nix: init singularity and docker images (#5056)

Exposes a few attributes demonstrating how to build [singularity](https://docs.sylabs.io/guides/latest/user-guide/)/[apptainer](https://apptainer.org/) and Docker images re-using llama.cpp's Nix expression.

Built locally on `x86_64-linux` with `nix build github:someoneserge/llama.cpp/feat/nix/images#llamaPackages.{docker,docker-min,sif,llama-cpp}` and it's fast and effective.

3 files changed, 67 insertions, 0 deletions

diff --git a/.devops/nix/docker.nix b/.devops/nix/docker.nix
new file mode 100644
index 00000000..d607b457
--- /dev/null
+++ b/.devops/nix/docker.nix
@@ -0,0 +1,37 @@
+{
+  lib,
+  dockerTools,
+  buildEnv,
+  llama-cpp,
+  interactive ? true,
+  coreutils,
+}:
+
+# A tar that can be fed into `docker load`:
+#
+# $ nix build .#llamaPackages.docker
+# $ docker load < result
+
+# For details and variations cf.
+# - https://nixos.org/manual/nixpkgs/unstable/#ssec-pkgs-dockerTools-buildLayeredImage
+# - https://discourse.nixos.org/t/a-faster-dockertools-buildimage-prototype/16922
+# - https://nixery.dev/
+
+# Approximate (compressed) sizes, at the time of writing, are:
+#
+# .#llamaPackages.docker: 125M;
+# .#llamaPackagesCuda.docker: 537M;
+# .#legacyPackages.aarch64-linux.llamaPackagesXavier.docker: 415M.
+
+dockerTools.buildLayeredImage {
+  name = llama-cpp.pname;
+  tag = "latest";
+
+  contents =
+    [ llama-cpp ]
+    ++ lib.optionals interactive [
+      coreutils
+      dockerTools.binSh
+      dockerTools.caCertificates
+    ];
+}
diff --git a/.devops/nix/scope.nix b/.devops/nix/scope.nix
index d295995a..78530c9e 100644
--- a/.devops/nix/scope.nix
+++ b/.devops/nix/scope.nix
@@ -12,5 +12,8 @@ lib.makeScope newScope (
   self: {
     inherit llamaVersion;
     llama-cpp = self.callPackage ./package.nix { };
+    docker = self.callPackage ./docker.nix { };
+    docker-min = self.callPackage ./docker.nix { interactive = false; };
+    sif = self.callPackage ./sif.nix { };
   }
 )
diff --git a/.devops/nix/sif.nix b/.devops/nix/sif.nix
new file mode 100644
index 00000000..7535ca0f
--- /dev/null
+++ b/.devops/nix/sif.nix
@@ -0,0 +1,27 @@
+{
+  lib,
+  singularity-tools,
+  llama-cpp,
+  bashInteractive,
+  interactive ? false,
+}:
+
+let
+    optionalInt = cond: x: if cond then x else 0;
+in
+singularity-tools.buildImage rec {
+  inherit (llama-cpp) name;
+  contents = [ llama-cpp ] ++ lib.optionals interactive [ bashInteractive ];
+
+  # These are excessive (but safe) for most variants. Building singularity
+  # images requires superuser privileges, so we build them inside a VM in a
+  # writable image of pre-determined size.
+  #
+  # ROCm is currently affected by https://github.com/NixOS/nixpkgs/issues/276846
+  #
+  # Expected image sizes:
+  # - cpu/blas: 150M,
+  # - cuda, all gencodes: 560M,
+  diskSize = 4096 + optionalInt llama-cpp.useRocm 16384;
+  memSize = diskSize;
+}