EVPath Performance Tests on the GTRI Parallel Software Testing and Evaluation Center (PASTEC) Cluster

Size: px
Start display at page:

Download "EVPath Performance Tests on the GTRI Parallel Software Testing and Evaluation Center (PASTEC) Cluster"

Transcription

1 EVPath Performance Tests on the GTRI Parallel Software Testing and Evaluation Center (PASTEC) Cluster Magdalena Slawinska, Greg Eisenhauer, Thomas M. Benson, Alan Nussbaum College of Computing, Georgia Institute of Technology, Atlanta, Georgia or Georgia Tech Research Institute, Atlanta, Georgia April 9, Abstract The document presents the performance results of running EVPath mtest/trans_test on the Georgia Tech Research Institute (GTRI) PASTEC (Parallel Software Testing and Evaluation Center) cluster [2] in a two-node setup over the EVPath sockets transport and in a single-node setup over the sockets and enet transport. In certain situations, the bandwidth obtained by EVPath was compared to the throughput achieved with the netperf benchmark [3]. The PASTEC cluster is owned by the School of Electrical and Computer Engineering at the Georgia Institute of Technology, and operated by the Sensors and Electromagnetics Laboratory of the GTRI. I. EXPERIMENT DESCRIPTION The experiments were run with EVPath trans_test and for certain experiments benchmarked with the independent tool netperf [3]. A. EVPath trans_test The trans_test is a part of the EVPath messaging constructor package and can be found at evpath-build/evpath/source/mtest/trans_test.c. The trans_test measures one-way delivered bandwidth by a particular transport. At the beginning the short start message is sent that initiates timing upon receive. Next, the actual messages are being sent specified by the -size parameter; the message is divided into vectors. After sending msg_count messages, the timing is terminated. The bandwidth is calculated based on the number of bytes transferred according to the formula: bandwidth [Mbps] = size [bytes] msg_count 8/time [sec]. (1) The experiments were conducted in the following setups: single-node over sockets transport (c4-xx, c3-xx) single-node over enet transport (c4-xx, c3-xx) two-nodes over sockets transport (c4-xx, c3-xx) The c4-xx nodes are equipped with 1 gigabit Ethernet NIC, and c3-xx nodes with specifications provided in Table I and the kernel parameter values for c3-xx nodes in Table II. The processors on c4-xx and c3-xx are different, Intel(R) Xeon(R) CPU X GHz and Intel(R) Core(TM) i7-3960x CPU 3.30GHz, respectively. More detail in this regard is provided in Table III. The EVPath enet transport is a reliable UDP transport based on an open source package ENet [1] that aims at providing a thin, yet robust network communication layer on top of UDP. The open source ENet package provides optional reliable, in-order delivery of packets. It does not support higher level networking features such as authentication, encryption, etc. All experiments with trans_test were conducted with the following command:

2 2 TABLE I: NIC specifications on c3-xx nodes. Port Bandwidth Vendor and Model Driver Ver Firmware Ver em1 1 GbE Broadcom adapter; model BCM sb p72p1 10 GbE Myricom adapter; model 10G-PCIE-8B p8p1 40 GbE Mellanox ConnectX-3 (part no. MCX354-FBCT) TABLE II: The kernel parameter values (via sysctl) on c3-xx nodes. Kernel Parameter Value net.core.optmem_max net.core.rmem_default net.core.rmem_max net.core.wmem_default net.core.wmem_max net.ipv4.tcp_mem net.ipv4.tcp_rmem net.ipv4.tcp_wmem /trans_test -transport $TRANSPORT -size $BYTES -vectors 1 -msg_count $MSG_COUNT \ -reuse_write_buffers 1 -take_receive_buffer 0 -timeout 60 The $TRANSPORT parameter was used to select the relevant transport, i.e., sockets or enet. The $BYTES parameter, which describes the size of the message in bytes, varied from 1024 to The $MSG_COUNT parameter was experimentally determined: on c4-xx nodes it was selected to ensure the duration of a single test to about 30 seconds, and for c3-xx it varied for different message sizes c.a. <= 3 seconds and c.a. 10 seconds. The two-node experiment was executed over ssh: 1) On c4-xx nodes an additional option -ssh $NODE was used to connect to the appropriate remote node; the CM_INTERFACE variable was not set; trans_test selected the default NIC. 2) On c3-xx nodes the -n option was used for trans_test because the -ssh option did not work due to the lack of reverse DNS for NICs on c3-xx nodes; the -n option outputs the explicit contact information that needs to be provided to trans_test on the remote node as an input parameter; TABLE III: The processor s specifications for c4-xx and c3-xx nodes as reported by /proc/cpuinfo. The nodes were running in powersave mode using the intel_pstate driver. In that cpuinfo snapshot, the CPU cores are downclocked, but they will have ramped up in clock frequency during the tests. The separate tests executed with iperf showed that switching the cores to performance mode did not make an appreciable difference; it previously helped to use performance mode with the older ACPI driver, but the intel_pstate driver seems to be more responsive in terms of dynamic clock adjustment. Characteristic c4-xx c3-xx vendor_id GenuineIntel GenuineIntel cpu family 6 6 model model name Intel(R) Xeon(R) CPU X GHz Intel(R) Core(TM) i7-3960x CPU 3.30GHz cpu MHz (see a comment in Table s Caption) cache size KB KB siblings cpu cores 6 6 cpuid level bogomips clflush size cache_alignment address sizes 40 bits physical, 48 bits virtual 46 bits physical, 48 bits virtual

3 3 #./trans_test client./trans_test -transport $TRANSPORT -size $MSG_SIZE -vectors 1 \ -msg_count $MSG_COUNT -reuse_write_buffers 1 -take_receive_buffer 0 \ -timeout 60 -n #./trans_test server on a remote node./trans_test -transport $TRANSPORT -size $MSG_SIZE -vectors 1 -msg_count $MSG_COUNT\ -reuse_write_buffers 1 -take_receive_buffer 0 -timeout 60 \ -n AAIAAJTJ8o29ZQAAATkCmAILqMA= 3) The selection of a NIC was performed via the environment variable CM_INTERFACE that was preceding the trans_test command. Specifically, it was set to p8p1, p72p1, em1, as e.g. indicated by ifconfig -a on c3-xx. For instance CM_INTERFACE=p8p1./trans_test -transport $TRANSPORT -size $MSG_SIZE -vectors 1 \ -msg_count $MSG_COUNT -reuse_write_buffers 1 -take_receive_buffer 0 \ -timeout 60 -n Apart from the single-node experiment with netperf on a c3-xx, each experiment was repeated five times, and the average was calculated and presented in figures and tables. B. Benchmark netperf For runs on c3-xx nodes the independent benchmark netperf [3], a benchmarking tool to measure the socket throughput, was executed. On a single node netperf was executed only once. Since the installation of netperf was accomplished in the user s space a client-server netperf version was used netserver was executing on the server node and netperf was running on the client node. The netserver was run as:./netserver The netperf was executed with the following command (the test_duration parameter indicates in seconds for the intended duration of the test):./netperf -H remote-hostname -l test_duration -- -m mesg_size II. RESULTS The results are presented in Fig. 1, and Fig. 2, for a single-node experiment, and for a two-node experiment, respectively. Table IV and Table V show the performance comparison between EVPath and netperf for c3-xx nodes, for the single-node experiment, and for the two-node experiment, respectively. It seems that in order to improve the performance results significantly the network driver parameters need to be modified, specifically an increase in the receive ring buffer size from the default 1024 to the maximum 8192 in the 40 GbE driver resulted in performance boost from ~10 Gbps to Gpbs for all test programs, i.e., netperf, iperf (the iperf results are not included in this study), and EVPath. A. Single-Node Experiment Results The single node experiment over the EVPath sockets transport for c4-xx shows that the bandwidth achieves the greatest value equal 45,692 Mbps for the 1 MiB message size, and then decreases to 26,000 Mbps for the 16 MiB message. It is a 43% decrease in comparison to the observed peak bandwidth. EVPath achieves the highest bandwidth of 65,000 Mbps for the 1 MiB message size in the single-node experiment for c3-xx over 1GbE NIC (em1), and it also deteriorates about 41% to 38,500 Mbps at the 16 MiB message size in comparison to the observed peak throughput.

