{"id":212,"date":"2020-02-02T05:03:44","date_gmt":"2020-02-01T20:03:44","guid":{"rendered":"https:\/\/wp.study3.biz\/?p=212"},"modified":"2020-02-02T05:04:56","modified_gmt":"2020-02-01T20:04:56","slug":"ubuntu16-04-5-tesla-v100-sxm2-16gb-x4-cuda-9-0-samples-simplep2p-p2pbandwidthlatencytest-%e4%bb%96%e3%82%92%e5%8b%95%e4%bd%9c%e3%81%95%e3%81%9b%e3%81%a6%e3%81%bf%e3%81%9f","status":"publish","type":"post","link":"https:\/\/wp.study3.biz\/?p=212","title":{"rendered":"Ubuntu16.04.5 TESLA V100-SXM2 16GB x4 CUDA 9.0 Samples simpleP2P p2pBandwidthLatencyTest \u4ed6\u3092\u52d5\u4f5c\u3055\u305b\u3066\u307f\u305f"},"content":{"rendered":"

user0018@10-111-146-12:~\/cuda-samples-9.0\/samples\/0_Simple\/simpleP2P$ .\/simpleP2P
\n[.\/simpleP2P] – Starting…
\nChecking for multiple GPUs…
\nCUDA-capable device count: 4
\n> GPU0 = “Tesla V100-SXM2-16GB” IS capable of Peer-to-Peer (P2P)
\n> GPU1 = “Tesla V100-SXM2-16GB” IS capable of Peer-to-Peer (P2P)
\n> GPU2 = “Tesla V100-SXM2-16GB” IS capable of Peer-to-Peer (P2P)
\n> GPU3 = “Tesla V100-SXM2-16GB” IS capable of Peer-to-Peer (P2P)<\/p>\n

Checking GPU(s) for support of peer to peer memory access…
\n> Peer access from Tesla V100-SXM2-16GB (GPU0) -> Tesla V100-SXM2-16GB (GPU1) : Yes
\n> Peer access from Tesla V100-SXM2-16GB (GPU0) -> Tesla V100-SXM2-16GB (GPU2) : Yes
\n> Peer access from Tesla V100-SXM2-16GB (GPU0) -> Tesla V100-SXM2-16GB (GPU3) : Yes
\n> Peer access from Tesla V100-SXM2-16GB (GPU1) -> Tesla V100-SXM2-16GB (GPU0) : Yes
\n> Peer access from Tesla V100-SXM2-16GB (GPU1) -> Tesla V100-SXM2-16GB (GPU2) : Yes
\n> Peer access from Tesla V100-SXM2-16GB (GPU1) -> Tesla V100-SXM2-16GB (GPU3) : Yes
\n> Peer access from Tesla V100-SXM2-16GB (GPU2) -> Tesla V100-SXM2-16GB (GPU0) : Yes
\n> Peer access from Tesla V100-SXM2-16GB (GPU2) -> Tesla V100-SXM2-16GB (GPU1) : Yes
\n> Peer access from Tesla V100-SXM2-16GB (GPU2) -> Tesla V100-SXM2-16GB (GPU3) : Yes
\n> Peer access from Tesla V100-SXM2-16GB (GPU3) -> Tesla V100-SXM2-16GB (GPU0) : Yes
\n> Peer access from Tesla V100-SXM2-16GB (GPU3) -> Tesla V100-SXM2-16GB (GPU1) : Yes
\n> Peer access from Tesla V100-SXM2-16GB (GPU3) -> Tesla V100-SXM2-16GB (GPU2) : Yes
\nEnabling peer access between GPU0 and GPU1…
\nChecking GPU0 and GPU1 for UVA capabilities…
\n> Tesla V100-SXM2-16GB (GPU0) supports UVA: Yes
\n> Tesla V100-SXM2-16GB (GPU1) supports UVA: Yes
\nBoth GPUs can support UVA, enabling…
\nAllocating buffers (64MB on GPU0, GPU1 and CPU Host)…
\nCreating event handles…
\ncudaMemcpyPeer \/ cudaMemcpy between GPU0 and GPU1: 44.84GB\/s<\/strong>
\nPreparing host buffer and memcpy to GPU0…
\nRun kernel on GPU1, taking source data from GPU0 and writing to GPU1…
\nRun kernel on GPU0, taking source data from GPU1 and writing to GPU0…
\nCopy data back to host from GPU0 and verify results…
\nDisabling peer access…
\nShutting down…
\nTest passed
\nuser0018@10-111-146-12:~\/cuda-samples-9.0\/samples\/0_Simple\/simpleP2P$ cd ~\/cuda-samples-9.0\/samples\/1_Utilities
\nuser0018@10-111-146-12:~\/cuda-samples-9.0\/samples\/1_Utilities$ ls
\nbandwidthTest deviceQueryDrv topologyQuery
\ndeviceQuery p2pBandwidthLatencyTest
\nuser0018@10-111-146-12:~\/cuda-samples-9.0\/samples\/1_Utilities$ cd p2pBandwidthLatencyTest
\nuser0018@10-111-146-12:~\/cuda-samples-9.0\/samples\/1_Utilities\/p2pBandwidthLatencyTest$ .\/p2pBandwidthLatencyTest
\n[P2P (Peer-to-Peer) GPU Bandwidth Latency Test]
\nDevice: 0, Tesla V100-SXM2-16GB, pciBusID: 61, pciDeviceID: 0, pciDomainID:0
\nDevice: 1, Tesla V100-SXM2-16GB, pciBusID: 62, pciDeviceID: 0, pciDomainID:0
\nDevice: 2, Tesla V100-SXM2-16GB, pciBusID: 89, pciDeviceID: 0, pciDomainID:0
\nDevice: 3, Tesla V100-SXM2-16GB, pciBusID: 8a, pciDeviceID: 0, pciDomainID:0
\nDevice=0 CAN Access Peer Device=1
\nDevice=0 CAN Access Peer Device=2
\nDevice=0 CAN Access Peer Device=3
\nDevice=1 CAN Access Peer Device=0
\nDevice=1 CAN Access Peer Device=2
\nDevice=1 CAN Access Peer Device=3
\nDevice=2 CAN Access Peer Device=0
\nDevice=2 CAN Access Peer Device=1
\nDevice=2 CAN Access Peer Device=3
\nDevice=3 CAN Access Peer Device=0
\nDevice=3 CAN Access Peer Device=1
\nDevice=3 CAN Access Peer Device=2<\/p>\n

