-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathbench.r
225 lines (175 loc) · 8.47 KB
/
bench.r
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
library(ggplot2)
library(tidyr)
library(dplyr)
library(readr)
setwd("~/Research/COMP-522/")
################## X86 - i9-12900K ##################
# Analysis of the benchmarking results
df_x86 <- read.table("bench_x86.tsv", header=TRUE, sep='\t') %>%
group_by(memory_model, n_threads) %>%
summarise(avg_dur_s = mean(dur_ms) / 1000)
# Plot the comparison
ggplot(df_x86, aes(x=n_threads, y=avg_dur_s, color=memory_model)) +
geom_line(linewidth=1.5) +
geom_point() +
theme(axis.title =element_text(size = rel(3.2)),
axis.text = element_text(colour = "black", size = rel(2.4)),
strip.text = element_text(size = rel(2.4))) +
xlab("Number of threads") + ylab("Duration (s)") + labs(group="Memory Model")
################## ARM - M1 Pro ##################
df_mac <- read.table("bench_mac.tsv", header=TRUE, sep='\t') %>%
group_by(memory_model, n_threads) %>%
summarise(avg_dur_s = mean(dur_ms) / 1000)
# Plot the comparison
ggplot(df_mac, aes(x=n_threads, y=avg_dur_s, color=memory_model)) +
geom_line(linewidth=1.5) +
geom_point() +
theme(axis.title =element_text(size = rel(3.2)),
axis.text = element_text(colour = "black", size = rel(2.4)),
strip.text = element_text(size = rel(2.4))) +
xlab("Number of threads") + ylab("Duration (s)") + labs(group="Memory Model")
################## IBM - POWER9 ##################
### Case with n_iter = 1000 and up to 20 threads
df_power9 <- read.table("bench_power9.tsv", header=TRUE, sep='\t') %>%
group_by(memory_model, n_threads) %>%
summarise(avg_dur_s = mean(dur_ms-tot_dur_work_ms) / 1000)
# Plot the comparison
ggplot(df_power9, aes(x=n_threads, y=avg_dur_s, color=memory_model)) +
geom_line(linewidth=1.5) +
geom_point() +
theme(axis.title =element_text(size = rel(3.2)),
axis.text = element_text(colour = "black", size = rel(2.4)),
strip.text = element_text(size = rel(2.4))) +
xlab("Number of threads") + ylab("Duration (s)") + labs(group="Memory Model")
### Case with n_iter = 1000 and up to 151 threads
df_power9_1000 <- read.table("bench_power9_1000.tsv", header=TRUE, sep='\t') %>%
group_by(memory_model, n_threads) %>%
summarise(avg_dur_s = mean(dur_ms-tot_dur_work_ms) / 1000)
# Plot the comparison
ggplot(df_power9_1000, aes(x=n_threads, y=avg_dur_s, color=memory_model)) +
geom_line(linewidth=1.5) +
geom_point() +
theme(axis.title =element_text(size = rel(3.2)),
axis.text = element_text(colour = "black", size = rel(2.4)),
strip.text = element_text(size = rel(2.4))) +
xlab("Number of threads") + ylab("Duration (s)") + labs(group="Memory Model")
### Case with n_iter = 20 and up to 160 threads
df_power9_20_50M <- read.table("bench_power9_20_50M.tsv", header=TRUE, sep='\t') %>%
group_by(memory_model, n_threads) %>%
summarise(avg_dur_s = mean(dur_ms-tot_dur_work_ms) / 1000)
# Plot the comparison
ggplot(df_power9_20_50M, aes(x=n_threads, y=avg_dur_s, color=memory_model)) +
geom_line(linewidth=1.5) +
geom_point() +
theme(axis.title =element_text(size = rel(3.2)),
axis.text = element_text(colour = "black", size = rel(2.4)),
strip.text = element_text(size = rel(2.4))) +
xlab("Number of threads") + ylab("Duration (s)") + labs(group="Memory Model")
### Case with n_iter = 100, 1M ops and up to 160 threads
df_power9_100_1M <- read.table("bench_power9_100_1M.tsv", header=TRUE, sep='\t') %>%
group_by(memory_model, n_threads) %>%
summarise(avg_dur_s = mean(dur_ms-tot_dur_work_ms) / 1000)
# Plot the comparison
ggplot(df_power9_100_1M, aes(x=n_threads, y=avg_dur_s, color=memory_model)) +
geom_line(linewidth=1.5) +
geom_point() +
theme(axis.title =element_text(size = rel(3.2)),
axis.text = element_text(colour = "black", size = rel(2.4)),
strip.text = element_text(size = rel(2.4))) +
xlab("Number of threads") + ylab("Duration (s)") + labs(group="Memory Model")
########### NEW RESULTS ###########
# x86
df_x86_1000_10M <- read.table("bench_x86_1000_10M.tsv", header=TRUE, sep='\t') %>%
group_by(memory_model, n_threads) %>%
summarise(avg_dur_s = mean(dur_ms-tot_dur_work_ms) / 1000)
# Plot the comparison
ggplot(df_x86_1000_10M, aes(x=n_threads, y=avg_dur_s, color=memory_model)) +
