Skip to content

Instantly share code, notes, and snippets.

@gengwg

gengwg/gist:29302ceff3aa2519d1619dd35b95ae7b

Created March 29, 2023 17:00
Show Gist options
  • Save gengwg/29302ceff3aa2519d1619dd35b95ae7b to your computer and use it in GitHub Desktop.
Save gengwg/29302ceff3aa2519d1619dd35b95ae7b to your computer and use it in GitHub Desktop.
Download ZIP
Raw
gistfile1.txt
# ./p2pBandwidthLatencyTest
[P2P (Peer-to-Peer) GPU Bandwidth Latency Test]
Device: 0, NVIDIA A100-SXM4-80GB, pciBusID: 7, pciDeviceID: 0, pciDomainID:0
Device: 1, NVIDIA A100-SXM4-80GB, pciBusID: f, pciDeviceID: 0, pciDomainID:0
Device: 2, NVIDIA A100-SXM4-80GB, pciBusID: 47, pciDeviceID: 0, pciDomainID:0
Device: 3, NVIDIA A100-SXM4-80GB, pciBusID: 4e, pciDeviceID: 0, pciDomainID:0
Device: 4, NVIDIA A100-SXM4-80GB, pciBusID: 87, pciDeviceID: 0, pciDomainID:0
Device: 5, NVIDIA A100-SXM4-80GB, pciBusID: 90, pciDeviceID: 0, pciDomainID:0
Device: 6, NVIDIA A100-SXM4-80GB, pciBusID: b7, pciDeviceID: 0, pciDomainID:0
Device: 7, NVIDIA A100-SXM4-80GB, pciBusID: bd, pciDeviceID: 0, pciDomainID:0
Device=0 CAN Access Peer Device=1
Device=0 CAN Access Peer Device=2
Device=0 CAN Access Peer Device=3
Device=0 CAN Access Peer Device=4
Device=0 CAN Access Peer Device=5
Device=0 CAN Access Peer Device=6
Device=0 CAN Access Peer Device=7
Device=1 CAN Access Peer Device=0
Device=1 CAN Access Peer Device=2
Device=1 CAN Access Peer Device=3
Device=1 CAN Access Peer Device=4
Device=1 CAN Access Peer Device=5
Device=1 CAN Access Peer Device=6
Device=1 CAN Access Peer Device=7
Device=2 CAN Access Peer Device=0
Device=2 CAN Access Peer Device=1
Device=2 CAN Access Peer Device=3
Device=2 CAN Access Peer Device=4
Device=2 CAN Access Peer Device=5
Device=2 CAN Access Peer Device=6
Device=2 CAN Access Peer Device=7
Device=3 CAN Access Peer Device=0
Device=3 CAN Access Peer Device=1
Device=3 CAN Access Peer Device=2
Device=3 CAN Access Peer Device=4
Device=3 CAN Access Peer Device=5
Device=3 CAN Access Peer Device=6
Device=3 CAN Access Peer Device=7
Device=4 CAN Access Peer Device=0
Device=4 CAN Access Peer Device=1
Device=4 CAN Access Peer Device=2
Device=4 CAN Access Peer Device=3
Device=4 CAN Access Peer Device=5
Device=4 CAN Access Peer Device=6
Device=4 CAN Access Peer Device=7
Device=5 CAN Access Peer Device=0
Device=5 CAN Access Peer Device=1
Device=5 CAN Access Peer Device=2
Device=5 CAN Access Peer Device=3
Device=5 CAN Access Peer Device=4
Device=5 CAN Access Peer Device=6
Device=5 CAN Access Peer Device=7
Device=6 CAN Access Peer Device=0
Device=6 CAN Access Peer Device=1
Device=6 CAN Access Peer Device=2
Device=6 CAN Access Peer Device=3
Device=6 CAN Access Peer Device=4
Device=6 CAN Access Peer Device=5
Device=6 CAN Access Peer Device=7
Device=7 CAN Access Peer Device=0
Device=7 CAN Access Peer Device=1
Device=7 CAN Access Peer Device=2
Device=7 CAN Access Peer Device=3
Device=7 CAN Access Peer Device=4
Device=7 CAN Access Peer Device=5
Device=7 CAN Access Peer Device=6
***NOTE: In case a device doesn't have P2P access to other one, it falls back to normal memcopy procedure.
So you can see lesser Bandwidth (GB/s) and unstable Latency (us) in those cases.
P2P Connectivity Matrix
D\D 0 1 2 3 4 5 6 7
0 1 1 1 1 1 1 1 1
1 1 1 1 1 1 1 1 1
2 1 1 1 1 1 1 1 1
3 1 1 1 1 1 1 1 1
4 1 1 1 1 1 1 1 1
5 1 1 1 1 1 1 1 1
6 1 1 1 1 1 1 1 1
7 1 1 1 1 1 1 1 1
Unidirectional P2P=Disabled Bandwidth Matrix (GB/s)
D\D 0 1 2 3 4 5 6 7
0 1539.41 16.37 16.83 17.50 16.74 17.18 17.00 17.27
1 15.31 1534.87 17.33 16.72 17.14 17.21 17.13 17.38
2 17.46 17.41 1537.89 16.18 16.92 17.18 17.14 17.36
