Rodrigo Arias Mallo
92cd88e365
The input dataset is not enough to determine which script produced a given plot.
112 lines
2.6 KiB
R
112 lines
2.6 KiB
R
library(ggplot2)
|
|
library(dplyr)
|
|
library(scales)
|
|
library(jsonlite)
|
|
|
|
args=commandArgs(trailingOnly=TRUE)
|
|
|
|
# Read the timetable from args[1]
|
|
input_file = "input.json"
|
|
if (length(args)>0) { input_file = args[1] }
|
|
if (length(args)>1) { output = args[2] } else { output = "?" }
|
|
|
|
# Load the dataset in NDJSON format
|
|
dataset = jsonlite::stream_in(file(input_file)) %>%
|
|
jsonlite::flatten()
|
|
|
|
particles = unique(dataset$config.particles)
|
|
|
|
# We only need the nblocks and time
|
|
df = select(dataset, config.nblocks, config.hw.cpusPerSocket, time) %>%
|
|
rename(nblocks=config.nblocks,
|
|
cpusPerSocket=config.hw.cpusPerSocket)
|
|
|
|
df = df %>% mutate(blocksPerCpu = nblocks / cpusPerSocket)
|
|
df$nblocks = as.factor(df$nblocks)
|
|
df$blocksPerCpuFactor = as.factor(df$blocksPerCpu)
|
|
|
|
# Normalize the time by the median
|
|
D=group_by(df, nblocks) %>%
|
|
mutate(tnorm = time / median(time) - 1) %>%
|
|
mutate(bad = max(ifelse(abs(tnorm) >= 0.01, 1, 0)))
|
|
|
|
D$bad = as.factor(D$bad)
|
|
|
|
#D$bad = as.factor(ifelse(abs(D$tnorm) >= 0.01, 2,
|
|
# ifelse(abs(D$tnorm) >= 0.005, 1, 0)))
|
|
|
|
bs_unique = unique(df$nblocks)
|
|
nbs=length(bs_unique)
|
|
|
|
print(D)
|
|
|
|
ppi=300
|
|
h=5
|
|
w=5
|
|
|
|
png("box.png", width=w*ppi, height=h*ppi, res=ppi)
|
|
#
|
|
#
|
|
#
|
|
# Create the plot with the normalized time vs nblocks
|
|
p = ggplot(data=D, aes(x=blocksPerCpuFactor, y=tnorm, color=bad)) +
|
|
|
|
# Labels
|
|
labs(x="Blocks/CPU", y="Normalized time",
|
|
title=sprintf("Nbody normalized time. Particles=%d", particles),
|
|
subtitle=output) +
|
|
|
|
|
|
# Center the title
|
|
#theme(plot.title = element_text(hjust = 0.5)) +
|
|
|
|
# Black and white mode (useful for printing)
|
|
#theme_bw() +
|
|
|
|
# Add the maximum allowed error lines
|
|
geom_hline(yintercept=c(-0.01, 0.01),
|
|
linetype="dashed", color="gray") +
|
|
|
|
# Draw boxplots
|
|
geom_boxplot() +
|
|
scale_color_manual(values=c("black", "brown")) +
|
|
|
|
#scale_y_continuous(breaks = scales::pretty_breaks(n = 10)) +
|
|
|
|
theme_bw() +
|
|
|
|
theme(plot.subtitle=element_text(size=8)) +
|
|
theme(legend.position = "none")
|
|
#theme(legend.position = c(0.85, 0.85))
|
|
|
|
|
|
|
|
|
|
# Render the plot
|
|
print(p)
|
|
|
|
## Save the png image
|
|
dev.off()
|
|
#
|
|
png("scatter.png", width=w*ppi, height=h*ppi, res=ppi)
|
|
#
|
|
## Create the plot with the normalized time vs nblocks
|
|
p = ggplot(D, aes(x=blocksPerCpuFactor, y=time)) +
|
|
|
|
labs(x="Blocks/CPU", y="Time (s)",
|
|
title=sprintf("Nbody granularity. Particles=%d", particles),
|
|
subtitle=output) +
|
|
theme_bw() +
|
|
theme(plot.subtitle=element_text(size=8)) +
|
|
theme(legend.position = c(0.5, 0.88)) +
|
|
|
|
geom_point(shape=21, size=3) +
|
|
#scale_x_continuous(trans=log2_trans()) +
|
|
scale_y_continuous(trans=log2_trans())
|
|
|
|
# Render the plot
|
|
print(p)
|
|
|
|
# Save the png image
|
|
dev.off()
|