4 4 Single node: sockets transport Bandwidth [Mbps] 60,000 40,000 20, KiB 4KiB 16KiB 64KiB 256KiB 1MiB 4MiB 16MiB Message size [KiB] or [MiB] c4-05 EVPath c3-00 em1 EVPath c3-00 p72p1 EVPath c3-00 p8p1 EVPath c3-00 em1 netperf c3-00 p72p1 netperf c3-00 p8p1 netperf (a) A single node over the sockets transport. For numerical values please refer to Table VI, Table VII, Table VIII, Table IX, and Table X. The netperf experiments were run only once per each NIC and lasted 30 seconds. 160 Single node: enet transport Bandwidth [Mbps] c4-05 EVPath c3-00 em1 EVPath c3-00 p72p1 EVPath c3-00 p8p1 EVPath 1KiB 4KiB 16KiB 64KiB 256KiB 1MiB 4MiB 16MiB Message size [KiB] or [MiB] (b) A single node over the enet transport. For numerical values please refer to Table XI, Table XII, Table XIII, and Table XIV. Fig. 1: Single node performance tests for Pastec over the sockets and enet; averaged over 5 experiments apart from netperf in Fig. 1a that was executed only once per each NIC and lasted 30 seconds; EVPath message count varies to allow for duration of the test for about 30 seconds. The c4-xx NIC is 1GbE, c3-xx em1 1GbE, p72p1 10GbE, and p8p1 40GbE. The observed EVPath trends are consistent with the throughput reports obtained from netperf, i.e., the greatest bandwidth of 54,289 Mbps over em1 on c3-xx was observed for the 1 MiB message size, and then 17% bandwidth deterioration for the 16 MiB message. The single node experiment over the enet transport demonstrates a trend similar to the single node over sockets experiment (to some extent). However, the highest bandwidth, 146 Mbps, is achieved at the 256 KiB message

5 5 TABLE IV: Average EVPath bandwidth versus netperf for a single c3-xx node over 1 GbE, 10 GbE, and 40 GbE. The percentage is calculated according to the formula: B EVPath B netperf 100. Bandwidth 40 GbE Bandwidth 10 GbE Bandwidth 1 GbE Message size EVPath [Mbps] netperf [Mbps] [%] EVPath [Mbps] netperf [Mbps] [%] EVPath [Mbps] netperf[mbps] [%] 1KiB 4,138 12, ,165 13, ,159 13, KiB 9,055 30, ,066 31, ,058 31, KiB 30,526 38, ,354 43, ,333 38, KiB 44,279 41, ,174 43, ,060 42, KiB 58,724 52, ,074 52, ,249 51, MiB 64,704 61, ,967 54, ,999 54, MiB 61,006 51, ,315 51, ,340 50, MiB 38,561 45, ,505 45, ,516 45, size, and then it starts decreasing to about 116 Mbps for 16 MiB on c4-xx. On the c3-xx node for all tested NICs, the bandwidth peaked at about 143 Mbps for the 1 MiB message (for 256 KiB it was very close and peaked at about 142 Mbps), and then deteriorated to about 136 [Mpbs] for 16 MiB. The bandwidth deterioration at 16 MiB message sizes on the single node over the enet transport is not so dramatic as in the case of the single-node sockets experiment, and it is about 20% in comparison to the highest bandwidth observed over 1GbE NIC on c4-xx, and 5% compared to the highest bandwidth achieved over 1GbE, 10GbE, and 40GbE on c3-xx. The bandwidth deterioration is surprising; the anticipated behavior is that the bandwidth should saturate at a certain message size and remain constant with the increase of the message size. The observed deterioration might result from the fact that the memory bandwidth on chips has been exceeded above a certain message size, which might indicate that some of the data stays within the cache hierarchy and thus some DRAM operations are avoided altogether. Once the data can no longer live within the L3 cache, performance becomes limited by the maximum bandwidth to memory. The single node over sockets transport allows to achieve c.a. 313x greater bandwidth than the single node over enet on c4-xx, i.e., 45,692 Mbps vs. 146 Mbps, respectively, and 456x on c3-xx, 65,000 Mbps vs. 143 Mbps. On a single c3-xx node the usage of CM_INTERFACE does not have any impact, as the results obtained for em1, p72p1, p8p1 are very similar. This is also confirmed by netperf the obtained throughput does not depend on the selected NIC. Although the throughput achieved over p8p1 for 1MiB by netperf is about 14% better than over p72p1 or em1; however, it has to be taken into account that there was only one sample of netperf run. On a single c3-xx node the EVPath performance is % of the netperf performance for 1 KiB 64 KiB message sizes, % of the netperf performance for 256 KiB 4 MiB, and 85% of the netperf performance for 16 MiB (Table IV). The duration of the tests has impact on the bandwidth achieved by netperf. In general, netperf reports better bandwidth for a default, 10 seconds test run, than for 30 seconds. The netperf tests lasted 30 seconds each, similarly to EVPath trans_test in this setup; although, it is not possible to configure EVPath to last for precisely 30 seconds, the execution time was chosen empirically by manipulating of the message count parameter for a particular message size to last for about 30 seconds. This might be one of the reasons why netperf reports higher bandwidths. The higher single-node performance for the c3-xx nodes may just be due to having newer CPUs (see Table III). The CPUs in c3-xx have significantly higher maximum memory bandwidth than those in c4-xx. The memory speed should not be a factor, as c4-xx and c3-xx are nominally the same speed (DDR3 1333MHz), although from different manufactures. In fact, we believe that the peak bandwidth results exceed the maximum memory bandwidth on those chips, which would seem to indicate that some of the data stays within the cache hierarchy and thus some DRAM operations are avoided altogether. As we stated earlier, this might also explain the degradation in results above a certain message size. Once the data can no longer live within the L3 cache, performance becomes limited by the maximum bandwidth to memory.

6 6 B. Two-Node Experiment Results Two nodes: sockets transport, 1GbE Bandwidth [Mbps] c4-05->c4-09 EVPath c3-01->c3-00 em1 EVPath c3-01->c3-00 em1 netperf 1KiB 4KiB 16KiB 64KiB 256KiB 1MiB 4MiB 16MiB Message size [KiB] or [MiB] (a) Two nodes over the sockets transport over 1GbE Ethernet interface. For numerical values please refer to Table XV, Table XVI, and Table XVII. Two nodes: sockets transport, p72p1 10GbE and p8p1 40GbE 20,000 Bandwidth [Mbps] 15,000 10,000 5,000 0 c3-00->c3-01 p8p1 EVPath c3-01->c3-00 p72p1 EVPath c3-00->c3-01 p8p1 netperf c3-01->c3-00 p72p1 netperf 1KiB 4KiB 16KiB 64KiB 256KiB 1MiB 4MiB 16MiB Message size [KiB] or [MiB] (b) Two nodes over the sockets transport over p8p1 and p72p1 interfaces. For numerical values please refer to Table XX, Table XVIII, Table XXI, and Table XIX. Fig. 2: Performance tests for two nodes c3-01->c3-00 or c3-00->c3-01 for Pastec over the sockets; average over 5 experiments; duration of the EVPath tests varies up to about 20 seconds for em1 and 15 seconds for p8p1, and p72p1; duration of the netperf is 3 seconds or 10 seconds. For 40 GbE the test direction had an impact that was demonstrated by both test programs. The achieved bandwidth c3-00->c3-01 was twice as high as for c3-01->c3-00. In this report only c3-00->c3-01 results over 40 GbE are included. The sockets transport for c4-xx in a two-node setup is stable for all message sizes and is about 941 Mbps. For

7 7 TABLE V: The average EVPath performance versus the average netperf performance on a two-node setup over 1 GbE, 10 GbE, and 40 GbE; c3-00->c3-01. The percentage is calculated according to the formula: B EVPath B netperf 100. Bandwidth 40 GbE Bandwidth 10 GbE Bandwidth 1 GbE Message size EVPath [Mbps] netperf [Mbps] [%] EVPath [Mbps] netperf [Mbps] [%] EVPath [Mbps] netperf[mbps] [%] 1KiB 2,250 8, ,355 9, KiB 4,738 12, ,409 9, KiB 10,923 21, ,871 9, KiB 15,445 21, ,902 9, KiB 20,698 21, ,868 9, MiB 19,047 19, ,714 9, MiB 18,758 19, ,647 9, MiB 17,891 19, ,556 9, c3-xx nodes, the observed peak bandwidth was 884 Mbps; 9,902 Mbps; 20,698 Mbps for 1 GbE (em1), 10 GbE (p72p1), and 40 GbE (p8p1), respectively. The results obtained for c3-xx were consistent with netperf reports 851 Mbps; 9,901 Mbps; 21,653 Mbps over em1, p72p1, and p8p1, respectively. There is about 100 Mbps difference in the achieved bandwidth between 1 GbE NICs on c4-xx and c3-xx. The lower performance over 1 GbE is likely due to a lower quality Broadcom-based NIC in the c3-xx nodes relative to the Intel chipset based NIC in c4-xx. The c3-xx nodes use a consumer ASRock motherboard whereas the c4-xx nodes use server-class Supermicro boards. For c3-xx nodes, the bandwidth deteriorates from 884 Mbps for the 1 KiB message size to 840 Mbps at the 16 KiB message size, and increases to 849 Mbps for 16 MiB message size over 1 GbE NIC. This behavior is similar to the one reported by netperf. However, we have observed the opposite behavior over 10 GbE and 40 GbE c3-xx NICs. The bandwidth reported by EVPath increases up to 9,800 Mbps for 10 GbE and stabilizes at the 16 KiB message size, although netperf performance remains above 9,000 Mbps through the entire message size range. However, for 40 GbE netperf demonstrates a similar behavior to EVPath, although it achieves its close-to-maximum bandwidth at the 16 KiB message size and remains in the range 19,000 22,000 Mbps, whereas EVPath bandwidth stabilizes at the 256 KiB message size and remains in the range from about 18,000 Mbps to less than 21,000 Mbps. Each of test programs achieves at maximum about 50% of available bandwidth over 40 GbE. The EVPath throughput is 94 98% of the netperf performance starting from the 256 KiB message size (Table V). For c3-xx, the bandwidth over 10 GbE NIC reported by EVPath is 24% and 55% of the netperf performance for message sizes 1 KiB and 4 KiB, respectively, and % for 16 KiB 16 MiB (Table V). For c3-xx, the throughputs over 1 GbE NIC reported by netperf and EVPath are almost identical, apart from message sizes 1 KiB and 4 KiB for which EVPath achieves 4+%, and 3% better bandwidth than netperf, respectively (Table V). III. ACKNOWLEDGMENTS This work was supported in part by the GTRI Independent Research and Development (IRAD) under the contract I , GTRI-SEAL-S2APO-DO-ATL ( A). REFERENCES [1] enet website. March [2] Parallel Software Testing and Evaluation Center at Georgia Tech Research Institute. March [3] Rick Jones. Netperf Homepage. March Release

8 8 Tables with numerical values. IV. APPENDIX TABLE VI: The EVPath experiment summary for a single node (c4-05); averaged over 5 samples; sockets transport. 1KiB , KiB , KiB , KiB ,530 5, KiB ,124 2, MiB ,692 2, MiB , MiB , TABLE VII: The EVPath experiment summary for a single node (c3-00); averaged over 5 samples; sockets transport; interface em1. 1KiB , KiB , KiB , KiB , KiB , MiB , MiB ,340 1, MiB ,

9 9 TABLE VIII: The EVPath experiment summary for a single node (c3-00); averaged over 5 samples; sockets transport; interface p72p1. 1KiB , KiB , KiB , KiB ,174 1, KiB , MiB , MiB ,315 1, MiB , TABLE IX: The EVPath experiment summary for a single node (c3-00); averaged over 5 samples; sockets transport; interface p8p1. 1KiB , KiB , KiB , KiB ,279 1, KiB , MiB , MiB ,006 1, MiB , TABLE X: The experiment summary for a single node (c3-00); only one sample for each NIC; sockets transport; netperf results; each test ran for 30 sec; received size socket was determined automatically by netperf and was bytes; send size socket was also determined automatically by netperf at bytes. Message size Bandwidth em1 [10 6 bits sec ] bits Bandwidth p72p1 [106 sec ] 1KiB 13,429 13,434 12,999 4KiB 31,302 31,549 30,609 16KiB 38,213 43,558 38,164 64KiB 42,095 43,489 41, KiB 51,793 52,255 52,338 1MiB 54,289 54,256 61,704 4MiB 50,704 51,000 51,591 16MiB 45,136 45,710 45,597 bits Bandwidth p8p1 [106 sec ] TABLE XI: The EVPath experiment summary for a single node (c4-05); averaged over 5 samples; enet transport. 1KiB KiB KiB KiB KiB MiB MiB MiB

10 10 TABLE XII: The EVPath experiment summary for a single node (c3-00); averaged over 5 samples; enet transport; interface em1. 1KiB KiB KiB KiB KiB MiB MiB MiB TABLE XIII: The EVPath experiment summary for a single node (c3-00); averaged over 5 samples; enet transport; interface p72p1. 1KiB KiB KiB KiB KiB MiB MiB MiB TABLE XIV: The EVPath experiment summary for a single node (c3-00); averaged over 5 samples; enet transport; interface p8p1. 1KiB KiB KiB KiB KiB MiB MiB MiB TABLE XV: The EVPath experiment summary for two nodes; the test started at c4-05 and ssh-ed to c4-09; averaged over 5 samples; sockets transport. 1KiB KiB KiB KiB KiB MiB MiB MiB

11 11 TABLE XVI: The EVPath experiment summary for two nodes (c3-01 -> c3-00); averaged over 5 samples; the socket transport; interface em1. 1KiB KiB KiB KiB KiB MiB MiB MiB TABLE XVII: The netperf experiment summary for two nodes (c3-01 -> c3-00); averaged over 5 samples; received size socket was determined automatically by netperf and was bytes; send size socket was also determined automatically by netperf at bytes; the socket transport; interface em1. Message size Time [secs] Bandwidth [Mbps] Std. dev. σ σ Bandwidth 1KiB KiB KiB KiB KiB MiB MiB MiB TABLE XVIII: The EVPath experiment summary for two nodes (c3-01 -> c3-00); averaged over 5 samples; the socket transport; interface p72p1. 1KiB , KiB , KiB , KiB , KiB , MiB , MiB , MiB , TABLE XIX: The netperf experiment summary for two nodes (c3-01 -> c3-00); averaged over 5 samples; received size socket was determined automatically by netperf and was bytes; send size socket was also determined automatically by netperf at bytes; the socket transport; interface p72p1. Message size Time [secs] Bandwidth [Mbps] Std. dev. σ σ Bandwidth 1KiB , KiB , KiB , KiB , KiB , MiB , MiB , MiB ,

12 12 TABLE XX: The EVPath experiment summary for two nodes (c3-00 -> c3-01); averaged over 5 samples; the socket transport; interface p8p1. 1KiB , KiB , KiB , KiB ,445 1, KiB , MiB , MiB , MiB , TABLE XXI: The netperf experiment summary for two nodes (c3-00 -> c3-01); averaged over 5 samples; received size socket was determined automatically by netperf and was bytes; send size socket was also determined automatically by netperf at bytes; the socket transport; interface p8p1. Message size Time [secs] Bandwidth [Mbps] Std. dev. σ σ Bandwidth 1KiB , KiB , KiB , KiB , KiB , MiB , MiB , MiB ,

46PaQ. Dimitris Miras, Saleem Bhatti, Peter Kirstein Networks Research Group Computer Science UCL. 46PaQ AHM 2005 UKLIGHT Workshop, 19 Sep

46PaQ. Dimitris Miras, Saleem Bhatti, Peter Kirstein Networks Research Group Computer Science UCL. 46PaQ AHM 2005 UKLIGHT Workshop, 19 Sep 46PaQ Dimitris Miras, Saleem Bhatti, Peter Kirstein Networks Research Group Computer Science UCL 46PaQ AHM 2005 UKLIGHT Workshop, 19 Sep 2005 1 Today s talk Overview Current Status and Results Future Work

More information

10GE network tests with UDP. Janusz Szuba European XFEL

10GE network tests with UDP. Janusz Szuba European XFEL 10GE network tests with UDP Janusz Szuba European XFEL Outline 2 Overview of initial DAQ architecture Slice test hardware specification Initial networking test results DAQ software UDP tests Summary 10GE

More information

Performance Characterization of the Dell Flexible Computing On-Demand Desktop Streaming Solution

Performance Characterization of the Dell Flexible Computing On-Demand Desktop Streaming Solution Performance Characterization of the Dell Flexible Computing On-Demand Desktop Streaming Solution Product Group Dell White Paper February 28 Contents Contents Introduction... 3 Solution Components... 4

More information

A System-Level Optimization Framework For High-Performance Networking. Thomas M. Benson Georgia Tech Research Institute

A System-Level Optimization Framework For High-Performance Networking. Thomas M. Benson Georgia Tech Research Institute A System-Level Optimization Framework For High-Performance Networking Thomas M. Benson Georgia Tech Research Institute thomas.benson@gtri.gatech.edu 1 Why do we need high-performance networking? Data flow

More information

ARISTA: Improving Application Performance While Reducing Complexity

ARISTA: Improving Application Performance While Reducing Complexity ARISTA: Improving Application Performance While Reducing Complexity October 2008 1.0 Problem Statement #1... 1 1.1 Problem Statement #2... 1 1.2 Previous Options: More Servers and I/O Adapters... 1 1.3

More information

Optimizing the GigE transfer What follows comes from company Pleora.

Optimizing the GigE transfer What follows comes from company Pleora. Optimizing the GigE transfer What follows comes from company Pleora. Selecting a NIC and Laptop Based on our testing, we recommend Intel NICs. In particular, we recommend the PRO 1000 line of Intel PCI

More information

CITY UNIVERSITY OF NEW YORK. Creating a New Project in IRBNet. i. After logging in, click Create New Project on left side of the page.

CITY UNIVERSITY OF NEW YORK. Creating a New Project in IRBNet. i. After logging in, click Create New Project on left side of the page. CITY UNIVERSITY OF NEW YORK Creating a New Project in IRBNet i. After logging in, click Create New Project on left side of the page. ii. Enter the title of the project, the principle investigator s (PI)

More information

LANCOM Techpaper IEEE n Indoor Performance

LANCOM Techpaper IEEE n Indoor Performance Introduction The standard IEEE 802.11n features a number of new mechanisms which significantly increase available bandwidths. The former wireless LAN standards based on 802.11a/g enable physical gross

More information

Mathematics Shape and Space: Polygon Angles

Mathematics Shape and Space: Polygon Angles a place of mind F A C U L T Y O F E D U C A T I O N Department of Curriculum and Pedagogy Mathematics Shape and Space: Polygon Angles Science and Mathematics Education Research Group Supported by UBC Teaching

More information

NFS/RDMA over 40Gbps iwarp Wael Noureddine Chelsio Communications

NFS/RDMA over 40Gbps iwarp Wael Noureddine Chelsio Communications NFS/RDMA over 40Gbps iwarp Wael Noureddine Chelsio Communications Outline RDMA Motivating trends iwarp NFS over RDMA Overview Chelsio T5 support Performance results 2 Adoption Rate of 40GbE Source: Crehan

More information

Mechanism Design using Creo Parametric 3.0

Mechanism Design using Creo Parametric 3.0 Mechanism Design using Creo Parametric 3.0 Overview Course Code Course Length TRN-4521-T 1 Day In this course, you will learn about creating mechanism connections, configuring the mechanism model, creating

More information

Work Project Report: Benchmark for 100 Gbps Ethernet network analysis

Work Project Report: Benchmark for 100 Gbps Ethernet network analysis Work Project Report: Benchmark for 100 Gbps Ethernet network analysis CERN Summer Student Programme 2016 Student: Iraklis Moutidis imoutidi@cern.ch Main supervisor: Balazs Voneki balazs.voneki@cern.ch

More information

EC-Bench: Benchmarking Onload and Offload Erasure Coders on Modern Hardware Architectures

EC-Bench: Benchmarking Onload and Offload Erasure Coders on Modern Hardware Architectures EC-Bench: Benchmarking Onload and Offload Erasure Coders on Modern Hardware Architectures Haiyang Shi, Xiaoyi Lu, and Dhabaleswar K. (DK) Panda {shi.876, lu.932, panda.2}@osu.edu The Ohio State University

More information

Lighting the Blue Touchpaper for UK e-science - Closing Conference of ESLEA Project The George Hotel, Edinburgh, UK March, 2007

Lighting the Blue Touchpaper for UK e-science - Closing Conference of ESLEA Project The George Hotel, Edinburgh, UK March, 2007 Working with 1 Gigabit Ethernet 1, The School of Physics and Astronomy, The University of Manchester, Manchester, M13 9PL UK E-mail: R.Hughes-Jones@manchester.ac.uk Stephen Kershaw The School of Physics