3 17.48 17.43 16.00 1567.20 16.97 17.11 17.01 17.17
4 17.37 17.36 17.58 17.55 1539.41 15.23 16.99 17.55
5 17.17 17.46 17.41 17.65 15.38 1568.78 17.02 17.53
6 17.22 17.58 17.50 17.74 16.54 16.60 1571.93 15.81
7 17.00 17.21 16.92 17.64 16.40 16.71 15.08 1567.20
Unidirectional P2P=Enabled Bandwidth (P2P Writes) Matrix (GB/s)
D\D 0 1 2 3 4 5 6 7
0 1530.36 271.35 270.27 274.38 272.04 274.28 274.60 272.17
1 271.51 1547.03 270.11 274.20 273.55 272.41 272.74 273.07
2 270.63 271.35 1540.93 274.02 273.56 273.79 273.08 273.07
3 270.86 269.09 271.75 1553.18 271.21 273.92 273.60 272.46
4 271.78 272.93 273.06 272.18 1591.14 274.94 274.40 274.06
5 271.62 270.81 273.05 272.38 274.82 1586.29 275.02 273.32
6 271.16 271.83 274.01 272.13 274.21 274.59 1583.08 273.17
7 273.01 272.77 270.98 273.17 273.04 275.15 273.53 1584.69
Bidirectional P2P=Disabled Bandwidth Matrix (GB/s)
D\D 0 1 2 3 4 5 6 7
0 1546.26 17.82 24.32 24.60 25.01 25.37 25.34 19.21
1 17.46 1565.63 18.96 18.72 18.71 18.90 18.60 24.82
2 24.51 18.91 1583.88 17.83 18.55 18.88 18.74 25.13
3 24.64 18.99 17.90 1587.10 18.77 18.92 18.82 25.14
4 24.73 19.21 19.19 19.14 1586.29 16.31 18.79 24.89
5 24.87 19.30 19.31 19.31 16.40 1591.95 18.78 24.79
6 24.75 19.15 19.02 19.13 18.33 18.41 1590.33 17.38
7 18.59 24.75 24.47 24.84 24.82 24.03 17.74 1585.49
Bidirectional P2P=Enabled Bandwidth Matrix (GB/s)
D\D 0 1 2 3 4 5 6 7
0 1552.41 425.00 425.94 426.23 434.28 434.59 434.30 434.41
1 423.27 1546.26 423.02 423.71 425.33 424.75 423.94 424.61
2 424.50 422.72 1548.56 424.08 423.83 424.06 423.40 424.50
3 424.03 423.96 422.68 1545.50 424.98 424.18 423.83 424.03
4 435.60 427.22 427.24 426.93 1591.14 517.78 519.90 518.98
5 437.11 426.97 426.51 426.56 518.29 1585.49 517.61 518.87
6 435.82 426.81 427.05 425.90 517.78 519.15 1586.29 517.78
7 437.11 426.65 425.82 425.21 518.24 519.72 518.47 1582.28
P2P=Disabled Latency Matrix (us)
GPU 0 1 2 3 4 5 6 7
0 2.76 24.81 25.54 25.55 24.84 25.50 24.94 25.59
1 24.75 3.15 25.51 25.49 24.94 25.44 25.33 25.71
2 24.66 24.58 2.92 24.66 24.60 24.65 24.60 24.59
3 24.60 24.59 26.37 2.86 24.56 24.63 24.53 24.60
4 24.60 24.60 24.61 24.60 2.27 24.67 24.59 24.60
5 24.62 24.59 24.56 24.67 24.56 2.37 24.60 24.60
6 24.62 24.59 24.59 24.63 24.57 24.60 2.40 24.58
7 24.77 25.25 25.06 25.35 24.59 24.66 24.59 2.57
CPU 0 1 2 3 4 5 6 7
0 3.47 10.55 10.14 10.08 10.74 10.77 10.50 11.00
1 10.21 3.44 10.02 10.01 10.76 10.71 10.62 10.98
2 10.06 9.88 3.29 9.74 10.53 10.55 10.32 10.58
3 9.97 10.02 9.72 3.25 10.38 10.34 10.24 10.42
4 10.35 10.28 10.00 10.05 3.54 10.75 10.73 10.88
5 10.30 10.27 10.02 10.04 10.66 3.48 10.69 10.87
6 10.13 10.22 10.04 9.98 10.69 10.60 3.48 10.82
7 10.73 10.53 12.31 10.45 11.13 11.11 10.98 3.67
P2P=Enabled Latency (P2P Writes) Matrix (us)
GPU 0 1 2 3 4 5 6 7
0 2.76 3.44 3.39 3.48 3.47 3.39 3.40 3.41
1 3.50 2.86 3.52 3.51 3.45 3.47 3.51 3.46
2 3.48 3.46 2.96 3.46 3.52 3.54 3.46 3.45
3 3.52 3.45 3.52 2.83 3.45 3.44 3.45 3.52
4 2.95 2.93 2.94 2.93 2.25 2.93 2.94 2.97
5 3.07 3.07 3.10 3.09 3.07 2.29 3.04 3.02
6 2.93 3.01 2.92 2.94 3.00 2.97 2.38 3.00
7 3.03 3.04 3.03 3.11 3.03 3.02 3.07 2.45
CPU 0 1 2 3 4 5 6 7
0 3.50 2.79 2.71 2.71 2.71 2.71 2.74 2.71
1 2.85 3.52 2.82 2.79 2.78 2.84 2.76 2.85
2 2.76 2.76 3.42 2.72 2.72 2.77 2.69 2.74
3 2.90 2.74 2.79 3.41 2.74 2.79 2.75 2.83
4 3.07 2.96 2.98 3.00 3.62 3.01 2.89 2.97
5 3.07 3.01 2.98 2.95 2.99 3.73 2.96 2.98
6 3.03 2.99 3.05 2.98 3.04 3.05 3.57 2.92
7 3.20 3.12 3.16 3.07 3.17 3.14 3.15 3.77
NOTE: The CUDA Samples are not meant for performance measurements. Results may vary when GPU Boost is enabled.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment