You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
This repository was archived by the owner on May 17, 2025. It is now read-only.
Copy file name to clipboardExpand all lines: README
+7-13Lines changed: 7 additions & 13 deletions
Original file line number
Diff line number
Diff line change
@@ -27,7 +27,7 @@ hopscotch/
27
27
| |---4_cache/ Benchmark for evaluating caching.
28
28
| |---common/ Source code common to all benchmarks.
29
29
| |---include/ Common header files.
30
-
| |---kernels/ Common kernels. May get used by different benchmarks.
30
+
| |---kernels/ Common kernels. Used by different benchmarks.
31
31
|
32
32
|---gpu/
33
33
| |
@@ -59,12 +59,12 @@ IV. CPU Benchmarks
59
59
60
60
1_roofline
61
61
==========
62
-
Measures the maximum attainable performance with varying arithmetic intensity.
62
+
Measures the maximum attainable performance with varying arithmetic intensity and the machine balance.
63
63
64
-
To run: ./roofline_cpu.py
64
+
To run: ./roofline.py
65
65
66
66
The python script will generate a pdf for the roofline plot. Available options can be found
67
-
using ./roofline_cpu.py --help
67
+
using ./roofline.py --help
68
68
69
69
70
70
2_bandwidth
@@ -74,10 +74,9 @@ IV. CPU Benchmarks
74
74
To run: a) make
75
75
b) ./bandwidth
76
76
77
-
Working set size can be changed by defining HS_ARRAY_ELEM, which denotes number of elements in the working set array.
78
-
HS_ARRAY_ELEM can be defined directy if manually compiling, or can be passes with USER_DEF.
77
+
Working set size can be changed by defining WSS_EXP. Number of elements in the working set is (2 ^ WSS_EXP). WSS_EXP can be defined directy if manually compiling, or can be passes with USER_DEFS.
79
78
80
-
Example: a) make USER_DEF="-DHS_ARRAY_ELEM=1000000"
79
+
Example: a) make USER_DEFS="-DWSS_EXP=32"
81
80
b) ./bandwidth
82
81
83
82
@@ -98,18 +97,13 @@ IV. CPU Benchmarks
98
97
To run: a) make
99
98
b) ./cache
100
99
101
-
Working set size can be changed by defining HS_ARRAY_ELEM, which denotes number of elements in the working set array.
102
-
HS_ARRAY_ELEM can be defined directy if manually compiling, or can be passes with USER_DEF.
103
100
104
-
Example: a) make USER_DEF="-DHS_ARRAY_ELEM=1000000"
105
-
b) ./cache
106
-
107
101
108
102
IV. GPU Benchmarks
109
103
110
104
1_roofline
111
105
==========
112
-
Measures the maximum attainable performance with varying arithmetic intensity.
106
+
Measures the maximum attainable performance with varying arithmetic intensity and the machine balance.
113
107
Supports single and double precision floating point operations.
0 commit comments