geom_line(linewidth=1.5) +
geom_point() +
theme(axis.title =element_text(size = rel(2)),
axis.text = element_text(colour = "black", size = rel(1)),
strip.text = element_text(size = rel(1)),
legend.text=element_text(size=rel(1)),
legend.title = element_text(size = rel(2))) +
xlab("Number of threads") + ylab("Duration (s)") + labs(color="Memory Model")
ggsave("res-x86.pdf", height = 6, width = 15)
df_x86_1000_10M_weak <- df_x86_1000_10M %>%
filter(memory_model == "weak") %>%
rename(dur_weak_ms = avg_dur_s)
df_x86_1000_10M_strong <- df_x86_1000_10M %>%
filter(memory_model == "strong") %>%
rename(dur_strong_ms = avg_dur_s)
df_x86_all <- merge(df_x86_1000_10M_weak, df_x86_1000_10M_strong, by="n_threads") %>%
select(-c(memory_model.x, memory_model.y)) %>%
mutate(percent_speedup = (1-dur_weak_ms/dur_strong_ms) * 100)
avg_speedup_x86 <- mean(df_x86_all$percent_speedup)
# POWER9
df_power9_1000_10M <- read.table("bench_power9_1000_10M.tsv", header=TRUE, sep='\t') %>%
group_by(memory_model, n_threads) %>%
summarise(avg_dur_s = mean(dur_ms-tot_dur_work_ms) / 1000)
# Plot the comparison
ggplot(df_power9_1000_10M, aes(x=n_threads, y=avg_dur_s, color=memory_model)) +
geom_line(linewidth=1.5) +
geom_point() +
theme(axis.title =element_text(size = rel(3.2)),
axis.text = element_text(colour = "black", size = rel(2.4)),
strip.text = element_text(size = rel(2.4))) +
xlab("Number of threads") + ylab("Duration (s)") + labs(color="Memory Model")
ggsave("res-power9.pdf", height = 6, width = 15)
df_power9_weak <- df_power9_1000_10M %>%
filter(memory_model == "weak") %>%
rename(dur_weak_ms = avg_dur_s)
df_power9_strong <- df_power9_1000_10M %>%
filter(memory_model == "strong") %>%
rename(dur_strong_ms = avg_dur_s)
df_power9_all <- merge(df_power9_weak, df_power9_strong, by="n_threads") %>%
select(-c(memory_model.x, memory_model.y)) %>%
mutate(percent_speedup = (1-dur_weak_ms/dur_strong_ms) * 100)
avg_speedup_power9 <- mean(df_power9_all$percent_speedup)
# Multithreading results
df_power9_1000_10M_320 <- read.table("bench_power9_1000_10M_320.tsv", header=TRUE, sep='\t') %>%
group_by(memory_model, n_threads) %>%
summarise(avg_dur_s = mean(dur_ms-tot_dur_work_ms) / 1000)
# Plot the comparison
ggplot(df_power9_1000_10M_320, aes(x=n_threads, y=avg_dur_s, color=memory_model)) +
geom_line(linewidth=1.5) +
geom_point() +
theme(axis.title =element_text(size = rel(3.2)),
axis.text = element_text(colour = "black", size = rel(2.4)),
strip.text = element_text(size = rel(2.4))) +
xlab("Number of threads") + ylab("Duration (s)") + labs(color="Memory Model")
ggsave("res-power9-multithreading.pdf", height = 6, width = 15)
df_power9_320_weak <- df_power9_1000_10M_320 %>%
filter(memory_model == "weak") %>%
rename(dur_weak_ms = avg_dur_s)
df_power9_320_strong <- df_power9_1000_10M_320 %>%
filter(memory_model == "strong") %>%
rename(dur_strong_ms = avg_dur_s)
df_power9_320_all <- merge(df_power9_320_weak, df_power9_320_strong, by="n_threads") %>%
select(-c(memory_model.x, memory_model.y)) %>%
mutate(percent_speedup = (1-dur_weak_ms/dur_strong_ms) * 100) %>%
filter(n_threads >= 160)
avg_speedup_power9_320 <- mean(df_power9_320_all$percent_speedup)
# Mac M1 Pro
df_mac <- read.table("bench_mac_1000_10M.tsv", header=TRUE, sep='\t') %>%
group_by(memory_model, n_threads) %>%
summarise(avg_dur_s = mean(dur_ms-tot_dur_work_ms) / 1000)
# Plot the comparison
ggplot(df_mac, aes(x=n_threads, y=avg_dur_s, color=memory_model)) +
geom_line(linewidth=1.5) +
geom_point() +
theme(axis.title =element_text(size = rel(3.2)),
axis.text = element_text(colour = "black", size = rel(2.4)),
strip.text = element_text(size = rel(2.4))) +
xlab("Number of threads") + ylab("Duration (s)") + labs(color="Memory Model")
ggsave("res-m1pro-multithreading.pdf", height = 6, width = 15)
df_mac_weak <- df_mac %>%
filter(memory_model == "weak") %>%
rename(dur_weak_ms = avg_dur_s)
df_mac_strong <- df_mac %>%
filter(memory_model == "strong") %>%
rename(dur_strong_ms = avg_dur_s)
df_mac_all <- merge(df_mac_weak, df_mac_strong, by="n_threads") %>%
select(-c(memory_model.x, memory_model.y)) %>%
mutate(percent_speedup = (1-dur_weak_ms/dur_strong_ms) * 100)
avg_speedup_mac<- mean(df_mac_all$percent_speedup)