Merge branch 'master' of https://gitea.thecrazyinsanity.win/TheCrazyInsanity/nixos-configv3

flake.lock

@@ -7,11 +7,11 @@
         ]
       },
       "locked": {
-        "lastModified": 1764998300,
+        "lastModified": 1772060133,
-        "narHash": "sha256-fZatn/KLfHLDXnF0wy7JxXqGaZmGDTVufT4o/AOlj44=",
+        "narHash": "sha256-VuyRptb8v1lVGMlLp4/1vRX3Efwec0CN0S6mKmDPzLg=",
         "owner": "nix-community",
         "repo": "home-manager",
-        "rev": "27a6182347ccae90a88231ae0dc5dfa7d15815bb",
+        "rev": "ce9b6e52500a0ea0ec48f0bbf6d7a3e431d9dfa4",
         "type": "github"
       },
       "original": {
@@ -21,13 +21,38 @@
         "type": "github"
       }
     },
-    "impermanence": {
+    "home-manager_2": {
+      "inputs": {
+        "nixpkgs": [
+          "impermanence",
+          "nixpkgs"
+        ]
+      },
       "locked": {
-        "lastModified": 1737831083,
+        "lastModified": 1768598210,
-        "narHash": "sha256-LJggUHbpyeDvNagTUrdhe/pRVp4pnS6wVKALS782gRI=",
+        "narHash": "sha256-kkgA32s/f4jaa4UG+2f8C225Qvclxnqs76mf8zvTVPg=",
+        "owner": "nix-community",
+        "repo": "home-manager",
+        "rev": "c47b2cc64a629f8e075de52e4742de688f930dc6",
+        "type": "github"
+      },
+      "original": {
+        "owner": "nix-community",
+        "repo": "home-manager",
+        "type": "github"
+      }
+    },
+    "impermanence": {
+      "inputs": {
+        "home-manager": "home-manager_2",
+        "nixpkgs": "nixpkgs"
+      },
+      "locked": {
+        "lastModified": 1769548169,
+        "narHash": "sha256-03+JxvzmfwRu+5JafM0DLbxgHttOQZkUtDWBmeUkN8Y=",
         "owner": "nix-community",
         "repo": "impermanence",
-        "rev": "4b3e914cdf97a5b536a889e939fb2fd2b043a170",
+        "rev": "7b1d382faf603b6d264f58627330f9faa5cba149",
         "type": "github"
       },
       "original": {
@@ -38,11 +63,27 @@
     },
     "nixpkgs": {
       "locked": {
-        "lastModified": 1764950072,
+        "lastModified": 1768564909,
-        "narHash": "sha256-BmPWzogsG2GsXZtlT+MTcAWeDK5hkbGRZTeZNW42fwA=",
+        "narHash": "sha256-Kell/SpJYVkHWMvnhqJz/8DqQg2b6PguxVWOuadbHCc=",
         "owner": "nixos",
         "repo": "nixpkgs",
-        "rev": "f61125a668a320878494449750330ca58b78c557",
+        "rev": "e4bae1bd10c9c57b2cf517953ab70060a828ee6f",
+        "type": "github"
+      },
+      "original": {
+        "owner": "nixos",
+        "ref": "nixos-unstable",
+        "repo": "nixpkgs",
+        "type": "github"
+      }
+    },
+    "nixpkgs_2": {
+      "locked": {
+        "lastModified": 1771848320,
+        "narHash": "sha256-0MAd+0mun3K/Ns8JATeHT1sX28faLII5hVLq0L3BdZU=",
+        "owner": "nixos",
+        "repo": "nixpkgs",
+        "rev": "2fc6539b481e1d2569f25f8799236694180c0993",
         "type": "github"
       },
       "original": {

global/default.nix

@@ -20,6 +20,66 @@
   ];
   nix.package = pkgs.lixPackageSets.stable.lix;
 
