dev merge

CHANGELOG.md

@@ -1,5 +1,17 @@
 # Change Log for SD.Next
 
+## Update for 2025-05-12
+
+Curious how your system is performing?  
+Run a built-in benchmark and compare to over 15k unique results world-wide: (Benchmark data)[https://vladmandic.github.io/sd-extension-system-info/pages/benchmark.html]!  
+From slowest 0.02 it/s running on 6th gen CPU without acceleration up to 275 it/s running on tuned GH100 system!  
+
+- **Wiki**  
+  - Updates for: *WSL, ZLUDA, ROCm*  
+- **Compute**  
+  - NNCF: added experimental support for direct INT8 MatMul  
+
+
 ## Update for 2025-05-12
 
 ### Highlights for 2025-05-12

README.md

@@ -1,7 +1,7 @@
 <div align="center">
 <img src="https://github.com/vladmandic/sdnext/raw/master/html/logo-transparent.png" width=200 alt="SD.Next">
 
-**Image Diffusion implementation with advanced features**
+# SD.Next: All-in-one WebUI for AI generative image and video creation
 
 ![Last update](https://img.shields.io/github/last-commit/vladmandic/sdnext?svg=true)
 ![License](https://img.shields.io/github/license/vladmandic/sdnext?svg=true)
@@ -29,9 +29,9 @@ All individual features are not listed here, instead check [ChangeLog](CHANGELOG
 - Multiple UIs!  
   ▹ **Standard | Modern**  
 - Multiple [diffusion models](https://vladmandic.github.io/sdnext-docs/Model-Support/)!  
-- Built-in Control for Text, Image, Batch and video processing!  
+- Built-in Control for Text, Image, Batch and Video processing!  
 - Multiplatform!  
- ▹ **Windows | Linux | MacOS | nVidia | AMD | IntelArc/IPEX | DirectML | OpenVINO | ONNX+Olive | ZLUDA**
+ ▹ **Windows | Linux | MacOS | nVidia CUDA | AMD ROCm | IntelArc/IPEX | DirectML | OpenVINO | ONNX+Olive | ZLUDA**
 - Platform specific autodetection and tuning performed on install  
 - Optimized processing with latest `torch` developments with built-in support for model compile, quantize and compress  
   Compile backends: *Triton | StableFast | DeepCache | OneDiff | TeaCache | etc.*  

javascript/extraNetworks.js

@@ -6,11 +6,11 @@ let totalCards = -1;
 
 const getENActiveTab = () => {
   let tabName = '';
-  if (gradioApp().getElementById('txt2img_prompt').checkVisibility()) return 'txt2img';
-  if (gradioApp().getElementById('img2img_prompt').checkVisibility()) return 'img2img';
-  if (gradioApp().getElementById('control_prompt').checkVisibility()) return 'control';
-  if (gradioApp().getElementById('video_prompt').checkVisibility()) return 'video';
-  if (gradioApp().getElementById('framepack_prompt_row').checkVisibility()) return 'framepack';
+  if (gradioApp().getElementById('txt2img_prompt')?.checkVisibility()) return 'txt2img';
+  if (gradioApp().getElementById('img2img_prompt')?.checkVisibility()) return 'img2img';
+  if (gradioApp().getElementById('control_prompt')?.checkVisibility()) return 'control';
+  if (gradioApp().getElementById('video_prompt')?.checkVisibility()) return 'video';
+  if (gradioApp().getElementById('framepack_prompt_row')?.checkVisibility()) return 'framepack';
   // legacy method
   if (gradioApp().getElementById('tab_txt2img').style.display === 'block') tabName = 'txt2img';
   else if (gradioApp().getElementById('tab_img2img').style.display === 'block') tabName = 'img2img';

modules/lora/lora_apply.py

@@ -151,7 +151,8 @@ def network_add_weights(self: Union[torch.nn.Conv2d, torch.nn.Linear, torch.nn.G
             new_weight = dequant_weight.to(devices.device, dtype=torch.float32) + lora_weights.to(devices.device, dtype=torch.float32)
             self.weight = torch.nn.Parameter(new_weight, requires_grad=False)
             self.pre_ops.pop("0")
-            self = nncf_compress_layer(self, num_bits, is_asym_mode, torch_dtype=devices.dtype, quant_conv=shared.opts.nncf_quantize_conv_layers)
+            self._custom_forward_fn = None
+            self = nncf_compress_layer(self, num_bits, is_asym_mode, torch_dtype=devices.dtype, quant_conv=shared.opts.nncf_quantize_conv_layers, group_size=shared.opts.nncf_compress_weights_group_size, use_int8_matmul=shared.opts.nncf_decompress_int8_matmul)
             self = self.to(device)
             del dequant_weight
         except Exception as e:

modules/model_quant.py

@@ -118,7 +118,11 @@ def create_nncf_config(kwargs = None, allow_nncf: bool = True, module: str = 'Mo
             diffusers.quantizers.auto.AUTO_QUANTIZATION_CONFIG_MAPPING["nncf"] = NNCFConfig
             transformers.quantizers.auto.AUTO_QUANTIZATION_CONFIG_MAPPING["nncf"] = NNCFConfig
 
-            nncf_config = NNCFConfig(weights_dtype=shared.opts.nncf_compress_weights_mode.lower())
+            nncf_config = NNCFConfig(
+                weights_dtype=shared.opts.nncf_compress_weights_mode.lower(),
+                group_size=shared.opts.nncf_compress_weights_group_size,
+                use_int8_matmul=shared.opts.nncf_decompress_int8_matmul,
+            )
             log.debug(f'Quantization: module="{module}" type=nncf dtype={shared.opts.nncf_compress_weights_mode}')
             if kwargs is None:
                 return nncf_config