.762 INFO AnyIO worker thread Scanning all model folders for models...                                   model_manager.py:418
17:28:07.035 INFO AnyIO worker thread Scanned 3 models. Nothing infected                                         model_manager.py:429
You have disabled the safety checker for <class 'diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline'> by passing `safety_checker=None`. Ensure that you abide to the conditions of the Stable Diffusion license and do not expose unfiltered results in services or applications open to the public. Both the diffusers team and Hugging Face strongly recommend to keep the safety filter enabled in all public facing circumstances, disabling it only for use-cases that involve analyzing network behavior or auditing its results. For more information, please have a look at https://github.com/huggingface/diffusers/pull/254 .
You have disabled the safety checker for <class 'diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.StableDiffusionImg2ImgPipeline'> by passing `safety_checker=None`. Ensure that you abide to the conditions of the Stable Diffusion license and do not expose unfiltered results in services or applications open to the public. Both the diffusers team and Hugging Face strongly recommend to keep the safety filter enabled in all public facing circumstances, disabling it only for use-cases that involve analyzing network behavior or auditing its results. For more information, please have a look at https://github.com/huggingface/diffusers/pull/254 .
Soft/easy-diffusion/installer_files/env/lib/python3.9/site-packages/diffusers/pipelines/deprecated/stable_diffusion_variants/pipeline_stable_diffusion_inpaint_legacy.py:141: FutureWarning: The class <class 'diffusers.pipelines.deprecated.stable_diffusion_variants.pipeline_stable_diffusion_inpaint_legacy.StableDiffusionInpaintPipelineLegacy'> is deprecated and will be removed in v1.0.0. You can achieve exactly the same functionalityby loading your model into `StableDiffusionInpaintPipeline` instead. See https://github.com/huggingface/diffusers/pull/3533for more information.
  deprecate("legacy is outdated", "1.0.0", deprecation_message, standard_warn=False)
You have disabled the safety checker for <class 'diffusers.pipelines.deprecated.stable_diffusion_variants.pipeline_stable_diffusion_inpaint_legacy.StableDiffusionInpaintPipelineLegacy'> by passing `safety_checker=None`. Ensure that you abide to the conditions of the Stable Diffusion license and do not expose unfiltered results in services or applications open to the public. Both the diffusers team and Hugging Face strongly recommend to keep the safety filter enabled in all public facing circumstances, disabling it only for use-cases that involve analyzing network behavior or auditing its results. For more information, please have a look at https://github.com/huggingface/diffusers/pull/254 .
17:28:12.505 INFO cuda Loaded on diffusers                                                                            __init__.py:434
17:28:12.621 INFO cuda loaded stable-diffusion model from                                                              __init__.py:56
Soft/easy-diffusion/models/stable-diffusion/sd-v1-5.safetensors to device: cuda                             
17:28:31.365 INFO cuda Session 1755955683724 starting task 139657484131632 on AMD Radeon RX 7600                  task_manager.py:283
17:28:31.649 INFO cuda unloaded stable-diffusion model from device: cuda                                               __init__.py:87
17:28:31.651 INFO cuda loading stable-diffusion model from                                                             __init__.py:52
Soft/easy-diffusion/models/stable-diffusion/sd-v1-5.safetensors to device: cuda                             
17:28:32.262 INFO cuda loading on diffusers                                                                           __init__.py:168
17:28:32.263 INFO cuda using config:                                                                                  __init__.py:170
Soft/easy-diffusion/installer_files/env/lib/python3.9/site-packages/sdkit/models/models_db/c                
onfigs/v1-inference.yaml                                                                                                             
17:28:32.275 INFO cuda using attn_precision: fp16                                                                     __init__.py:186
You have disabled the safety checker for <class 'diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline'> by passing `safety_checker=None`. Ensure that you abide to the conditions of the Stable Diffusion license and do not expose unfiltered results in services or applications open to the public. Both the diffusers team and Hugging Face strongly recommend to keep the safety filter enabled in all public facing circumstances, disabling it only for use-cases that involve analyzing network behavior or auditing its results. For more information, please have a look at https://github.com/huggingface/diffusers/pull/254 .
You have disabled the safety checker for <class 'diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.StableDiffusionImg2ImgPipeline'> by passing `safety_checker=None`. Ensure that you abide to the conditions of the Stable Diffusion license and do not expose unfiltered results in services or applications open to the public. Both the diffusers team and Hugging Face strongly recommend to keep the safety filter enabled in all public facing circumstances, disabling it only for use-cases that involve analyzing network behavior or auditing its results. For more information, please have a look at https://github.com/huggingface/diffusers/pull/254 .
You have disabled the safety checker for <class 'diffusers.pipelines.deprecated.stable_diffusion_variants.pipeline_stable_diffusion_inpaint_legacy.StableDiffusionInpaintPipelineLegacy'> by passing `safety_checker=None`. Ensure that you abide to the conditions of the Stable Diffusion license and do not expose unfiltered results in services or applications open to the public. Both the diffusers team and Hugging Face strongly recommend to keep the safety filter enabled in all public facing circumstances, disabling it only for use-cases that involve analyzing network behavior or auditing its results. For more information, please have a look at https://github.com/huggingface/diffusers/pull/254 .
17:28:45.413 INFO cuda Loaded on diffusers                                                                            __init__.py:434
17:28:45.469 INFO cuda loaded stable-diffusion model from                                                              __init__.py:56
Soft/easy-diffusion/models/stable-diffusion/sd-v1-5.safetensors to device: cuda                             
Soft/easy-diffusion/installer_files/env/lib/python3.9/site-packages/pydantic/main.py:463: UserWarning: Pydantic serializer warnings:
  PydanticSerializationUnexpectedValue(Unexpected Value)
  PydanticSerializationUnexpectedValue(Expected `list[float]` - serialized value may not be as expected [input_value=0, input_type=int])
  return self.__pydantic_serializer__.to_python(
Soft/easy-diffusion/installer_files/env/lib/python3.9/site-packages/pydantic/main.py:463: UserWarning: Pydantic serializer warnings:
  PydanticSerializationUnexpectedValue(Expected `str` - serialized value may not be as expected [input_value=139657484131632, input_type=int])
  return self.__pydantic_serializer__.to_python(
17:28:45.478 INFO cuda request: {'clip_skip': False,                                                             render_images.py:179
 'control_alpha': None,                                                                                                              
 'guidance_scale': 7.5,                                                                                                              
 'height': 512,                                                                                                                      
 'negative_prompt': '',                                                                                                              
 'num_inference_steps': 25,                                                                                                          
 'prompt': 'a photograph of an astronaut riding a horse',                                                                            
 'sampler_name': 'euler_a',                                                                                                          
 'seed': 0,                                                                                                                          
 'tiling': None,                                                                                                                     
 'use_controlnet_model': None,                                                                                                       
 'use_embeddings_model': None,                                                                                                       
 'use_face_correction': None,                                                                                                        
 'use_lora_model': None,                                                                                                             
 'use_stable_diffusion_model': 'sd-v1-5',                                                                                            
 'use_upscale': None,                                                                                                                
 'use_vae_model': '',                                                                                                                
 'width': 512}                                                                                                                       