+  nixpkgs.config.packageOverrides = pkgs: {
+
+    # Stolen from https://www.nijho.lt/post/llama-nixos/
+    llama-cpp =
+      (pkgs.llama-cpp.override {
+        cudaSupport = true;
+        rocmSupport = false;
+        metalSupport = false;
+        # Enable BLAS for optimized CPU layer performance (OpenBLAS)
+        # This is crucial for models using split-mode or CPU offloading
+        blasSupport = true;
+      }).overrideAttrs
+        (oldAttrs: rec {
+          version = "8184";
+          src = pkgs.fetchFromGitHub {
+            owner = "aagit";
+            repo = "llama.cpp";
+            # tag = "b${version}";
+            rev = "6ebf2e0d00d31acfc1a1fa9662e9a7d38bd07bf7"; # https://github.com/ggml-org/llama.cpp/pull/19970
+            hash = "sha256-xryajW0Cs1d+WDijspMTW21FDaZP9Grkb+uErMQCQ48=";
+            leaveDotGit = true;
+            postFetch = ''
+              git -C "$out" rev-parse --short HEAD > $out/COMMIT
+              find "$out" -name .git -print0 | xargs -0 rm -rf
+            '';
+          };
+          # Must update npm deps hash to match the new version's webui dependencies
+          npmDepsHash = "sha256-FKjoZTKm0ddoVdpxzYrRUmTiuafEfbKc4UD2fz2fb8A=";
+          # Enable native CPU optimizations for massively better CPU performance
+          # This enables AVX, AVX2, AVX-512, FMA, etc. for your specific CPU
+          # NOTE: This is intentionally opposite of nixpkgs (which uses -DGGML_NATIVE=off
+          # for reproducible builds). We sacrifice portability for faster CPU layers.
+          cmakeFlags = (oldAttrs.cmakeFlags or [ ]) ++ [
+            "-DGGML_NATIVE=ON"
+            "-DCMAKE_CUDA_ARCHITECTURES=86" # RTX 3090 - needed since sandbox has no GPU
+          ];
+
+          # Disable Nix's NIX_ENFORCE_NO_NATIVE which strips -march=native flags
+          # See: https://github.com/NixOS/nixpkgs/issues/357736
+          # See: https://github.com/NixOS/nixpkgs/pull/377484 (intentionally contradicts this)
+          preConfigure = ''
+            export NIX_ENFORCE_NO_NATIVE=0
+            ${oldAttrs.preConfigure or ""}
+          '';
+        });
+
+    # llama-swap from GitHub releases
+    llama-swap = pkgs.runCommand "llama-swap" { } ''
+      mkdir -p $out/bin
+      tar -xzf ${
+        pkgs.fetchurl {
+          url = "https://github.com/mostlygeek/llama-swap/releases/download/v190/llama-swap_190_linux_amd64.tar.gz";
+          hash = "sha256-WAfmJ4YiVH/UYq++l2Ut6oLqkd270HgG7eV+6FG/0Oc=";
+        }
+      } -C $out/bin
+      chmod +x $out/bin/llama-swap
+    '';
+
+  };
+
   # 🇺🇸
   i18n.defaultLocale = "en_US.UTF-8";
   i18n.extraLocaleSettings = {
@@ -57,7 +117,7 @@
     jack.enable = true;
   };
 
-  programs.adb.enable = true; # Bit heavy but you never know when you might need it...
+  # programs.adb.enable = true; # Bit heavy but you never know when you might need it...
 
   users.users.laythe = {
     isNormalUser = true;
@@ -110,11 +170,11 @@
   environment.systemPackages = with pkgs; [
     git
     wget
-    wineWowPackages.stable # Heavy but really annoying to not have when you need it
+    wineWow64Packages.stable # Heavy but really annoying to not have when you need it
     winetricks
     gparted
     unrar
-    electrum
+    # electrum
     qpwgraph
     libimobiledevice
     ifuse
@@ -144,6 +204,7 @@
     xmodmap
     nixfmt
     units
+    android-tools
   ];
 
   services.usbmuxd = {

machines/homepc/configuration.nix

@@ -39,10 +39,81 @@
   # };
 
