osu: adjust figures for publication

garlic/fig/osu/bw.R

@@ -2,6 +2,9 @@ library(ggplot2)
 library(dplyr, warn.conflicts = FALSE)
 library(scales)
 library(jsonlite)
+library(stringr)
+#library(extrafont)
+#library(Cairo)
 
 args=commandArgs(trailingOnly=TRUE)
 
@@ -15,7 +18,9 @@ dataset = jsonlite::stream_in(file(input_file), verbose=FALSE) %>%
 
 # We only need the nblocks and time
 df = select(dataset, config.unitName, config.nodes, config.ntasksPerNode, config.cpusPerTask, size, bw) %>%
-	rename(unitName=config.unitName)
+	rename(unitName=config.unitName) %>%
+  mutate(bw=bw / 1024.0) %>%
+  mutate(unitName=str_replace(unitName, "osu-bw-", ""))
 
 nodes = unique(df$config.nodes)
 tasksPerNode = unique(df$config.ntasksPerNode)
@@ -24,42 +29,69 @@ df$unitName = as.factor(df$unitName)
 df$sizeFactor = as.factor(df$size)
 
 df = group_by(df, unitName, sizeFactor) %>%
-  mutate(medianBw = median(bw)) %>%
+  mutate(median.bw = median(bw)) %>%
   ungroup()
 
-breaks = 10^(-10:10)
-minor_breaks <- rep(1:9, 21)*(10^rep(-10:10, each=9))
-
-p = ggplot(data=df, aes(x=size, y=bw)) +
-	labs(x="Size (bytes)", y="Bandwidth (MB/s)",
-              title=sprintf("OSU bandwidth benchmark: nodes=%d tasksPerNode=%d cpusPerTask=%d",
-			    nodes, tasksPerNode, cpusPerTask), 
-              subtitle=input_file) +
-	geom_boxplot(aes(color=unitName, group=interaction(unitName, sizeFactor))) +
-	scale_x_continuous(trans=log2_trans()) +
-	#scale_y_log10(breaks = breaks, minor_breaks = minor_breaks) +
-	theme_bw() +
-	theme(legend.position = c(0.8, 0.2))
-
 ppi=300
-h=4
+h=3
-w=8
+w=6
-ggsave("boxplot.pdf", plot=p, width=w, height=h, dpi=ppi)
-ggsave("boxplot.png", plot=p, width=w, height=h, dpi=ppi)
 
-p = ggplot(data=df, aes(x=size, y=medianBw)) +
+p = ggplot(data=df, aes(x=size, y=median.bw)) +
-	labs(x="Size (bytes)", y="Bandwidth (MB/s)",
+	labs(x="Message size", y="Bandwidth (GB/s)",
-              title=sprintf("OSU benchmark: osu_bw",
+    #title=sprintf("OSU benchmark: osu_bw", nodes, tasksPerNode, cpusPerTask), 
-			    nodes, tasksPerNode, cpusPerTask), 
+    subtitle=gsub("-", "\uad", input_file)) +
-              subtitle=input_file) +
+	geom_line(aes(linetype=unitName)) +
-	geom_line(aes(color=unitName, linetype=unitName)) +
+	geom_point(aes(shape=unitName), size=1.5) +
-	geom_point(aes(color=unitName, shape=unitName)) +
+  scale_shape_discrete(name = "MPI version") +
-  geom_hline(yintercept = 100e3 / 8, color="red") +
+  scale_linetype_discrete(name = "MPI version") +
-  annotate("text", x = 8, y = (100e3 / 8) * 0.95, label = "12.5GB/s (100Gb/s)") +
+  #scale_color_discrete(name = "MPI version") +
-	scale_x_continuous(trans=log2_trans()) +
+  geom_hline(yintercept=12.5, color="red") +
+  annotate("text", x=1, y=12.5 * .95,
+    label="Max: 12.5GB/s (100Gbps)",
+    hjust=0, vjust=1, size=3) +
+	#scale_x_continuous(trans=log2_trans()) +
+  scale_x_continuous(trans=log2_trans(),
+    labels=label_bytes("auto_binary"),
+    n.breaks = 12,
+    #breaks=unique(df$size),
+    #minor_breaks=NULL
+  ) +
 	#scale_y_log10(breaks = breaks, minor_breaks = minor_breaks) +
 	theme_bw() +
-	theme(legend.position = c(0.8, 0.2))
+  theme(plot.subtitle = element_text(size=8, family="mono")) +
+	theme(legend.justification = c(1,0), legend.position = c(0.99, 0.01)) +
+  theme(axis.text.x = element_text(angle=-45, hjust=0))
 
