Fix large matrix sizes and add GPU hardware detection warnings

- Updated large_matrices size range to go up to 10000 (was 2000)
- Added sizes: 2500:500:5000, 6000:1000:10000 for GPU benchmarking
- Added warnings when CUDA hardware is detected but CUDA.jl not loaded
- Added warnings when Apple Silicon detected but Metal.jl not loaded
- Improved GPU detection with environment variable and system file checks
- Total of 45 matrix sizes when large_matrices=true, max size 10000

🤖 Generated with [Claude Code](https://claude.ai/code)

Co-Authored-By: Claude <[email protected]>

Author: ChrisRackauckas

lib/LinearSolveAutotune/src/benchmarking.jl

@@ -86,8 +86,9 @@ Get the matrix sizes to benchmark based on the large_matrices flag.
 """
 function get_benchmark_sizes(large_matrices::Bool = false)
     if large_matrices
-        # For GPU benchmarking, include much larger sizes
-        return vcat(4:8:128, 150:50:500, 600:100:1000, 1200:200:2000)
+        # For GPU benchmarking, include much larger sizes up to 10000
+        return vcat(4:8:128, 150:50:500, 600:100:1000,
+            1200:200:2000, 2500:500:5000, 6000:1000:10000)
     else
         # Default sizes similar to existing benchmarks
         return 4:8:500

lib/LinearSolveAutotune/src/gpu_detection.jl

@@ -4,11 +4,23 @@
     is_cuda_available()
 
 Check if CUDA hardware and packages are available.
+Issues warnings if CUDA hardware is detected but packages aren't loaded.
 """
 function is_cuda_available()
     # Check if CUDA extension is loaded
     ext = Base.get_extension(LinearSolve, :LinearSolveCUDAExt)
     if ext === nothing
+        # Check if we might have CUDA hardware but missing packages
+        try
+            # Try to detect NVIDIA GPUs via nvidia-smi or similar system indicators
+            if haskey(ENV, "CUDA_VISIBLE_DEVICES") ||
+               (Sys.islinux() && isfile("/proc/driver/nvidia/version")) ||
+               (Sys.iswindows() && success(`where nvidia-smi`))
+                @warn "CUDA hardware may be available but CUDA.jl extension is not loaded. Consider adding `using CUDA` to enable GPU algorithms."
+            end
+        catch
+            # Silently continue if detection fails
+        end
         return false
     end
 
@@ -25,16 +37,23 @@ end
     is_metal_available()
 
 Check if Metal (Apple Silicon) hardware and packages are available.
+Issues warnings if Metal hardware is detected but packages aren't loaded.
 """
 function is_metal_available()
     # Check if we're on macOS with Apple Silicon
     if !Sys.isapple()
         return false
     end
 
+    # Check if this is Apple Silicon
+    is_apple_silicon = Sys.ARCH == :aarch64
+
     # Check if Metal extension is loaded
     ext = Base.get_extension(LinearSolve, :LinearSolveMetalExt)
     if ext === nothing
+        if is_apple_silicon
+            @warn "Apple Silicon hardware detected but Metal.jl extension is not loaded. Consider adding `using Metal` to enable GPU algorithms."
+        end
         return false
     end