Linpack Benchmark Results

Compare Linpack floating-point performance across Single Board Computers

78 boards tested · scores from 2.7 to 156.4 (Performance (GFLOPS))

78 results

Full Results

Compare#
Board
Performance (GFLOPS)
GFLOPS/W
1
Radxa
Dragon Q8B
32GB LPDDR4
156.3803.097
2
Radxa
Orion O6N
32GB LPDDR5
137.1603.516
3
Orange Pi
6 Plus
32GB LPDDR5
136.7603.122
4
Orange Pi
6
16GB LPDDR5
135.2503.251
5
Radxa
Fogwise Airbox Q900
36GB LPDDR5
125.0705.670
6
LattePanda
Mu
8GB LPDDR5
50.1342.060
7
ArmSoM
W3
8GB LPDDR4X
48.9052.275
8
ArmSoM
AIM7
8GB LPDDR4X
48.6092.493
9
Orange Pi
5 Pro
16GB LPDDR4
48.2682.758
10
Radxa
ROCK 5T
12GB LPDDR5
48.2362.540
11
Orange Pi
5 Plus
8GB LPDDR4X
46.0682.934
12
Radxa
ROCK 5B+
16GB LPDDR5
44.1502.299
13
Radxa
ROCK 5 ITX
8GB LPDDR5
43.8542.171
14
Radxa
NX5
8GB LPDDR4X
43.4942.289
15
ADLINK
I-Pi SMARC 1200
4GB LPDDR4X
41.6722.526
16
Radxa
ROCK 5B
8GB LPDDR4X
41.5812.166
17
Radxa
ROCK 5A
4GB LPDDR4X
41.0262.613
18
ArmSoM
Sige7
8GB LPDDR4X
40.2151.991
19
Radxa
Dragon Q6A
6GB LPDDR5
38.6593.828
20
Orange Pi
5B
4GB
38.1992.497
21
Cool Pi
4B
4GB LPDDR4X
38.1252.230
22
Raspberry Pi
5
16GB LPDDR4X
37.9333.188
23
Raspberry Pi
5
8GB LPDDR4X
36.5993.155
24
Raspberry Pi
500
8GB LPDDR4X
36.4773.409
25
Raspberry Pi
500+
16GB LPDDR4X
35.1462.663
26
Radxa
ROCK 5C
4GB LPDDR4X
34.4232.100
27
Raspberry Pi
Compute Module 5
4GB LPDDR4X
34.4033.156
28
Raspberry Pi
5
4GB LPDDR4X
34.1742.712
29
Radxa
X4L
8GB LPDDR5
31.024N/A
30
Raspberry Pi
5
2GB LPDDR4X
28.1752.780
31
Radxa
X4
8GB LPDDR5
27.1941.230
32
Radxa
CM5 Lite
8GB LPDDR4X
25.3422.185
33
ArmSoM
CM5
8GB LPDDR5
23.198N/A
34
Radxa
ROCK 4D
8GB LPDDR5
22.8693.267
35
ArmSoM
Sige5
8GB LPDDR4X
22.6112.610
36
Libre Computer
Alta
4GB LPDDR4
21.6842.390
37
Orange Pi
4A
4GB LPDDR4X
21.6672.100
38
Dshan Pi
A1
4GB LPDDR4X
21.5262.790
39
Banana Pi
CM4
4GB LPDDR4
21.2412.150
40
Banana Pi
M2S (S922X)
4GB LPDDR4
20.9442.094
41
HardKernel
ODROID N2+
4GB DDR4
20.688N/A
42
HardKernel
ODROID N2+
2GB DDR4
20.295N/A
43
Radxa
Cubie A7A
6GB LPDDR5
18.7672.040
44
Banana Pi
R4
8GB LPDDR4
15.7102.014
45
Raspberry Pi
400
4GB LPDDR4
15.3881.998
46
Forlinx
FET3576-C
2GB LPDDR4X
15.2252.050
47
Radxa
ROCK 4B+
4GB LPDDR4
14.5470.915
48
Libre Computer
Renegade Elite
4GB LPDDR4
13.557N/A
49
Radxa
ROCK 4C+
4GB LPDDR4
12.9320.990
50
Raspberry Pi
4
2GB LPDDR4
12.5741.830
51
Raspberry Pi
Compute Module 4 (CM4)
2GB
12.4572.224
52
DEBIX
Model A
4GB LPDDR4
12.3332.250
53
Radxa
4 SE
4GB LPDDR4
11.6401.000
54
Orange Pi
Zero 3
4GB LPDDR4
10.8082.740
55
Orange Pi
Zero 2W
4GB LPDDR4
10.6932.673
56
Banana Pi
M4 Zero
4GB LPDDR4
10.6832.640
57
NVIDIA
Jetson Nano
2GB LPDDR4
10.2601.127
58
ArmSoM
Sige1
2GB LPDDR4X
9.8681.974
59
Radxa
ROCK 2A
4GB LPDDR4
9.8142.080
60
Libre Computer
Solitude
4GB LPDDR4
9.7631.992
61
Banana Pi
M5
4GB LPDDR4
9.2662.200
62
Radxa
ROCK 2F
2GB LPDDR4
9.1462.210
63
Orange Pi
3B
4GB LPDDR4
8.5211.272
64
Radxa
ROCK 3C
2GB LPDDR4
8.2031.370
65
Radxa
ROCK 3B
2GB LPDDR4
8.0741.009
66
Libre Computer
Renegade
2GB
7.7471.336
67
Radxa
CM3I
2GB LPDDR4
7.7410.832
68
Orange Pi
CM4
2GB LPDDR4
7.7031.410
69
Orange Pi
RV2
4GB LPDDR4X
7.2100.784
70
Radxa
Zero 3W
8GB LPDDR4
7.1571.746
71
DEBIX
Model C
2GB LPDDR4
6.6832.230
72
Libre Computer
Le Potato
2GB DDR3
6.0911.610
73
Sipeed
LicheePi 4A
8GB LPDDR4X
5.2780.580
74
Khadas
VIM1S
2GB LPDDR4
4.4201.890
75
Milk-V
Mars
4GB LPDDR4
3.1890.693
76
Orange Pi
RV
4GB LPDDR4
3.1870.490
77
StarFive
VisionFive 2
4GB LPDDR4
3.1720.470
78
StarFive
VisionFive 2 Lite
4GB LPDDR4
2.659N/A
Showing 78 results

Frequently Asked Questions

What is the fastest single board computer in Linpack?

The Radxa Dragon Q8B 32GB currently tops our Linpack ranking with 156.4 (Performance (GFLOPS)), measured on real hardware with our standardised test setup.

What is a good Linpack score for an SBC?

Across the 78 boards we have tested, Linpack results range from 2.7 to 156.4 (Performance (GFLOPS)), with a median of 21.2. Anything above the median puts a board in the faster half of the market.

About Linpack testing

Linpack measures raw floating point throughput in GFLOPS by solving dense systems of linear equations, the same family of workload used to rank the world's supercomputers. It leans hard on the CPU's vector units and memory subsystem, making it a strong indicator for scientific computing, signal processing, and other maths-heavy workloads. Where we have power measurements, we also report GFLOPS per watt, which turns Linpack into one of the best efficiency comparisons between boards.