***NOTE: In case a device doesn’t have P2P access to other one, it falls back to normal memcopy procedure.
\nSo you can see lesser Bandwidth (GB\/s) in those cases.<\/p>\n

P2P Connectivity Matrix
\nD\\D 0 1 2 3
\n0 1 1 1 1
\n1 1 1 1 1
\n2 1 1 1 1
\n3 1 1 1 1
\nUnidirectional P2P=Disabled Bandwidth Matrix (GB\/s)
\nD\\D 0 1 2 3
\n0 748.32 9.92 11.02 11.03
\n1 9.96 742.63 11.02 11.07
\n2 11.05 11.03 744.05 9.94
\n3 11.04 11.01 9.92 745.47
\nUnidirectional P2P=Enabled Bandwidth Matrix (GB\/s)
\nD\\D 0 1 2 3
\n0 744.05 47.91 47.95 47.89
\n1 47.90 745.47 47.96 47.96
\n2 47.91 47.97 746.89 48.00
\n3 47.97 47.90 47.97 745.47
\nBidirectional P2P=Disabled Bandwidth Matrix (GB\/s)
\nD\\D 0 1 2 3
\n0 764.43 10.38 18.92 17.99
\n1 10.42 765.93 18.08 17.56
\n2 18.79 18.23 763.69 10.41
\n3 17.86 17.49 10.43 762.94
\nBidirectional P2P=Enabled Bandwidth Matrix (GB\/s)
\nD\\D 0 1 2 3
\n0 762.20 95.58 95.67 95.58
\n1 95.37 765.93 95.67 95.65
\n2 95.37 95.39 765.93 95.44
\n3 95.48 95.46 95.51 769.70
\nP2P=Disabled Latency Matrix (us)
\nD\\D 0 1 2 3
\n0 4.21 20.29 19.84 19.75
\n1 20.36 4.27 20.36 20.39
\n2 20.30 20.31 3.92 18.73
\n3 19.48 19.49 18.69 3.26
\nP2P=Enabled Latency Matrix (us)
\nD\\D 0 1 2 3
\n0 4.14 7.38 7.32 7.02
\n1 8.32 4.37 7.45 7.52
\n2 6.80 6.84 4.04 6.79
\n3 6.81 6.75 6.88 3.39<\/p>\n

NOTE: The CUDA Samples are not meant for performance measurements. Results may vary when GPU Boost is enabled.
\nuser0018@10-111-146-12:~\/cuda-samples-9.0\/samples\/1_Utilities\/p2pBandwidthLatencyTest$<\/p>\n

Ubuntu16.04.5 TESLA V100-SXM2 16GB x4 CUDA 9.0 Samples simpleP2P p2pBandwidthLatencyTest\u3000\u4ed6<\/a><\/p>\n","protected":false},"excerpt":{"rendered":"

user0018@10-111-146-12:~\/cuda-samples-9.0\/samples\/0_Simple\/simpleP2P$ .\/simpleP2P [.\/simpleP2P] – Starti … \u7d9a\u304d\u3092\u8aad\u3080 →<\/span><\/a><\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_jetpack_memberships_contains_paid_content":false,"footnotes":""},"categories":[18,17],"tags":[],"class_list":["post-212","post","type-post","status-publish","format-standard","hentry","category-nvidia","category-ubuntu"],"jetpack_featured_media_url":"","jetpack_sharing_enabled":true,"_links":{"self":[{"href":"https:\/\/wp.study3.biz\/index.php?rest_route=\/wp\/v2\/posts\/212","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/wp.study3.biz\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/wp.study3.biz\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/wp.study3.biz\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/wp.study3.biz\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=212"}],"version-history":[{"count":4,"href":"https:\/\/wp.study3.biz\/index.php?rest_route=\/wp\/v2\/posts\/212\/revisions"}],"predecessor-version":[{"id":219,"href":"https:\/\/wp.study3.biz\/index.php?rest_route=\/wp\/v2\/posts\/212\/revisions\/219"}],"wp:attachment":[{"href":"https:\/\/wp.study3.biz\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=212"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/wp.study3.biz\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=212"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/wp.study3.biz\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=212"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}