wip on unified cublas integration, add all the small libraries but exclude the large ones

make_old_pyinstaller_cuda.bat

@@ -1,4 +1,4 @@
 echo This file is only for my own usage, please do not use it. I am lazy.
 
 set PATH=d:\\MainApplications\\KoboldAIGPT\\KoboldAI-Horde-Bridge\\python;d:\\MainApplications\\KoboldAIGPT\\KoboldAI-Horde-Bridge\\python\\Scripts;%PATH%
-PyInstaller --noconfirm --onefile --clean --console --icon "./niko.ico" --add-data "./klite.embd;." --add-data "./koboldcpp.dll;." --add-data "./cublas64_11.dll;." --add-data "./cublasLt64_11.dll;." --add-data "./cudart64_110.dll;." --add-data "./msvcp140.dll;." --add-data "./vcruntime140.dll;." --add-data "./vcruntime140_1.dll;." --add-data "./rwkv_vocab.embd;." --add-data "./rwkv_world_vocab.embd;." "./koboldcpp.py" -n "koboldcpp.exe"
+PyInstaller --noconfirm --onefile --clean --console --icon "./nikogreen.ico" --add-data "./klite.embd;." --add-data "./koboldcpp.dll;." --add-data "./koboldcpp_openblas.dll;." --add-data "./koboldcpp_failsafe.dll;." --add-data "./koboldcpp_openblas_noavx2.dll;." --add-data "./libopenblas.dll;." --add-data "./koboldcpp_clblast.dll;." --add-data "./clblast.dll;." --add-data "./koboldcpp_cublas.dll;." --add-data "./cublas64_11.dll;." --add-data "./cublasLt64_11.dll;." --add-data "./cudart64_110.dll;." --add-data "./msvcp140.dll;." --add-data "./vcruntime140.dll;." --add-data "./vcruntime140_1.dll;." --add-data "./rwkv_vocab.embd;." --add-data "./rwkv_world_vocab.embd;." "./koboldcpp.py" -n "koboldcpp.exe"

otherarch/llama_v2.cpp

@@ -1063,6 +1063,8 @@ static void llama_v2_model_load_internal(
 #if defined(GGML_USE_CUBLAS)
     {
         const int n_gpu = std::min(n_gpu_layers, int(hparams.n_layer));
+        if(GetQuantsUnshuffled())
+        {
 
         fprintf(stderr, "%s: [old cublas] offloading %d layers to GPU\n", __func__, n_gpu);
 
@@ -1086,6 +1088,14 @@ static void llama_v2_model_load_internal(
 
         fprintf(stderr, "%s: [old cublas] total VRAM used: %zu MB\n", __func__, vram_total / 1024 / 1024);
         }
+        else
+        {
+            if(n_gpu>0)
+            {
+                printf("\n[WARNING: Old format does not support GPU offloading! It will be deactivated!]\n");
+            }
+        }
+    }
 #elif defined(GGML_USE_CLBLAST)
     {
         const int n_gpu = std::min(n_gpu_layers, int(hparams.n_layer));