From eacee115660997eba33c1c1358ddd99b014e46f6 Mon Sep 17 00:00:00 2001
From: osbm <osbm@osbm.dev>
Date: Wed, 27 Aug 2025 01:18:05 +0300
Subject: [PATCH] its working now

---
 flake.nix | 51 ++++++++++++++++++++++++++++++++++-----------------
 main.cu   | 20 ++++++++++----------
 2 files changed, 44 insertions(+), 27 deletions(-)

diff --git a/flake.nix b/flake.nix
index 249b75b..230130d 100644
--- a/flake.nix
+++ b/flake.nix
@@ -29,23 +29,40 @@
     packages."${system}" = {
       # just one main.cu file for now
       # compile it and place it to $out/bin/cuda-kernel
-      cuda-kernel = pkgs.cudaPackages.backendStdenv.mkDerivation {
-        name = "cuda-kernel";
-        src = ./.;
-        buildInputs = with pkgs; [
-          cudatoolkit
-          cudaPackages.cuda_cudart
-          linuxPackages.nvidia_x11_beta
-        ];
-        cudaSupport = true;
-        requiredSystemFeatures = [ "cuda" ];
-        LD_LIBRARY_PATH = "${pkgs.linuxPackages.nvidia_x11_beta}/lib";
-        buildPhase = ''
-          nvcc -o main main.cu
-          mkdir -p $out/bin
-          mv main $out/bin/cuda-kernel
-        '';
-      };
+      cuda-kernel = let
+        # Use CUDA 12.0 which should be compatible with driver 580.76.05
+        cudaPackages = pkgs.cudaPackages;
+        cudatoolkit = cudaPackages.cudatoolkit;
+      in
+        cudaPackages.backendStdenv.mkDerivation {
+          name = "cuda";
+          src = ./main.cu;
+          unpackPhase = "true";
+          buildInputs = [
+            cudatoolkit
+            cudaPackages.cuda_cudart
+          ];
+          nativeBuildInputs = [
+            cudatoolkit
+            pkgs.autoAddDriverRunpath
+          ];
+          LD_LIBRARY_PATH = "${cudatoolkit}/lib:${cudaPackages.cuda_cudart}/lib";
+          buildPhase = ''
+            echo "GCC Version:"
+            gcc --version
+            echo "NVCC Version:"
+            nvcc --version
+
+            echo "Building CUDA program..."
+            nvcc -arch=sm_89 -o main $src
+            mkdir -p $out/bin
+            cp main $out/bin/
+          '';
+          meta = {
+            mainProgram = "main";
+            platforms = ["x86_64-linux"];
+          };
+        };
       default = self.packages."${system}".cuda-kernel;
     };
   };
diff --git a/main.cu b/main.cu
index d99fd87..6ce6ea2 100644
--- a/main.cu
+++ b/main.cu
@@ -1,13 +1,13 @@
-#include <unistd.h>
 #include <iostream>
+#include "cuda.h"
 
-
-int main(){
-
-    size_t mf, ma;
-    cudaMemGetInfo(&mf, &ma);
-    std::cout << "Free memory (mb): " << mf/1024/1024 << std::endl;
-    std::cout << "Total memory (mb): " << ma/1024/1024 << std::endl;
-
-    return 0;
+int main() {
+  size_t free_t, total_t;
+  cudaError_t err;
+  err = cudaMemGetInfo(&free_t, &total_t);
+  if (err != cudaSuccess) { printf("cudaMalloc d_steps_arr error: %s\n", cudaGetErrorString(err)); return 1; }
+  std::cout << "Free mem: " << free_t / (1024 * 1024) << " MB Total mem: " << total_t / (1024 * 1024) << " MB" << std::endl;
+  std::cout << "Free mem: " << free_t / (1024 * 1024 * 1024) << " GB Total mem: " << total_t / (1024 * 1024 * 1024) << " GB" << std::endl;
+  return 0;
 }
+