Fix LinearSolveAutotune printout to use correct benchmark size ranges (#691)

- Updated categorize_results() to use actual benchmark size categories (tiny, small, medium, large, big) instead of hardcoded ranges (0-128, 128-256, etc.)
- Added proper size range labels with cutoffs: tiny (5-20), small (20-100), medium (100-300), large (300-1000), big (10000+)
- Updated telemetry formatting to handle new range format and sort them in logical order
- Now the issue printout correctly reflects the same sizing used in the benchmark

Co-authored-by: ChrisRackauckas <[email protected]>

Author: ChrisRackauckas-Claude

lib/LinearSolveAutotune/src/benchmarking.jl

@@ -229,12 +229,14 @@ function categorize_results(df::DataFrame)
 
     categories = Dict{String, String}()
 
-    # Define size ranges
+    # Define size ranges based on actual benchmark categories
+    # These align with the sizes defined in get_benchmark_sizes()
     ranges = [
-        ("0-128", 1:128),
-        ("128-256", 129:256),
-        ("256-512", 257:512),
-        ("512+", 513:10000)
+        ("tiny (5-20)", 5:20),
+        ("small (20-100)", 21:100),
+        ("medium (100-300)", 101:300),
+        ("large (300-1000)", 301:1000),
+        ("big (10000+)", 10000:typemax(Int))
     ]
 
     # Get unique element types

lib/LinearSolveAutotune/src/telemetry.jl

@@ -192,7 +192,7 @@ function format_categories_markdown(categories::Dict{String, String})
     eltype_categories = Dict{String, Dict{String, String}}()
 
     for (key, algorithm) in categories
-        # Parse key like "Float64_0-128" -> eltype="Float64", range="0-128"
+        # Parse key like "Float64_tiny (5-20)" -> eltype="Float64", range="tiny (5-20)"
         if contains(key, "_")
             eltype, range = split(key, "_", limit=2)
             if !haskey(eltype_categories, eltype)
@@ -208,14 +208,34 @@ function format_categories_markdown(categories::Dict{String, String})
         end
     end
 
+    # Define the proper order for size ranges
+    size_order = ["tiny (5-20)", "small (20-100)", "medium (100-300)", "large (300-1000)", "big (10000+)"]
+    
+    # Custom sort function for ranges
+    function sort_ranges(ranges_dict)
+        sorted_pairs = []
+        for size in size_order
+            if haskey(ranges_dict, size)
+                push!(sorted_pairs, (size, ranges_dict[size]))
+            end
+        end
+        # Add any other ranges not in our predefined order (for backward compatibility)
+        for (range, algo) in ranges_dict
+            if !(range in size_order)
+                push!(sorted_pairs, (range, algo))
+            end
+        end
+        return sorted_pairs
+    end
+    
     # Format each element type
     for (eltype, ranges) in sort(eltype_categories)
         push!(lines, "#### Recommendations for $eltype")
         push!(lines, "")
         push!(lines, "| Size Range | Best Algorithm |")
         push!(lines, "|------------|----------------|")
 
-        for (range, algorithm) in sort(ranges)
+        for (range, algorithm) in sort_ranges(ranges)
             push!(lines, "| $range | $algorithm |")
         end
         push!(lines, "")