 ggsave("median-lines.png", plot=p, width=w, height=h, dpi=ppi)
 ggsave("median-lines.pdf", plot=p, width=w, height=h, dpi=ppi)
+#ggsave("median-lines-cairo.pdf", plot=p, width=w, height=h, dpi=ppi, device=cairo_pdf)
+#CairoPDF(file="median-lines-Cairo.pdf", width=w, height=h)
+#print(p)
+#dev.off()
+
+
+p = ggplot(data=df, aes(x=size, y=bw)) +
+	labs(x="Message size", y="Bandwidth (MB/s)",
+    #title=sprintf("OSU benchmark: osu_bw", nodes, tasksPerNode, cpusPerTask),
+    subtitle=input_file) +
+	geom_line(aes(y=median.bw, linetype=unitName, group=unitName)) +
+	geom_point(aes(shape=unitName), size=2) +
+  scale_shape(solid = FALSE) +
+  geom_hline(yintercept = 100e3 / 8, color="red") +
+  annotate("text", x = 8, y = (100e3 / 8) * 0.95,
+    label = "Max: 12.5GB/s (100Gbps)") +
+	#scale_x_continuous(trans=log2_trans()) +
+  scale_x_continuous(trans=log2_trans(),
+    labels=label_bytes("auto_binary"),
+    breaks=unique(df$size),
+    minor_breaks=NULL) +
+	#scale_y_log10(breaks = breaks, minor_breaks = minor_breaks) +
+	theme_bw() +
+  theme(plot.subtitle = element_text(size=4)) +
+	theme(legend.position = c(0.2, 0.6)) +
+  theme(axis.text.x = element_text(angle=-45, hjust=0))
+
+ggsave("bw.png", plot=p, width=w, height=h, dpi=ppi)
+ggsave("bw.pdf", plot=p, width=w, height=h, dpi=ppi)
+
+warnings()

garlic/fig/osu/latency.R

@@ -2,6 +2,7 @@ library(ggplot2)
 library(dplyr, warn.conflicts = FALSE)
 library(scales)
 library(jsonlite)
+library(stringr)
 
 args=commandArgs(trailingOnly=TRUE)
 
@@ -15,7 +16,8 @@ dataset = jsonlite::stream_in(file(input_file), verbose=FALSE) %>%
 
 # We only need the nblocks and time
 df = select(dataset, config.unitName, config.nodes, config.ntasksPerNode, config.cpusPerTask, size, latency) %>%
-	rename(unitName=config.unitName)
+	rename(unitName=config.unitName) %>%
+  mutate(unitName=str_replace(unitName, "osu-latency-", ""))
 
 nodes = unique(df$config.nodes)
 tasksPerNode = unique(df$config.ntasksPerNode)
@@ -30,34 +32,45 @@ df = group_by(df, unitName, sizeFactor) %>%
 breaks = 10^(-10:10)
 minor_breaks <- rep(1:9, 21)*(10^rep(-10:10, each=9))
 
-p = ggplot(data=df, aes(x=size, y=latency)) +
-	labs(x="Size (bytes)", y="Latency (us)",
-              title=sprintf("OSU latency benchmark nodes=%d tasksPerNode=%d cpusPerTask=%d",
-			    nodes, tasksPerNode, cpusPerTask), 
-              subtitle=input_file) +
-	geom_boxplot(aes(color=unitName, group=interaction(unitName, sizeFactor))) +
-	scale_x_continuous(trans=log2_trans()) +
-	scale_y_log10(breaks = breaks, minor_breaks = minor_breaks) +
-	theme_bw() +
-	theme(legend.position = c(0.8, 0.2))
-
 ppi=300
-h=4
+h=3
-w=8
+w=6
-ggsave("boxplot.png", plot=p, width=w, height=h, dpi=ppi)
-ggsave("boxplot.pdf", plot=p, width=w, height=h, dpi=ppi)
 
 p = ggplot(data=df, aes(x=size, y=medianLatency)) +
-	labs(x="Size (bytes)", y="Latency (us)",
+	labs(x="Message size", y="Median latency (µs)",
-              title=sprintf("OSU benchmark: osu_latency",
+    #title=sprintf("OSU benchmark: osu_latency", nodes, tasksPerNode, cpusPerTask), 
-			    nodes, tasksPerNode, cpusPerTask), 
+    subtitle=gsub("-", "\uad", input_file)) +
-              subtitle=input_file) +
+	geom_line(aes(linetype=unitName)) +
-	geom_line(aes(color=unitName, linetype=unitName)) +
+	geom_point(aes(shape=unitName), size=2) +
-	geom_point(aes(color=unitName, shape=unitName)) +
-	scale_x_continuous(trans=log2_trans()) +
 	scale_y_log10(breaks = breaks, minor_breaks = minor_breaks) +
+  scale_x_continuous(trans=log2_trans(),
+    labels=label_bytes("auto_binary"),
+    n.breaks = 12)+
+  scale_shape_discrete(name = "MPI version") +
+  scale_linetype_discrete(name = "MPI version") +
 	theme_bw() +
-	theme(legend.position = c(0.2, 0.8))
+  theme(plot.subtitle = element_text(size=8, family="mono")) +
+	theme(legend.justification = c(0,1), legend.position = c(0.01, 0.99)) +
+  theme(axis.text.x = element_text(angle=-45, hjust=0))
 
