cleanup

Signed-off-by: vladmandic <mandic00@live.com>

CHANGELOG.md

@@ -15,17 +15,19 @@ Plus some significant under-the-hood changes to improve code coverage and qualit
 - **Models**
   - [Flux.2 Klein](https://bfl.ai/blog/flux2-klein-towards-interactive-visual-intelligence)  
     Flux.2-Klein is a new family of compact models from BFL in *4B and 9B sizes* and avaialable as *destilled and base* variants  
-    also includes are *SDNQ prequantized variants*  
+    also includes are *sdnq prequantized variants*  
   - [Qwen-Image-2512](https://qwen.ai/blog?id=qwen-image-2512)  
     Qwen-Image successor, significantly reduces the AI-generated look and adds finer natural detailils and improved text rendering  
     available in both *original*, *sdnq-svd prequantized* and *sdnq-dynamic prequantized*  variants  
+    thanks @CalamitousFelicitousness  
   - [LTX-2 19B Dev](https://ltx.io/model/ltx-2)  
     LTX-2 is a new very large 19B parameter video generation model from Lightricks using Gemma-3 text encoder  
-    available for T2I/I2I workflows in original and SDNQ prequantized variants  
+    available for T2I/I2I workflows in original and sdnq prequantized variants  
     *note*: audio generation and upsampling are not yet supported (soon)  
   - [GLM-Image](https://z.ai/blog/glm-image)  
     GLM-image is a new image generation model that adopts a hybrid autoregressive with diffusion decoder architecture  
-    available in both *original* and *sdnq-dynamic prequantized* variants, thanks @CalamitousFelicitousness  
+    available in both *original* and *sdnq-dynamic prequantized* variants  
+    thanks @CalamitousFelicitousness  
     *note*: model requires pre-release versions of `transformers` package:  
     > pip install --upgrade git+https://github.com/huggingface/transformers.git  
     > ./webui.sh --experimental  

installer.py

@@ -1400,6 +1400,7 @@ def set_environment():
     log.debug('Setting environment tuning')
     os.environ.setdefault('ACCELERATE', 'True')
     os.environ.setdefault('ATTN_PRECISION', 'fp16')
+    os.environ.setdefault('ClDeviceGlobalMemSizeAvailablePercent', '100')
     os.environ.setdefault('CUDA_AUTO_BOOST', '1')
     os.environ.setdefault('CUDA_CACHE_DISABLE', '0')
     os.environ.setdefault('CUDA_DEVICE_DEFAULT_PERSISTING_L2_CACHE_PERCENTAGE_LIMIT', '0')
@@ -1410,20 +1411,27 @@ def set_environment():
     os.environ.setdefault('GRADIO_ANALYTICS_ENABLED', 'False')
     os.environ.setdefault('K_DIFFUSION_USE_COMPILE', '0')
     os.environ.setdefault('KINETO_LOG_LEVEL', '3')
+    os.environ.setdefault('NEOReadDebugKeys', '1')
     os.environ.setdefault('NUMEXPR_MAX_THREADS', '16')
     os.environ.setdefault('PYTHONHTTPSVERIFY', '0')
+    os.environ.setdefault('PYTORCH_ENABLE_MPS_FALLBACK', '1')
+    os.environ.setdefault('PYTORCH_ENABLE_XPU_FALLBACK', '1')
+    os.environ.setdefault('RUNAI_STREAMER_CHUNK_BYTESIZE', '2097152')
+    os.environ.setdefault('RUNAI_STREAMER_LOG_LEVEL', 'DEBUG' if os.environ.get('SD_LOAD_DEBUG') else 'WARNING')
+    os.environ.setdefault('RUNAI_STREAMER_MEMORY_LIMIT', '-1')
     os.environ.setdefault('SAFETENSORS_FAST_GPU', '1')
+    os.environ.setdefault('SYCL_CACHE_PERSISTENT', '1')
     os.environ.setdefault('TF_CPP_MIN_LOG_LEVEL', '2')
     os.environ.setdefault('TF_ENABLE_ONEDNN_OPTS', '0')
+    os.environ.setdefault('TOKENIZERS_PARALLELISM', '0')
     os.environ.setdefault('TORCH_CUDNN_V8_API_ENABLED', '1')
     os.environ.setdefault('TORCH_FORCE_NO_WEIGHTS_ONLY_LOAD', '1')
+    os.environ.setdefault('TORCH_ROCM_AOTRITON_ENABLE_EXPERIMENTAL', '1')
+    os.environ.setdefault('UR_L0_ENABLE_RELAXED_ALLOCATION_LIMITS', '1')
     os.environ.setdefault('USE_TORCH', '1')
     os.environ.setdefault('UV_INDEX_STRATEGY', 'unsafe-any-match')
     os.environ.setdefault('UV_NO_BUILD_ISOLATION', '1')
     os.environ.setdefault('UVICORN_TIMEOUT_KEEP_ALIVE', '60')
-    os.environ.setdefault('RUNAI_STREAMER_CHUNK_BYTESIZE', '2097152')
-    os.environ.setdefault('RUNAI_STREAMER_MEMORY_LIMIT', '-1')
-    os.environ.setdefault('RUNAI_STREAMER_LOG_LEVEL', 'DEBUG' if os.environ.get('SD_LOAD_DEBUG') else 'WARNING')
     allocator = f'garbage_collection_threshold:{opts.get("torch_gc_threshold", 80)/100:0.2f},max_split_size_mb:512'
     if opts.get("torch_malloc", "native") == 'cudaMallocAsync':
         allocator += ',backend:cudaMallocAsync'
@@ -1433,14 +1441,6 @@ def set_environment():
     os.environ.setdefault('PYTORCH_CUDA_ALLOC_CONF', allocator)
     os.environ.setdefault('PYTORCH_HIP_ALLOC_CONF', allocator)
     log.debug(f'Torch allocator: "{allocator}"')
-    os.environ.setdefault('TORCH_ROCM_AOTRITON_ENABLE_EXPERIMENTAL', '1')
-    os.environ.setdefault('NEOReadDebugKeys', '1')
-    os.environ.setdefault('ClDeviceGlobalMemSizeAvailablePercent', '100')
-    os.environ.setdefault('SYCL_CACHE_PERSISTENT', '1')
-    os.environ.setdefault('UR_L0_ENABLE_RELAXED_ALLOCATION_LIMITS', '1')
-    os.environ.setdefault('PYTORCH_ENABLE_XPU_FALLBACK', '1')
-    os.environ.setdefault('PYTORCH_ENABLE_MPS_FALLBACK', '1')
-    os.environ.setdefault('TOKENIZERS_PARALLELISM', '0')
 
 
 def check_extensions():

launch.py

@@ -68,8 +68,7 @@ def get_custom_args():
         installer.log.trace(f'Environment: {installer.print_dict(env)}')
     env = [f'{k}={v}' for k, v in os.environ.items() if k.startswith('SD_')]
     ld = [f'{k}={v}' for k, v in os.environ.items() if k.startswith('LD_')]
-    compute = [f'{k}={v}' for k, v in os.environ.items() if 'TORCH' in k or 'CUDA' in k or 'ROCM' in k or 'MIOPEN' in k]
-    installer.log.debug(f'Flags: sd={env} ld={ld} compute={compute}')
+    installer.log.debug(f'Flags: sd={env} ld={ld}')
     rec('args')