Chart consistency + visual cleanup

2026-07-26 23:04:36 +02:00
parent df83623101
commit aad3b85d8f
10 changed files with 326 additions and 293 deletions
@@ -0,0 +1,120 @@
+library(ggplot2)
+library(dplyr)
+library(readr)
+library(viridisLite)
+
+# Usage: Rscript combined_instr_fault_correlation_heatmap.r exp_abspath1 ... [faults_file]
+
+args <- commandArgs(trailingOnly = TRUE)
+if (length(args) < 1) {
+  stop("Need at least 1 experiment")
+}
+
+csv_suffix <- if (grepl("\\.csv$", args[length(args)])) {
+  args[length(args)]
+} else {
+  "faults.csv"
+}
+exp_args <- if (grepl("\\.csv$", args[length(args)])) {
+  args[-length(args)]
+} else {
+  args
+}
+
+all_data <- data.frame()
+all_mnem <- data.frame()
+
+for (arg in exp_args) {
+  csv_file <- file.path(arg, csv_suffix)
+  if (!file.exists(csv_file)) {
+    warning(paste("Missing:", csv_file))
+    next
+  }
+
+  df <- read_csv(csv_file, col_types = cols())
+  df$experiment <- basename(arg)
+  all_data <- bind_rows(all_data, df)
+
+  # TODO: This is ignoring any filters currently
+  mnem_file <- file.path(arg, "mnemonics.csv")
+  if (!file.exists(mnem_file)) {
+    warning(paste("Missing:", mnem_file))
+    next
+  }
+  mdf <- read_csv(mnem_file, col_types = cols())
+  all_mnem <- bind_rows(all_mnem, mdf)
+}
+
+if (nrow(all_data) == 0) {
+  stop("No faults.csv data loaded")
+}
+
+if (nrow(all_mnem) == 0) {
+  stop("No mnemonics.csv data loaded")
+}
+
+# no OK_MARKER, sum GROUP1 + TRAP.
+all_data <- all_data |>
+  filter(!is.na(mnemonic), mnemonic != "NULL") |>
+  filter(resulttype != "OK_MARKER") |>
+  mutate(
+    resulttype = ifelse(resulttype == "GROUP1_MARKER", "TRAP", resulttype)
+  )
+
+if (nrow(all_data) == 0) {
+  stop("No failure-marker data to plot")
+}
+
+# Sum faults per (mnemonic, marker) pair for all experiments
+heat <- all_data |>
+  group_by(mnemonic, resulttype) |>
+  summarise(faults = sum(faults, na.rm = TRUE), .groups = "drop")
+
+# Sum mnemonic counts for all experiments
+mnem_counts <- all_mnem |>
+  filter(!is.na(mnemonic), mnemonic != "NULL") |>
+  group_by(mnemonic) |>
+  summarise(count = sum(count, na.rm = TRUE), .groups = "drop")
+
+# Normalize by mnemonic count
+heat <- heat |>
+  left_join(mnem_counts, by = "mnemonic") |>
+  filter(!is.na(count), count > 0) |>
+  mutate(fault_rate = faults / count)
+
+if (nrow(heat) == 0) {
+  stop("Heat join failed")
+}
+
+# Order by fault rate
+mnem_order <- heat |>
+  group_by(mnemonic) |>
+  summarise(total = sum(fault_rate), .groups = "drop") |>
+  arrange(desc(total)) |>
+  pull(mnemonic)
+
+heat <- heat |>
+  mutate(mnemonic = factor(mnemonic, levels = mnem_order))
+
+plot <- ggplot(
+  heat,
+  aes(x = mnemonic, y = resulttype, fill = fault_rate)
+) +
+  geom_tile(colour = "white") +
+  scale_fill_viridis_c(name = "Fault rate", trans = "log10") +
+  labs(
+    title = "Instruction / Fault Rate Heatmap (Normalized)",
+    x = "Instruction",
+    y = "Fault Type"
+  ) +
+  theme_minimal() +
+  theme(
+    axis.text.x = element_text(angle = 90, hjust = 1),
+    panel.grid = element_blank(),
+    plot.title = element_text(size = 14, face = "bold")
+  )
+
+suffix <- gsub("^faults|\\.csv$", "", csv_suffix)
+outfile <- paste0("injections/instr_fault_rate_heatmap", suffix, ".svg")
+ggsave(outfile, plot = plot, width = 12, height = 6)
+print(paste("Saved", outfile))