Fix benchmark postprocess and plot script

viercc · Dec 14, 2024 · c8efcd5 · c8efcd5
1 parent 465ec34
commit c8efcd5
Show file tree

Hide file tree

Showing 10 changed files with 169 additions and 157 deletions.
diff --git a/bench/analyze.R b/bench/analyze.R
@@ -3,10 +3,20 @@ library(ggplot2)
 srcfile <- 'benchdata/bench_preprocessed.csv'
 
 data <- read.csv(srcfile)
+pico = 1E-12
+data$Mean <- data$Mean * pico
+data$StdDev <- data$SDx2 * 0.5 * pico 
+data$MeanLB <- pmax(pico, data$Mean - data$StdDev)
+data$MeanUB <- data$Mean + data$StdDev
 
 plotComparison <- function(data, filename) {
   plt <- ggplot(data) +
-    aes(x=Method, y=MeanTime, ymin=MeanLB, ymax=MeanUB, color=Name) +
+    aes(
+      x=Method,
+      y=Mean,
+      ymin=MeanLB,
+      ymax=MeanUB,
+      color=Name) +
     scale_y_log10(breaks=10^(0:-8)) +
     geom_errorbar(width=0.5) +
     geom_point()
@@ -17,11 +27,11 @@ plotRatio <- function(data, target_name, base_name, filename) {
   data_baseline <- subset(data, Name == base_name)
   data_target   <- subset(data, Name == target_name)
   ratio_data <- data.frame(data_target)
-  ratio_data$MeanTime <- ratio_data$MeanTime / data_baseline$MeanTime
+  ratio_data$Mean <- ratio_data$Mean / data_baseline$Mean
   ratio_data$MeanLB <- ratio_data$MeanLB / data_baseline$MeanUB
   ratio_data$MeanUB <- ratio_data$MeanUB / data_baseline$MeanLB
   plt <- ggplot(ratio_data) +
-    aes(x=Method, y=MeanTime, ymin=MeanLB, ymax=MeanUB) +
+    aes(x=Method, y=Mean, ymin=MeanLB, ymax=MeanUB) +
     scale_y_log10(limits=c(0.001,1000), breaks=10^(-2:2)) +
     geom_hline(linewidth=1, color=alpha('red', .6), yintercept=1) + 
     xlab("") +
@@ -32,16 +42,14 @@ plotRatio <- function(data, target_name, base_name, filename) {
   ggsave(file=filename, plot=plt, dpi=100, width=14, height=8)
 }
 
-data$Method <- with(data, reorder(Method, MeanTime))
-
-# isEnglish <- data$Dataset=='English'
-# isWiki    <- data$Dataset=='Wiki'
-# onSetOps <- data$Name=='Set' | data$Name=='TSet'
-# onMapOps <- data$Name=='Map' | data$Name=='TMap'
-# plotComparison(data[isEnglish & onSetOps,], "benchdata/set-english.png")
-# plotComparison(data[isEnglish & onMapOps,], "benchdata/map-english.png")
-# plotComparison(data[isWiki & onSetOps,], "benchdata/set-wiki.png")
-# plotComparison(data[isWiki & onMapOps,], "benchdata/map-wiki.png")
+isEnglish <- data$Dataset=='English'
+isWiki    <- data$Dataset=='Wiki'
+onSetOps <- data$Name=='Set' | data$Name=='TSet'
+onMapOps <- data$Name=='Map' | data$Name=='TMap'
+plotComparison(data[isEnglish & onSetOps,], "benchdata/set-english.png")
+plotComparison(data[isEnglish & onMapOps,], "benchdata/map-english.png")
+plotComparison(data[isWiki & onSetOps,], "benchdata/set-wiki.png")
+plotComparison(data[isWiki & onMapOps,], "benchdata/map-wiki.png")
 
 plotRatio(subset(data, Method != 'stringCount'), 'TSet', 'Set', "benchdata/ratio-set.png")
 plotRatio(subset(data, Method != 'stringCount'), 'TMap', 'Map', "benchdata/ratio-map.png")
diff --git a/doc/benchmark-generation-date b/doc/benchmark-generation-date
@@ -1 +1 @@
-2023-06-08T00:44:16+09:00
+2024-12-14T13:53:16+09:00
diff --git a/doc/benchmark.csv b/doc/benchmark.csv
diff --git a/doc/map-english.png b/doc/map-english.png
diff --git a/doc/map-wiki.png b/doc/map-wiki.png
diff --git a/doc/ratio-map.png b/doc/ratio-map.png
diff --git a/doc/ratio-set.png b/doc/ratio-set.png
diff --git a/doc/set-english.png b/doc/set-english.png
diff --git a/doc/set-wiki.png b/doc/set-wiki.png
diff --git a/run_bench.sh b/run_bench.sh
@@ -23,9 +23,13 @@ fi
 # Run benchmark
 datestr="$(date -Iseconds)"
 benchfile="benchdata/bench-${datestr}.csv"
-cabal run trie-benchmark -- -s "--csv=$benchfile"
+cabal run trie-benchmark -- "--csv=$benchfile"
 # Preprocess the output file
-sed -e 's/Name,Mean/Dataset\/Name\/Category\/Method,MeanTime/' "$benchfile" | tr '/' ',' > ./benchdata/bench_preprocessed.csv
+sed \
+  -e '1c Dataset,Name,Method,Mean,SDx2' \
+  -e 's/^All\.\([[:alnum:]]*\)\.\([[:alnum:]]*\)\.\([[:alnum:]]*\)\.\([[:alnum:]]*\)/\1,\2,"\3.\4"/g' \
+  "$benchfile" \
+  > ./benchdata/bench_preprocessed.csv
 # Draw charts using R
 R CMD BATCH bench/analyze.R
Original file line number	Diff line number	Diff line change
		@@ -1 +1 @@
		2023-06-08T00:44:16+09:00
		2024-12-14T13:53:16+09:00