Comments (6)
Wait for vulkan backend, rocm is joke on team red.
from stable-diffusion.cpp.
To use rocBLAS with an unsupported GPU on Windows you have to build rocBLAS from source. It's not too hard if you follow the steps:
https://www.reddit.com/r/LocalLLaMA/comments/16d1hi0/guide_build_llamacpp_on_windows_with_amd_gpus_and/
from stable-diffusion.cpp.
For your convenience, I have built the rocBLAS library with gfx803, gfx900, gfx1010, gfx1031, gfx1032 support, using the HIP 5.7.1 SDK.
https://1drv.ms/u/s!ApduJQeHcf7MgfMKUM3PC1T-jqDIIA?e=qnWLHT
from stable-diffusion.cpp.
For your convenience, I have built the rocBLAS library with gfx803, gfx900, gfx1010, gfx1031, gfx1032 support, using the HIP 5.7.1 SDK. https://1drv.ms/u/s!ApduJQeHcf7MgfMKUM3PC1T-jqDIIA?e=qnWLHT
Tanks for the file, I've tried to replace it in /ROCm/5.7/bin, then compiled and execute sd, but still not work:
(regex_error(error_backref): The expression contained an invalid back reference.)
Probably the dll file is not all what is missing in my configuration. So as soon as I have time I will try to compile rocBLAS from source
Yes, the rocBLAS library files are also needed to be placed in /ROCm/5.7/bin, you can get it here:
https://1drv.ms/u/s!ApduJQeHcf7MgfMMIctu47Cs05ROxg?e=jRoh8F
Hope this helps!
from stable-diffusion.cpp.
For your convenience, I have built the rocBLAS library with gfx803, gfx900, gfx1010, gfx1031, gfx1032 support, using the HIP 5.7.1 SDK. https://1drv.ms/u/s!ApduJQeHcf7MgfMKUM3PC1T-jqDIIA?e=qnWLHT
Tanks for the file, I've tried to replace it in /ROCm/5.7/bin, then compiled and execute sd, but still not work: (regex_error(error_backref): The expression contained an invalid back reference.)
Probably the dll file is not all what is missing in my configuration. So as soon as I have time I will try to compile rocBLAS from source
from stable-diffusion.cpp.
For your convenience, I have built the rocBLAS library with gfx803, gfx900, gfx1010, gfx1031, gfx1032 support, using the HIP 5.7.1 SDK. https://1drv.ms/u/s!ApduJQeHcf7MgfMKUM3PC1T-jqDIIA?e=qnWLHT
Tanks for the file, I've tried to replace it in /ROCm/5.7/bin, then compiled and execute sd, but still not work:
(regex_error(error_backref): The expression contained an invalid back reference.)
Probably the dll file is not all what is missing in my configuration. So as soon as I have time I will try to compile rocBLAS from source
Yes, the rocBLAS library files are also needed to be placed in /ROCm/5.7/bin, you can get it here:
https://1drv.ms/u/s!ApduJQeHcf7MgfMMIctu47Cs05ROxg?e=jRoh8F
Hope this helps!
Thanks, now work.
For the moment, times are similar to direct ml (1.8 t/s for sd 1.4 at 512x512 and 3.9 s/t for 1024x1024. without other setting, all default).
For the quality, it seems that the same prompt gives worse results than directml, but I haven't done enough testing to be sure yet
from stable-diffusion.cpp.
Related Issues (20)
- Split the api to support comfyui like workflow HOT 1
- memsize was hardcoded in preprocess_canny function
- Macos binary is using an absolute library path HOT 1
- Vulkan support HOT 1
- [Feature Request] Enable Flash Attention in the released binary
- Enabling Flash Attention completely breaks prompt following ("Dog" and "Cat" generates identical image)
- [Feature Request] Support for SDXS-512, allowing for real-time image generation on the CPU (~0.6 seconds per image) HOT 4
- Much higher RAM usage (2-3 times) compared to FastSDCPU when using the exact same models/settings HOT 2
- Any GUI / webui for this? HOT 7
- Access GPU data of result
- Error when compile with latest GGML: ggml_quantize_chunk HOT 2
- Fixed random generator HOT 2
- TTI(Text 2 Image) inference using stable-diffusion.cpp doesn't work properly on Android phone HOT 2
- Support TCD Sampling method
- using the VAE HOT 3
- Opening model slows down inference HOT 2
- [Future Issue] New CPU SILU does not work with stable-diffusion.cpp HOT 1
- Cuda build with only AVX instruct detection bug. HOT 1
- ability to mix sdxl lighting loras with quantized models HOT 1
- Link error when building HOT 2
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from stable-diffusion.cpp.