More information

SNAP Performance Benchmark and Profiling. April 2014

SNAP Performance Benchmark and Profiling. April 2014 SNAP Performance Benchmark and Profiling April 2014 Note The following research was performed under the HPC Advisory Council activities Participating vendors: HP, Mellanox For more information on the supporting

More information

Performance Analysis and Evaluation of Mellanox ConnectX InfiniBand Architecture with Multi-Core Platforms

Performance Analysis and Evaluation of Mellanox ConnectX InfiniBand Architecture with Multi-Core Platforms Performance Analysis and Evaluation of Mellanox ConnectX InfiniBand Architecture with Multi-Core Platforms Sayantan Sur, Matt Koop, Lei Chai Dhabaleswar K. Panda Network Based Computing Lab, The Ohio State

More information

Report: A Comparison of Synchrophasor Protocols

Report: A Comparison of Synchrophasor Protocols J. Ritchie Carroll Grid Protection Alliance Report: A Comparison of Synchrophasor Protocols NASPI Fall Meeting Philadelphia, PA October 24, 2018 Abstract Paper compares three protocols used for continuous

More information

Security Standards for Information Systems

Security Standards for Information Systems Security Standards for Information Systems Area: Information Technology Services Number: IT-3610-00 Subject: Information Systems Management Issued: 8/1/2012 Applies To: University Revised: 4/1/2015 Sources:

More information

iwarp Learnings and Best Practices

iwarp Learnings and Best Practices iwarp Learnings and Best Practices Author: Michael Fenn, Penn State Date: March 28, 2012 www.openfabrics.org 1 Introduction Last year, the Research Computing and Cyberinfrastructure group at Penn State

More information

Disclaimer This presentation may contain product features that are currently under development. This overview of new technology represents no commitme

Disclaimer This presentation may contain product features that are currently under development. This overview of new technology represents no commitme NET1343BU NSX Performance Samuel Kommu #VMworld #NET1343BU Disclaimer This presentation may contain product features that are currently under development. This overview of new technology represents no

More information

RoCE vs. iwarp Competitive Analysis

RoCE vs. iwarp Competitive Analysis WHITE PAPER February 217 RoCE vs. iwarp Competitive Analysis Executive Summary...1 RoCE s Advantages over iwarp...1 Performance and Benchmark Examples...3 Best Performance for Virtualization...5 Summary...6

More information

Performance Characteristics on Gigabit networks

Performance Characteristics on Gigabit networks Version 4.6 Impairment Emulator Software for IP Networks (IPv4 & IPv6) Performance Characteristics on Gigabit networks ZTI / 1 boulevard d'armor / BP 20254 / 22302 Lannion Cedex / France Phone: +33 2 9648

More information

Linux Network Tuning Guide for AMD EPYC Processor Based Servers

Linux Network Tuning Guide for AMD EPYC Processor Based Servers Linux Network Tuning Guide for AMD EPYC Processor Application Note Publication # 56224 Revision: 1.00 Issue Date: November 2017 Advanced Micro Devices 2017 Advanced Micro Devices, Inc. All rights reserved.

More information

CITY UNIVERSITY OF NEW YORK. i. Visit:

CITY UNIVERSITY OF NEW YORK. i. Visit: CITY UNIVERSITY OF NEW YORK I. ACCESSING IRB NET (New Registration) i. Visit: https://www.irbnet.org/release/index.html ii. New users: Click on New Registration in the top right corner iii. Fill-out the

More information

PC BASED REAL TIME DATA EXCHANGE ON 10GbE OPTICAL NETWORK USING RTOS*

PC BASED REAL TIME DATA EXCHANGE ON 10GbE OPTICAL NETWORK USING RTOS* THIC Software and Hardware Technology THCC03 PC BASED REAL TIME DATA EXCHANGE ON 10GbE OPTICAL NETWORK USING RTOS* Ninth International Workshop on Personal Computers and Particle Accelerator Controls (PCaPAC

More information

Performance Characteristics on Gigabit networks

Performance Characteristics on Gigabit networks Version 4.7 Impairment Emulator Software for IP Networks (IPv4 & IPv6) Performance Characteristics on Gigabit networks ZTI Communications / 1 rue Ampère / 22300 LANNION / France Phone: +33 2 9613 4003

More information

TABLE OF CONTENTS CHAPTER NO. TITLE PAGE NO. ABSTRACT 5 LIST OF TABLES LIST OF FIGURES LIST OF SYMBOLS AND ABBREVIATIONS xxi

TABLE OF CONTENTS CHAPTER NO. TITLE PAGE NO. ABSTRACT 5 LIST OF TABLES LIST OF FIGURES LIST OF SYMBOLS AND ABBREVIATIONS xxi ix TABLE OF CONTENTS CHAPTER NO. TITLE PAGE NO. ABSTRACT 5 LIST OF TABLES xv LIST OF FIGURES xviii LIST OF SYMBOLS AND ABBREVIATIONS xxi 1 INTRODUCTION 1 1.1 INTRODUCTION 1 1.2 WEB CACHING 2 1.2.1 Classification

More information

LANCOM Techpaper Routing Performance

LANCOM Techpaper Routing Performance LANCOM Techpaper Routing Performance Applications for communications and entertainment are increasingly based on IP networks. In order to ensure that the necessary bandwidth performance can be provided

More information

Title: Collaborative research: End-to-End Provisioned Optical Network Testbed for Large-Scale escience Applications

Title: Collaborative research: End-to-End Provisioned Optical Network Testbed for Large-Scale escience Applications Year 1 Activities report for the NSF project EIN-0335190 Title: Collaborative research: End-to-End Provisioned Optical Network Testbed for Large-Scale escience Applications Date: July 29, 2004 (this is

More information

A Simulation: Improving Throughput and Reducing PCI Bus Traffic by. Caching Server Requests using a Network Processor with Memory

A Simulation: Improving Throughput and Reducing PCI Bus Traffic by. Caching Server Requests using a Network Processor with Memory Shawn Koch Mark Doughty ELEC 525 4/23/02 A Simulation: Improving Throughput and Reducing PCI Bus Traffic by Caching Server Requests using a Network Processor with Memory 1 Motivation and Concept The goal

More information

DPDK Performance Report Release Test Date: Nov 16 th 2016

DPDK Performance Report Release Test Date: Nov 16 th 2016 Test Date: Nov 16 th 2016 Revision History Date Revision Comment Nov 16 th, 2016 1.0 Initial document for release 2 Contents Audience and Purpose... 4 Test setup:... 4 Intel Xeon Processor E5-2699 v4 (55M

More information

D1.1 Server Scalibility

D1.1 Server Scalibility D1.1 Server Scalibility Ronald van der Pol and Freek Dijkstra SARA Computing & Networking Services, Science Park 121, 1098 XG Amsterdam, The Netherlands March 2010 ronald.vanderpol@sara.nl,freek.dijkstra@sara.nl

More information

Linux Network Tuning Guide for AMD EPYC Processor Based Servers

Linux Network Tuning Guide for AMD EPYC Processor Based Servers Linux Network Tuning Guide for AMD EPYC Processor Application Note Publication # 56224 Revision: 1.10 Issue Date: May 2018 Advanced Micro Devices 2018 Advanced Micro Devices, Inc. All rights reserved.