   # This is here because I don't have another computer that could run local AI, and regardless the packages would be different on every one.
+  # TODO: honestly, while I currently only have one pc that can run local AI, might change in the future.
+  # And this config is getting a bit complicated for a single pc config
+  # Should be moved to it's own shit
   environment.systemPackages = with pkgs; [
     ollama-cuda
+    opencode
+    llama-cpp
+    llama-swap
   ];
 
+  services.ollama = {
+    enable = true;
+    package = pkgs.ollama-cuda;
+    environmentVariables = {
+      OLLAMA_NUM_PARALLEL = "1";
+      OLLAMA_FLASH_ATTENTION = "1";
+      OLLAMA_KV_CACHE_TYPE = "q4_0";
+      OLLAMA_CONTEXT_LENGTH = "16384";
+    };
+  };
+
+  # Configure llama-swap as a systemd service
+  systemd.services.llama-swap = {
+    description = "llama-swap - OpenAI compatible proxy with automatic model swapping";
+    after = [ "network.target" ];
+    wantedBy = [ "multi-user.target" ];
+
+    serviceConfig = {
+      Type = "simple";
+      User = "laythe";
+      Group = "users";
+      # Point to your declarative config file
+      ExecStart = "${pkgs.llama-swap}/bin/llama-swap --config /etc/llama-swap/config.yaml --listen 0.0.0.0:9292 --watch-config";
+      Restart = "always";
+      RestartSec = 10;
+
+      # Environment for CUDA support
+      Environment = [
+        "PATH=/run/current-system/sw/bin"
+        "LD_LIBRARY_PATH=/run/opengl-driver/lib:/run/opengl-driver-32/lib"
+      ];
+    };
+  };
+
+  # As long as this is here the models are declarative. llama-server will grab them if not downloaded already.
+  environment.etc."llama-swap/config.yaml".text = ''
+    models:
+      "Qwen3.5-35B-A3B-GGUF":
+        cmd: llama-server --port ''${PORT} -hf unsloth/Qwen3.5-35B-A3B-GGUF:Q4_K_S --ctx-size 128000 --temp 0.6 --top-p 0.95 --top-k 20 --min-p 0.00 -np 1  --cpu-moe --fit-target 256 --ubatch-size 1024 -fa on  --slots --slot-save-path /home/laythe/llamapcache  --jinja -kvu --no-mmproj --swa-checkpoints 32
+        ttl: 2400
+      "Qwen3-1.7B-GGUF":
+        cmd: llama-server --port ''${PORT} -hf unsloth/Qwen3-1.7B-GGUF:Q4_K_S --ctx-size 32768 --temp 0.6 --top-p 0.95 --top-k 20 --min-p 0.00 -fa on --jinja -kvu
+        ttl: 120
+      "Qwen3-8B-GGUF":
+        cmd: llama-server --port ''${PORT} -hf unsloth/Qwen3-8B-GGUF:Q4_K_S --ctx-size 32768 --temp 0.6 --top-p 0.95 --top-k 20 --min-p 0.00 -fa on --jinja -kvu -np 1
+        ttl: 120
+      "Qwen3-4B-Claude-Opus-Distill":
+        cmd: llama-server --port ''${PORT} -hf TeichAI/Qwen3-4B-Thinking-2507-Claude-4.5-Opus-High-Reasoning-Distill-GGUF:Q4_K_M --ctx-size 32768 --temp 0.6 --top-p 0.95 --top-k 20 --min-p 0.00 -fa on --jinja -kvu -np 1 --fit-target 256
+        ttl: 120
+      "Qwen3.5-9B-Thinking":
+        cmd: "llama-server --port ''${PORT} -hf unsloth/Qwen3.5-9B-GGUF:Q4_K_M --ctx-size 32000 --temp 0.6 --top-p 0.95 --top-k 20 --min-p 0.00 -fa on --jinja -kvu -np 1 --fit-target 256 --chat-template-kwargs '{\"enable_thinking\": true}' --no-mmproj"
+        ttl: 120
+      "Qwen3.5-4B-Thinking":
+        cmd: "llama-server --port ''${PORT} -hf unsloth/Qwen3.5-4B-GGUF:Q4_K_M --ctx-size 64000 --temp 0.6 --top-p 0.95 --top-k 20 --min-p 0.00 -fa on --jinja -kvu -np 1 --fit-target 256 --chat-template-kwargs '{\"enable_thinking\": true}'"
+        ttl: 120
+      "Qwen3.5-9B-Non-Thinking":
+        cmd: "llama-server --port ''${PORT} -hf unsloth/Qwen3.5-9B-GGUF:Q4_K_M --ctx-size 32000 --temp 0.7 --top-p 0.8 --top-k 20 --min-p 0.00 -fa on --jinja -kvu -np 1 --fit-target 256 --no-mmproj"
+        ttl: 120
+      "Qwen3.5-4B-Non-Thinking":
+        cmd: "llama-server --port ''${PORT} -hf unsloth/Qwen3.5-4B-GGUF:Q4_K_M --ctx-size 64000 --temp 0.7 --top-p 0.8 --top-k 20 --min-p 0.00 -fa on --jinja -kvu -np 1 --fit-target 256"
+        ttl: 120
+      "Qwen3.5-0.8B-Non-Thinking":
+        cmd: "llama-server --port ''${PORT} -hf unsloth/Qwen3.5-0.8B-GGUF:Q4_K_M --ctx-size 64000 --temp 0.7 --top-p 0.8 --top-k 20 --min-p 0.00 -fa on --jinja -kvu -np 1 --fit-target 256"
+        ttl: 120'';
+
   # Set your time zone.
   # time.timeZone = "Europe/Amsterdam";
 