[cmdr2]

Thanks @ZenonEl Let's try ROCm 5.7, since a few users have reported that it works with 7600:

1. Run ./dev_console.sh
2. Run python -m pip uninstall torch torchvision
3. Run python -m pip install --upgrade --force torch torchvision --index-url https://download.pytorch.org/whl/rocm5.7
4. Run torchruntime info
5. Run torchruntime test

And then try generating an image. Thanks!

[ZenonEl]

Well, maybe I’m just not one of them :(

• Run ./dev_console.sh
• Run python -m pip uninstall torch torchvision
• Run python -m pip install --upgrade --force torch torchvision --index-url https://download.pytorch.org/whl/rocm5.7
• Run torchruntime info
• Run torchruntime test

(base) [zenonel@cachyos-zenonel easy-diffusion]$ python -m pip uninstall torch torchvision
Found existing installation: torch 2.8.0+rocm6.4
Uninstalling torch-2.8.0+rocm6.4:
  Would remove:
    Soft/easy-diffusion/installer_files/env/bin/torchfrtrace
    Soft/easy-diffusion/installer_files/env/bin/torchrun
    Soft/easy-diffusion/installer_files/env/lib/python3.9/site-packages/functorch/*
    Soft/easy-diffusion/installer_files/env/lib/python3.9/site-packages/torch-2.8.0+rocm6.4.dist-info/*
    Soft/easy-diffusion/installer_files/env/lib/python3.9/site-packages/torch/*
    Soft/easy-diffusion/installer_files/env/lib/python3.9/site-packages/torchgen/*
Proceed (Y/n)? 
  Successfully uninstalled torch-2.8.0+rocm6.4
Found existing installation: torchvision 0.23.0+rocm6.4
Uninstalling torchvision-0.23.0+rocm6.4:
  Would remove:
    Soft/easy-diffusion/installer_files/env/lib/python3.9/site-packages/torchvision-0.23.0+rocm6.4.dist-info/*
    Soft/easy-diffusion/installer_files/env/lib/python3.9/site-packages/torchvision.libs/libjpeg.bd6b9199.so.8
    Soft/easy-diffusion/installer_files/env/lib/python3.9/site-packages/torchvision.libs/libpng16.0481ee11.so.16
    Soft/easy-diffusion/installer_files/env/lib/python3.9/site-packages/torchvision.libs/libsharpyuv.b609dd4c.so.0
    Soft/easy-diffusion/installer_files/env/lib/python3.9/site-packages/torchvision.libs/libwebp.58a855fe.so.7
    Soft/easy-diffusion/installer_files/env/lib/python3.9/site-packages/torchvision.libs/libz.622bbd06.so.1
    Soft/easy-diffusion/installer_files/env/lib/python3.9/site-packages/torchvision/*
Proceed (Y/n)? 
  Successfully uninstalled torchvision-0.23.0+rocm6.4
(base) [zenonel@cachyos-zenonel easy-diffusion]$ python -m pip install --upgrade --force torch torchvision --index-url https://download.pytorch.org/whl/rocm5.7
Looking in indexes: https://download.pytorch.org/whl/rocm5.7
Collecting torch
  Downloading https://download.pytorch.org/whl/rocm5.7/torch-2.3.1%2Brocm5.7-cp39-cp39-linux_x86_64.whl (1905.1 MB)
     ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 1.9/1.9 GB 6.6 MB/s  0:03:29
Collecting torchvision
  Downloading https://download.pytorch.org/whl/rocm5.7/torchvision-0.18.1%2Brocm5.7-cp39-cp39-linux_x86_64.whl (65.5 MB)
     ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 65.5/65.5 MB 8.6 MB/s  0:00:07
Collecting filelock (from torch)
  Using cached https://download.pytorch.org/whl/filelock-3.13.1-py3-none-any.whl.metadata (2.8 kB)
Collecting typing-extensions>=4.8.0 (from torch)
  Using cached https://download.pytorch.org/whl/typing_extensions-4.12.2-py3-none-any.whl.metadata (3.0 kB)
Collecting sympy (from torch)
  Using cached https://download.pytorch.org/whl/sympy-1.13.3-py3-none-any.whl.metadata (12 kB)
Collecting networkx (from torch)
  Using cached https://download.pytorch.org/whl/networkx-3.2.1-py3-none-any.whl.metadata (5.2 kB)
Collecting jinja2 (from torch)
  Using cached https://download.pytorch.org/whl/Jinja2-3.1.4-py3-none-any.whl.metadata (2.6 kB)
Collecting fsspec (from torch)
  Using cached https://download.pytorch.org/whl/fsspec-2024.6.1-py3-none-any.whl.metadata (11 kB)
Collecting pytorch-triton-rocm==2.3.1 (from torch)
  Downloading https://download.pytorch.org/whl/pytorch_triton_rocm-2.3.1-cp39-cp39-linux_x86_64.whl (234.3 MB)
     ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 234.3/234.3 MB 9.8 MB/s  0:00:24
Collecting numpy (from torchvision)
  Using cached https://download.pytorch.org/whl/numpy-1.26.3-cp39-cp39-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (18.2 MB)
Collecting pillow!=8.3.*,>=5.3.0 (from torchvision)
  Using cached https://download.pytorch.org/whl/pillow-11.0.0-cp39-cp39-manylinux_2_28_x86_64.whl.metadata (9.1 kB)
Collecting MarkupSafe>=2.0 (from jinja2->torch)
  Using cached https://download.pytorch.org/whl/MarkupSafe-2.1.5-cp39-cp39-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (25 kB)
Collecting mpmath<1.4,>=1.1.0 (from sympy->torch)
  Using cached https://download.pytorch.org/whl/mpmath-1.3.0-py3-none-any.whl (536 kB)
Using cached https://download.pytorch.org/whl/pillow-11.0.0-cp39-cp39-manylinux_2_28_x86_64.whl (4.4 MB)
Using cached https://download.pytorch.org/whl/typing_extensions-4.12.2-py3-none-any.whl (37 kB)
Using cached https://download.pytorch.org/whl/filelock-3.13.1-py3-none-any.whl (11 kB)
Using cached https://download.pytorch.org/whl/fsspec-2024.6.1-py3-none-any.whl (177 kB)
Using cached https://download.pytorch.org/whl/Jinja2-3.1.4-py3-none-any.whl (133 kB)
Using cached https://download.pytorch.org/whl/networkx-3.2.1-py3-none-any.whl (1.6 MB)
Using cached https://download.pytorch.org/whl/sympy-1.13.3-py3-none-any.whl (6.2 MB)
Installing collected packages: mpmath, typing-extensions, sympy, pillow, numpy, networkx, MarkupSafe, fsspec, filelock, pytorch-triton-rocm, jinja2, torch, torchvision
  Attempting uninstall: mpmath
    Found existing installation: mpmath 1.3.0
    Uninstalling mpmath-1.3.0:
      Successfully uninstalled mpmath-1.3.0
  Attempting uninstall: typing-extensions
    Found existing installation: typing_extensions 4.12.2
    Uninstalling typing_extensions-4.12.2:
      Successfully uninstalled typing_extensions-4.12.2
  Attempting uninstall: sympy
    Found existing installation: sympy 1.13.3
    Uninstalling sympy-1.13.3:
      Successfully uninstalled sympy-1.13.3
  Attempting uninstall: pillow
    Found existing installation: pillow 11.0.0
    Uninstalling pillow-11.0.0:
      Successfully uninstalled pillow-11.0.0
  Attempting uninstall: numpy
    Found existing installation: numpy 1.26.3
    Uninstalling numpy-1.26.3:
      Successfully uninstalled numpy-1.26.3
  Attempting uninstall: networkx
    Found existing installation: networkx 3.2.1
    Uninstalling networkx-3.2.1:
      Successfully uninstalled networkx-3.2.1
  Attempting uninstall: MarkupSafe
    Found existing installation: MarkupSafe 2.1.5
    Uninstalling MarkupSafe-2.1.5:
      Successfully uninstalled MarkupSafe-2.1.5
  Attempting uninstall: fsspec
    Found existing installation: fsspec 2024.6.1
    Uninstalling fsspec-2024.6.1:
      Successfully uninstalled fsspec-2024.6.1
  Attempting uninstall: filelock
    Found existing installation: filelock 3.13.1
    Uninstalling filelock-3.13.1:
      Successfully uninstalled filelock-3.13.1
  Attempting uninstall: pytorch-triton-rocm
    Found existing installation: pytorch-triton-rocm 3.4.0
    Uninstalling pytorch-triton-rocm-3.4.0:
      Successfully uninstalled pytorch-triton-rocm-3.4.0
  Attempting uninstall: jinja2
    Found existing installation: Jinja2 3.1.4
    Uninstalling Jinja2-3.1.4:
      Successfully uninstalled Jinja2-3.1.4
Successfully installed MarkupSafe-2.1.5 filelock-3.13.1 fsspec-2024.6.1 jinja2-3.1.4 mpmath-1.3.0 networkx-3.2.1 numpy-1.26.3 pillow-11.0.0 pytorch-triton-rocm-2.3.1 sympy-1.13.3 torch-2.3.1+rocm5.7 torchvision-0.18.1+rocm5.7 typing-extensions-4.12.2
(base) [zenonel@cachyos-zenonel easy-diffusion]$ torchruntime info
--- GPUs ---
0. GPU(vendor_id='1002', vendor_name='Advanced Micro Devices, Inc. [AMD/ATI]', device_id='7480', device_name='Navi 33 [Radeon RX 7600/7600 XT/7600M XT/7600S/7700S / PRO W7600]', is_discrete=True)