More information

High bandwidth, Long distance. Where is my throughput? Robin Tasker CCLRC, Daresbury Laboratory, UK

High bandwidth, Long distance. Where is my throughput? Robin Tasker CCLRC, Daresbury Laboratory, UK High bandwidth, Long distance. Where is my throughput? Robin Tasker CCLRC, Daresbury Laboratory, UK [r.tasker@dl.ac.uk] DataTAG is a project sponsored by the European Commission - EU Grant IST-2001-32459

More information

Mainstream Computer System Components CPU Core 2 GHz GHz 4-way Superscaler (RISC or RISC-core (x86): Dynamic scheduling, Hardware speculation

Mainstream Computer System Components CPU Core 2 GHz GHz 4-way Superscaler (RISC or RISC-core (x86): Dynamic scheduling, Hardware speculation Mainstream Computer System Components CPU Core 2 GHz - 3.0 GHz 4-way Superscaler (RISC or RISC-core (x86): Dynamic scheduling, Hardware speculation One core or multi-core (2-4) per chip Multiple FP, integer

More information

FPGAs and Networking

FPGAs and Networking FPGAs and Networking Marc Kelly & Richard Hughes-Jones University of Manchester 12th July 27 1 Overview of Work Looking into the usage of FPGA's to directly connect to Ethernet for DAQ readout purposes.

More information

HP ProLiant BladeSystem Gen9 vs Gen8 and G7 Server Blades on Data Warehouse Workloads

HP ProLiant BladeSystem Gen9 vs Gen8 and G7 Server Blades on Data Warehouse Workloads HP ProLiant BladeSystem Gen9 vs Gen8 and G7 Server Blades on Data Warehouse Workloads Gen9 server blades give more performance per dollar for your investment. Executive Summary Information Technology (IT)

More information

Quality of Service Mechanism for MANET using Linux Semra Gulder, Mathieu Déziel

Quality of Service Mechanism for MANET using Linux Semra Gulder, Mathieu Déziel Quality of Service Mechanism for MANET using Linux Semra Gulder, Mathieu Déziel Semra.gulder@crc.ca, mathieu.deziel@crc.ca Abstract: This paper describes a QoS mechanism suitable for Mobile Ad Hoc Networks

More information

Mainstream Computer System Components

Mainstream Computer System Components Mainstream Computer System Components Double Date Rate (DDR) SDRAM One channel = 8 bytes = 64 bits wide Current DDR3 SDRAM Example: PC3-12800 (DDR3-1600) 200 MHz (internal base chip clock) 8-way interleaved

More information

ASPERA HIGH-SPEED TRANSFER. Moving the world s data at maximum speed

ASPERA HIGH-SPEED TRANSFER. Moving the world s data at maximum speed ASPERA HIGH-SPEED TRANSFER Moving the world s data at maximum speed ASPERA HIGH-SPEED FILE TRANSFER Aspera FASP Data Transfer at 80 Gbps Elimina8ng tradi8onal bo

More information

BurnInTest results BurnInTest Version Date Logging level Licenced System summary System summary General CPU

BurnInTest results BurnInTest Version Date Logging level Licenced System summary System summary General CPU BurnInTest results BurnInTest Version V8.0 Std 1040 Date Thu Apr 23 12:56:00 2015 Logging level rmal Licenced *** Unlicensed trial version *** System summary System summary Windows 8.1 Professional Edition

More information

Analytics of Wide-Area Lustre Throughput Using LNet Routers

Analytics of Wide-Area Lustre Throughput Using LNet Routers Analytics of Wide-Area Throughput Using LNet Routers Nagi Rao, Neena Imam, Jesse Hanley, Sarp Oral Oak Ridge National Laboratory User Group Conference LUG 2018 April 24-26, 2018 Argonne National Laboratory

More information

Improving Performance of 100G Data Transfer Nodes

Improving Performance of 100G Data Transfer Nodes Improving Performance of 100G Data Transfer Nodes Brian Tierney, Consultant Nate Hanford, ESnet bltierney@gmail.com http://fasterdata.es.net APAN, Singapore March 28, 2018 Recent TCP changes 2 Observation

More information

SSH Bulk Transfer Performance. Allan Jude --

SSH Bulk Transfer Performance. Allan Jude -- SSH Bulk Transfer Performance Allan Jude -- allanjude@freebsd.org Introduction 15 Years as FreeBSD Server Admin FreeBSD src/doc committer (ZFS, bhyve, ucl, xo) FreeBSD Core Team (July 2016-2018) Co-Author

More information

A Case Study in Optimizing GNU Radio s ATSC Flowgraph

A Case Study in Optimizing GNU Radio s ATSC Flowgraph A Case Study in Optimizing GNU Radio s ATSC Flowgraph Presented by Greg Scallon and Kirby Cartwright GNU Radio Conference 2017 Thursday, September 14 th 10am ATSC FLOWGRAPH LOADING 3% 99% 76% 36% 10% 33%

More information

BMEGUI Tutorial 1 Spatial kriging

BMEGUI Tutorial 1 Spatial kriging BMEGUI Tutorial 1 Spatial kriging 1. Objective The primary objective of this exercise is to get used to the basic operations of BMEGUI using a purely spatial dataset. The analysis will consist in an exploratory

More information

DPDK Vhost/Virtio Performance Report Release 18.05

DPDK Vhost/Virtio Performance Report Release 18.05 DPDK Vhost/Virtio Performance Report Test Date: Jun 1 2018 Author: Intel DPDK Validation Team Revision History Date Revision Comment Jun 1st, 2018 1.0 Initial document for release 2 Release 18.02 Contents

More information

Performance Optimizations via Connect-IB and Dynamically Connected Transport Service for Maximum Performance on LS-DYNA

Performance Optimizations via Connect-IB and Dynamically Connected Transport Service for Maximum Performance on LS-DYNA Performance Optimizations via Connect-IB and Dynamically Connected Transport Service for Maximum Performance on LS-DYNA Pak Lui, Gilad Shainer, Brian Klaff Mellanox Technologies Abstract From concept to

More information

Bandwidth estimations

Bandwidth estimations Bandwidth estimations for Test Traffic Measurements project Henk Uijterwaal Mark Santcroos RIPE NCC New Projects Group BEst2003, December 9 1 RIPE NCC Regional Internet Registry for Europe Based in Amsterdam,

More information

Oracle Exadata Recipes

Oracle Exadata Recipes Oracle Exadata Recipes A Problem-Solution Approach John Clarke Apress- Contents About the Author About the Technical Reviewer Acknowledgments Introduction xxxiii xxxv xxxvii xxxix Part 1: Exadata Architecture

More information

Firewall Performance Evaluation Secure Computing Sidewinder vs. Check Point NGX

Firewall Performance Evaluation Secure Computing Sidewinder vs. Check Point NGX Firewall Performance Evaluation vs. Check Point NGX June 1, 27 Author: Bill Bosen Firewall Performance Evaluation 7 vs. R62 June 1, 27 In view of the fact that firewall performance is becoming more and

More information

Introduction to PTC Windchill ProjectLink 11.0

Introduction to PTC Windchill ProjectLink 11.0 Introduction to PTC Windchill ProjectLink 11.0 Overview Course Code Course Length TRN-4756-T 8 Hours In this course, you will learn how to participate in and manage projects using Windchill ProjectLink

More information

Microsoft Windows 2016 Mellanox 100GbE NIC Tuning Guide

Microsoft Windows 2016 Mellanox 100GbE NIC Tuning Guide Microsoft Windows 2016 Mellanox 100GbE NIC Tuning Guide Publication # 56288 Revision: 1.00 Issue Date: June 2018 2018 Advanced Micro Devices, Inc. All rights reserved. The information contained herein

More information

716 West Ave Austin, TX USA

716 West Ave Austin, TX USA Fundamentals of Computer and Internet Fraud GLOBAL Headquarters the gregor building 716 West Ave Austin, TX 78701-2727 USA TABLE OF CONTENTS I. INTRODUCTION What Is Computer Crime?... 2 Computer Fraud

More information

Certification Document macle GmbH IBM System xx3650 M4 03/06/2014. macle GmbH IBM System x3650 M4

Certification Document macle GmbH IBM System xx3650 M4 03/06/2014. macle GmbH IBM System x3650 M4 macle GmbH IBM System x3650 M4 1 Executive summary After performing all tests, the Certification Document macle GmbH IBM System x3650 M4 system has been officially certified according to the Open-E Hardware

More information

Selecting the right Tesla/GTX GPU from a Drunken Baker's Dozen

Selecting the right Tesla/GTX GPU from a Drunken Baker's Dozen Selecting the right Tesla/GTX GPU from a Drunken Baker's Dozen GPU Computing Applications Here's what Nvidia says its Tesla K20(X) card excels at doing - Seismic processing, CFD, CAE, Financial computing,

More information

DTN End Host performance and tuning

DTN End Host performance and tuning DTN End Host performance and tuning 1 Gigabit Ethernet & NVME Disks Richard- Hughes Jones Senior Network Advisor, Office of the CTO GEANT AssociaOon - Cambridge Workshop: Moving My Data at High Speeds

More information

DPDK Vhost/Virtio Performance Report Release 18.11

DPDK Vhost/Virtio Performance Report Release 18.11 DPDK Vhost/Virtio Performance Report Test Date: December 3st 2018 Author: Intel DPDK Validation Team Revision History Date Revision Comment December 3st, 2018 1.0 Initial document for release 2 Contents

More information

Mellanox Technologies Maximize Cluster Performance and Productivity. Gilad Shainer, October, 2007

Mellanox Technologies Maximize Cluster Performance and Productivity. Gilad Shainer, October, 2007 Mellanox Technologies Maximize Cluster Performance and Productivity Gilad Shainer, shainer@mellanox.com October, 27 Mellanox Technologies Hardware OEMs Servers And Blades Applications End-Users Enterprise

More information

Performance Characteristics on Fast Ethernet and Gigabit networks

Performance Characteristics on Fast Ethernet and Gigabit networks Version 2.5 Traffic Generator and Measurement Tool for IP Networks (IPv4 & IPv6) FTTx, LAN, MAN, WAN, WLAN, WWAN, Mobile, Satellite, PLC, etc Performance Characteristics on Fast Ethernet and Gigabit networks

More information

Lenovo RD240 storage system

Lenovo RD240 storage system Certification Document Lenovo RD20 8 th June 2010 ver. 1.0 Lenovo RD20 storage system 1 Certification Document Lenovo RD20 8 th June 2010 ver. 1.0 Table of contents Test description and environment 3 Test

More information

Contents. Part I Introduction. Acknowledgments About the Authors. Chapter 1 Introduction to DTrace 1

Contents. Part I Introduction. Acknowledgments About the Authors. Chapter 1 Introduction to DTrace 1 Gregg.book Page v Wednesday, February 2, 2011 12:35 PM Foreword Preface Acknowledgments About the Authors xxi xxv xxxi xxxv Part I Introduction Chapter 1 Introduction to DTrace 1 What Is DTrace? 1 Why

More information

IT 341 Introduction to System Administration Project I Installing Ubuntu Server on a Virtual Machine

IT 341 Introduction to System Administration Project I Installing Ubuntu Server on a Virtual Machine IT 341 Introduction to System Administration Project I Installing Ubuntu Server on a Virtual Machine Here we create a new virtual machine and install Ubuntu 16.04 LTS Server on it. In this instance, we

More information

CA AppLogic and Dell PowerEdge R620 Equipment Validation

CA AppLogic and Dell PowerEdge R620 Equipment Validation CA AppLogic and Dell PowerEdge R620 Equipment Validation Opening Statement Last Updated May 14, 2013 The Dell PowerEdge R620 servers described below are compatible and validated with both CA AppLogic 3.1.14

More information

Independent consultant. (Ex-) Oracle ACE Director. Member of OakTable Network. Performance Troubleshooting In-house workshops

Independent consultant. (Ex-) Oracle ACE Director. Member of OakTable Network. Performance Troubleshooting In-house workshops Independent consultant Performance Troubleshooting In-house workshops Cost-Based Optimizer Performance By Design (Ex-) Oracle ACE Director 2009-2016 Alumni Member of OakTable Network http://oracle-randolf.blogspot.com

More information

Testing the Channel Aggregation Capability of the MPT Multipath Communication Library

Testing the Channel Aggregation Capability of the MPT Multipath Communication Library Testing the Channel Aggregation Capability of the MPT Multipath Communication Library Gábor Lencse Department of Telecommunications Széchenyi István University Győr, Hungary lencse@sze.hu Ákos Kovács Department

More information

Introduction... xiii Chapter 1: Introduction to Computer Networks and Internet Computer Networks Uses of Computer Networks...

Introduction... xiii Chapter 1: Introduction to Computer Networks and Internet Computer Networks Uses of Computer Networks... Table of Contents Introduction... xiii Chapter 1: Introduction to Computer Networks and Internet... 1 1.1 Computer Networks... 1 1.1.1 Advantages of Computer Networks... 2 1.1.2 Disadvantages of Computer

More information

PAC094 Performance Tips for New Features in Workstation 5. Anne Holler Irfan Ahmad Aravind Pavuluri

PAC094 Performance Tips for New Features in Workstation 5. Anne Holler Irfan Ahmad Aravind Pavuluri PAC094 Performance Tips for New Features in Workstation 5 Anne Holler Irfan Ahmad Aravind Pavuluri Overview of Talk Virtual machine teams 64-bit guests SMP guests e1000 NIC support Fast snapshots Virtual

More information

IBM WebSphere MQ Low Latency Messaging Software Tested With Arista 10 Gigabit Ethernet Switch and Mellanox ConnectX

IBM WebSphere MQ Low Latency Messaging Software Tested With Arista 10 Gigabit Ethernet Switch and Mellanox ConnectX IBM WebSphere MQ Low Latency Messaging Software Tested With Arista 10 Gigabit Ethernet Switch and Mellanox ConnectX -2 EN with RoCE Adapter Delivers Reliable Multicast Messaging With Ultra Low Latency

More information

Performance Characteristics on Fast Ethernet, Gigabit and 10 Gigabits networks

Performance Characteristics on Fast Ethernet, Gigabit and 10 Gigabits networks Versions 2.6 Traffic Generator for IP Networks (IPv4 & IPv6) FTTx, LAN, MAN, WAN, WLAN, WWAN, Mobile, Satellite, PLC, etc. Performance Characteristics on Fast Ethernet, Gigabit and 10 Gigabits networks

More information

IBM V7000 Unified R1.4.2 Asynchronous Replication Performance Reference Guide

IBM V7000 Unified R1.4.2 Asynchronous Replication Performance Reference Guide V7 Unified Asynchronous Replication Performance Reference Guide IBM V7 Unified R1.4.2 Asynchronous Replication Performance Reference Guide Document Version 1. SONAS / V7 Unified Asynchronous Replication