@@ -113,7 +184,7 @@
   # networking.firewall.allowedTCPPorts = [ ... ];
   # networking.firewall.allowedUDPPorts = [ ... ];
   # Or disable the firewall altogether.
-  # networking.firewall.enable = false;
+  networking.firewall.enable = false;
 
   # Copy the NixOS configuration file and link it from the resulting system
   # (/run/current-system/configuration.nix). This is useful in case you

modules/communication/default.nix

@@ -5,5 +5,6 @@
     jami
     element-desktop
     signal-desktop
+    praat
   ];
 }

modules/development/default.nix

@@ -2,10 +2,13 @@
 {
   environment.systemPackages = with pkgs; [
     antigravity # Free Gemini 3 credits was too good of an offer to not take up, even if it requires selling my soul.
+    chromium
   ];
 
   home-manager.sharedModules = [
-    ({ pkgs, ... }: {
+    (
+      { pkgs, ... }:
+      {
         programs.vscode = {
           enable = true;
           extensions = with pkgs.vscode-extensions; [
@@ -15,6 +18,7 @@
             "editor.formatOnSave" = true;
           };
         };
-    })
+      }
+    )
   ];
 }

modules/fabrication/default.nix

@@ -2,6 +2,6 @@
 {
   environment.systemPackages = with pkgs; [
     kicad
-    cura
+    # cura
   ];
 }

modules/kde/default.nix

@@ -9,6 +9,7 @@
 
   environment.systemPackages = with pkgs; [
     kdePackages.kate
+    kdePackages.filelight
     # kdePackages.kdeconnect-kde
   ];
 

modules/virtualization/default.nix

@@ -14,5 +14,6 @@
   virtualisation.podman = {
     enable = true;
     dockerCompat = true;
+    enableNvidia = true;
   };
 }