 ggsave("median-lines.png", plot=p, width=w, height=h, dpi=ppi)
 ggsave("median-lines.pdf", plot=p, width=w, height=h, dpi=ppi)
+
+p = ggplot(data=df, aes(x=size, y=latency)) +
+	labs(x="Size (bytes)", y="Latency (us)",
+    #title=sprintf("OSU benchmark: osu_latency", nodes, tasksPerNode, cpusPerTask), 
+    subtitle=input_file) +
+	geom_line(aes(y=medianLatency, linetype=unitName, group=unitName)) +
+	geom_point(aes(shape=unitName), size=2) +
+	scale_y_log10(breaks = breaks, minor_breaks = minor_breaks) +
+  scale_x_continuous(trans=log2_trans(),
+    labels=label_bytes("auto_binary"),
+    breaks=unique(df$size),
+    minor_breaks=NULL) +
+	theme_bw() +
+  theme(plot.subtitle = element_text(color="gray50")) +
+  theme(axis.text.x = element_text(angle=-45, hjust=0)) +
+	theme(legend.position = c(0.2, 0.8))
+
+ggsave("latency.png", plot=p, width=w, height=h, dpi=ppi)
+ggsave("latency.pdf", plot=p, width=w, height=h, dpi=ppi)

garlic/fig/osu/mtu.R

@@ -24,8 +24,9 @@ df = select(dataset,
 	    size, bw, config.iterations) %>%
 	rename(unitName=config.unitName,
 	       iterations=config.iterations,
-	       PSM2_MQ_EAGER_SDMA_SZ=config.PSM2_MQ_EAGER_SDMA_SZ,
+	       PSM2_MQ_EAGER_SDMA_SZ.val=config.PSM2_MQ_EAGER_SDMA_SZ,
-	       PSM2_MTU=config.PSM2_MTU)
+	       PSM2_MTU.val=config.PSM2_MTU) %>%
+  mutate(bw = bw / 1000.0)
 
 nodes = unique(df$config.nodes)
 tasksPerNode = unique(df$config.ntasksPerNode)
@@ -33,33 +34,35 @@ cpusPerTask = unique(df$config.cpusPerTask)
 df$unitName = as.factor(df$unitName)
 df$sizeFactor = as.factor(df$size)
 df$sizeKB = df$size / 1024
-df$PSM2_MQ_EAGER_SDMA_SZ.f = as.factor(df$PSM2_MQ_EAGER_SDMA_SZ)
+df$PSM2_MQ_EAGER_SDMA_SZ = as.factor(df$PSM2_MQ_EAGER_SDMA_SZ.val)
-df$PSM2_MTU.f = as.factor(df$PSM2_MTU)
+df$PSM2_MTU = as.factor(df$PSM2_MTU.val)
 
 iterations = unique(df$iterations)
 
 df = group_by(df, unitName, sizeFactor) %>%
-  mutate(medianBw = median(bw)) %>%
+  mutate(median.bw = median(bw)) %>%
   ungroup()
 
 breaks = 10^(-10:10)
 minor_breaks <- rep(1:9, 21)*(10^rep(-10:10, each=9))
 
-ppi=150
+ppi=300
-h=6
+h=3
-w=8
+w=6
 
 p = ggplot(data=df, aes(x=sizeKB, y=bw)) +
-  labs(x="Message size (KB)", y="Bandwidth (MB/s)",
+  geom_vline(aes(xintercept = PSM2_MQ_EAGER_SDMA_SZ.val/1024), color="blue") +
-    title=sprintf("OSU benchmark: osu_bw --iterations %d", iterations), 
+  geom_vline(aes(xintercept = PSM2_MTU.val/1024), color="red") +
-    subtitle=input_file) +
+  labs(x="Message size (KiB)", y="Bandwidth (GB/s)",
-  geom_point(shape=21, size=3) +
+    #title=sprintf("OSU benchmark: osu_bw --iterations %d", iterations), 
-  geom_vline(aes(xintercept = PSM2_MQ_EAGER_SDMA_SZ/1024), color="blue") +
+    subtitle=gsub("-", "\uad", input_file)) +
-  geom_vline(aes(xintercept = PSM2_MTU / 1024), color="red") +
+  geom_point(shape=21, size=2) +
   #annotate("text", x = 10.2, y = 8.5e3, label = "MTU = 10KB", color="red", hjust=0) +
-  facet_wrap(vars(PSM2_MTU.f), nrow=3, labeller = "label_both") +
+  facet_wrap(vars(PSM2_MTU), nrow=3, labeller = "label_both") +
-  scale_x_continuous(breaks = unique(df$sizeKB), minor_breaks=NULL) +
+  #scale_x_continuous(breaks = unique(df$sizeKB), minor_breaks=NULL) +
-  theme_bw()
+  scale_x_continuous(n.breaks = 12) +
+  theme_bw() +
+  theme(plot.subtitle = element_text(size=8, family="mono"))
 
 ggsave("bw.png", plot=p, width=w, height=h, dpi=ppi)
 ggsave("bw.pdf", plot=p, width=w, height=h, dpi=ppi)