--- RECOMMENDED TORCH PLATFORM ---
rocm6.2

--- CONFIGURATION ---
[INFO] Setting env variable HSA_OVERRIDE_GFX_VERSION=11.0.0
[INFO] Setting env variable HIP_VISIBLE_DEVICES=0
(base) [zenonel@cachyos-zenonel easy-diffusion]$ torchruntime test
[INFO] Setting env variable HSA_OVERRIDE_GFX_VERSION=11.0.0
[INFO] Setting env variable HIP_VISIBLE_DEVICES=0
--- IMPORT TEST ---
Traceback (most recent call last):
  Soft/easy-diffusion/installer_files/env/bin/torchruntime", line 7, in <module>
    sys.exit(main())
  Soft/easy-diffusion/installer_files/env/lib/python3.9/site-packages/torchruntime/__main__.py", line 63, in main
    test(subcommand)
  Soft/easy-diffusion/installer_files/env/lib/python3.9/site-packages/torchruntime/utils/torch_test/__init__.py", line 15, in test
    test_fn()
  Soft/easy-diffusion/installer_files/env/lib/python3.9/site-packages/torchruntime/utils/torch_test/__init__.py", line 20, in test_all
    fn()
  Soft/easy-diffusion/installer_files/env/lib/python3.9/site-packages/torchruntime/utils/torch_test/__init__.py", line 26, in test_import
    import torch
  Soft/easy-diffusion/installer_files/env/lib/python3.9/site-packages/torch/__init__.py", line 239, in <module>
    from torch._C import *  # noqa: F403
ImportError: libamdhip64.so: cannot enable executable stack as shared object requires: Invalid argument
(base) [zenonel@cachyos-zenonel easy-diffusion]$

[cmdr2]

Sorry for the extreme delay in replying to this.

Please try Easy Diffusion's new v4 engine, it has significantly better support for AMD (using Vulkan, and the internal engine powering llama.cpp) - https://github.com/easydiffusion/easydiffusion/wiki/v4

ROCm has been quite a nuisance to maintain, so going forward, the v4 engine (with Vulkan) is the recommended way for AMD gpus.

Please feel free to reopen or create a new issue if v4 doesn't work on your GPU, thanks!