From 86e6037e599c2df2a08b3456d7fe3bca337ed5b9 Mon Sep 17 00:00:00 2001
From: Christoph Urlacher <christoph.urlacher@protonmail.com>
Date: Sun, 26 Jul 2026 23:35:04 +0200
Subject: [PATCH] Group fault rates / instruction plot per experiment

---
 injections/fault_rates_per_instruction.svg    |  4 +-
 .../combined_fault_rates_per_instruction.r    | 39 +++++++++++++++----
 2 files changed, 33 insertions(+), 10 deletions(-)

diff --git a/injections/fault_rates_per_instruction.svg b/injections/fault_rates_per_instruction.svg
index aa51617..842ca20 100644
--- a/injections/fault_rates_per_instruction.svg
+++ b/injections/fault_rates_per_instruction.svg
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:47819df1d31fba4225b71cd23f1e2f7f1cc781868f572f2e679c3900c4e0d0fb
-size 13619
+oid sha256:aa323120393b929cd1defc143d2e4f5b4fe0d2731377385c6747ab78ef428cc9
+size 9711
diff --git a/scripts/charts/combined_fault_rates_per_instruction.r b/scripts/charts/combined_fault_rates_per_instruction.r
index 0a57b7a..ad65438 100644
--- a/scripts/charts/combined_fault_rates_per_instruction.r
+++ b/scripts/charts/combined_fault_rates_per_instruction.r
@@ -1,6 +1,7 @@
 library(ggplot2)
 library(dplyr)
 library(readr)
+library(stringr)
 
 # Usage: Rscript combined_fault_rates.r exp_abspath1 exp_abspath2 ... [faults_file]
 
@@ -51,10 +52,23 @@ for (arg in exp_args) {
     next
   }
 
+  match <- str_match(
+    basename(arg),
+    "^\\d{2}-\\d{2}_\\d{2}-\\d{2}-\\d{2}_(.+?)_(c|aot|interp)_"
+  )
+  if (is.na(match[1, 1])) {
+    base_name <- basename(arg)
+    variant <- "unknown"
+  } else {
+    base_name <- match[1, 2]
+    variant <- match[1, 3]
+  }
+
   rates <- bind_rows(
     rates,
     data.frame(
-      experiment = basename(arg),
+      base_name = base_name,
+      variant = variant,
       fault_rate = total_faults / total_instrs
     )
   )
@@ -64,25 +78,34 @@ if (nrow(rates) == 0) {
   stop("No data loaded")
 }
 
-# Order by fault rate
+# Order base_names by their max fault rate
+base_order <- rates |>
+  group_by(base_name) |>
+  summarise(max_rate = max(fault_rate), .groups = "drop") |>
+  arrange(desc(max_rate)) |>
+  pull(base_name)
+
 rates <- rates |>
-  arrange(desc(fault_rate)) |>
-  mutate(experiment = factor(experiment, levels = experiment))
+  mutate(
+    base_name = factor(base_name, levels = base_order),
+    variant = factor(variant, levels = c("c", "aot", "interp", "unknown"))
+  )
 
 plot <- ggplot(
   rates,
-  aes(x = experiment, y = fault_rate, fill = experiment)
+  aes(x = base_name, y = fault_rate, fill = variant)
 ) +
-  geom_col() +
+  geom_col(position = position_dodge(preserve = "single")) +
+  scale_y_log10() +
   labs(
     title = "Fault Rate per Instruction",
     x = "Experiment",
-    y = "Faults / Instruction Count"
+    y = "Faults / Instruction Count",
+    fill = "Variant"
   ) +
   theme_minimal() +
   theme(
     axis.text.x = element_text(angle = 90, hjust = 1),
-    legend.position = "none",
     plot.title = element_text(size = 14, face = "bold")
   )