More information

Computer System Components

Computer System Components Computer System Components CPU Core 1 GHz - 3.2 GHz 4-way Superscaler RISC or RISC-core (x86): Deep Instruction Pipelines Dynamic scheduling Multiple FP, integer FUs Dynamic branch prediction Hardware

More information

Reducing Network Contention with Mixed Workloads on Modern Multicore Clusters

Reducing Network Contention with Mixed Workloads on Modern Multicore Clusters Reducing Network Contention with Mixed Workloads on Modern Multicore Clusters Matthew Koop 1 Miao Luo D. K. Panda matthew.koop@nasa.gov {luom, panda}@cse.ohio-state.edu 1 NASA Center for Computational

More information

Introduction to Ethernet Latency

Introduction to Ethernet Latency Introduction to Ethernet Latency An Explanation of Latency and Latency Measurement The primary difference in the various methods of latency measurement is the point in the software stack at which the latency

More information

Dell PowerEdge R720xd with PERC H710P: A Balanced Configuration for Microsoft Exchange 2010 Solutions

Dell PowerEdge R720xd with PERC H710P: A Balanced Configuration for Microsoft Exchange 2010 Solutions Dell PowerEdge R720xd with PERC H710P: A Balanced Configuration for Microsoft Exchange 2010 Solutions A comparative analysis with PowerEdge R510 and PERC H700 Global Solutions Engineering Dell Product

More information

Use of the Internet SCSI (iscsi) protocol

Use of the Internet SCSI (iscsi) protocol A unified networking approach to iscsi storage with Broadcom controllers By Dhiraj Sehgal, Abhijit Aswath, and Srinivas Thodati In environments based on Internet SCSI (iscsi) and 10 Gigabit Ethernet, deploying

More information

CA AppLogic and Dell PowerEdge R420 Equipment Validation

CA AppLogic and Dell PowerEdge R420 Equipment Validation CA AppLogic and Dell PowerEdge R420 Equipment Validation Opening Statement Last Updated May 3, 2013 The Dell PowerEdge R420 servers described below are compatible and validated with both CA AppLogic 3.1.14

More information

An FPGA-Based Optical IOH Architecture for Embedded System

An FPGA-Based Optical IOH Architecture for Embedded System An FPGA-Based Optical IOH Architecture for Embedded System Saravana.S Assistant Professor, Bharath University, Chennai 600073, India Abstract Data traffic has tremendously increased and is still increasing

More information

TIBCO, HP and Mellanox High Performance Extreme Low Latency Messaging

TIBCO, HP and Mellanox High Performance Extreme Low Latency Messaging TIBCO, HP and Mellanox High Performance Extreme Low Latency Messaging Executive Summary: With the recent release of TIBCO FTL TM, TIBCO is once again changing the game when it comes to providing high performance

More information

"Charting the Course... MOC A Developing Microsoft SQL Server 2012 Databases. Course Summary

Charting the Course... MOC A Developing Microsoft SQL Server 2012 Databases. Course Summary Course Summary Description This 5-day instructor-led course introduces SQL Server 2012 and describes logical table design, indexing and query plans. It also focuses on the creation of database objects

More information

Intra-MIC MPI Communication using MVAPICH2: Early Experience

Intra-MIC MPI Communication using MVAPICH2: Early Experience Intra-MIC MPI Communication using MVAPICH: Early Experience Sreeram Potluri, Karen Tomko, Devendar Bureddy, and Dhabaleswar K. Panda Department of Computer Science and Engineering Ohio State University

More information

Eglobal Technology Co., Limited

Eglobal Technology Co., Limited Model No.: V8 6100U/V8 6200U Specifications Feature Full aluminium silver or black color shell, Exquisite production craft on ouside design Fanless system without cooling fan and slim design 200*200*40MM

More information

Performance Benefits of OpenVMS V8.4 Running on BL8x0c i2 Server Blades

Performance Benefits of OpenVMS V8.4 Running on BL8x0c i2 Server Blades Performance Benefits of OpenVMS V8.4 Running on BL8xc i2 Server Blades A detailed review of performance features and test results for OpenVMS V8.4. March 211 211, TechWise Research. All Rights Reserved

More information

Stacked Vlan: Performance Improvement and Challenges

Stacked Vlan: Performance Improvement and Challenges Stacked Vlan: Performance Improvement and Challenges Toshiaki Makita NTT Tokyo, Japan makita.toshiaki@lab.ntt.co.jp Abstract IEEE 802.1ad vlan protocol type was introduced in kernel 3.10, which has encouraged

More information

Gen 6 Fibre Channel Evaluation of Products from Emulex and Brocade

Gen 6 Fibre Channel Evaluation of Products from Emulex and Brocade Gen 6 Fibre Channel Evaluation of Products from Emulex and Brocade Gen 6 Fibre Channel provides new speeds and features for enterprise datacenters. Executive Summary Large enterprises choose Fibre Channel

More information

Energy Efficient Servers

Energy Efficient Servers Energy Efficient Servers Blueprints for Data Center Optimization Corey Gough Ian Steiner Winston Saunders Contents J About the Authors About the Technical Reviewers Contributing Authors Acknowledgments

More information

Annexure I: Contact Details:

Annexure I: Contact Details: Ref: CO/IT-BPR/CSC Date:.09.2017 Annexure I: Contact Details: a) Name of the company b) Company s address in India c) Contact person d) Telephone no. e) Fax f) E-mail address g) Service tax registration

More information

Evaluating the Impact of RDMA on Storage I/O over InfiniBand

Evaluating the Impact of RDMA on Storage I/O over InfiniBand Evaluating the Impact of RDMA on Storage I/O over InfiniBand J Liu, DK Panda and M Banikazemi Computer and Information Science IBM T J Watson Research Center The Ohio State University Presentation Outline

More information

CROSS-REFERENCE TABLE ASME A Including A17.1a-1997 Through A17.1d 2000 vs. ASME A

CROSS-REFERENCE TABLE ASME A Including A17.1a-1997 Through A17.1d 2000 vs. ASME A CROSS-REFERENCE TABLE ASME Including A17.1a-1997 Through A17.1d 2000 vs. ASME 1 1.1 1.1 1.1.1 1.2 1.1.2 1.3 1.1.3 1.4 1.1.4 2 1.2 3 1.3 4 Part 9 100 2.1 100.1 2.1.1 100.1a 2.1.1.1 100.1b 2.1.1.2 100.1c

More information

6.9. Communicating to the Outside World: Cluster Networking

6.9. Communicating to the Outside World: Cluster Networking 6.9 Communicating to the Outside World: Cluster Networking This online section describes the networking hardware and software used to connect the nodes of cluster together. As there are whole books and

More information

Introduction to Creo Elements/Direct 19.0 Modeling

Introduction to Creo Elements/Direct 19.0 Modeling Introduction to Creo Elements/Direct 19.0 Modeling Overview Course Code Course Length TRN-4531-T 3 Day In this course, you will learn the basics about 3-D design using Creo Elements/Direct Modeling. You

More information

BFW-037 B.Sc. FOOTWEAR TECHNOLOGY (BSCFWT) Term-End Examination. December, 2014

BFW-037 B.Sc. FOOTWEAR TECHNOLOGY (BSCFWT) Term-End Examination. December, 2014 No. of Printed Pages : 8 BFW-037 B.Sc. FOOTWEAR TECHNOLOGY (BSCFWT) Term-End Examination 00190 December, 2014 BFW-037 : COMPUTER SCIENCE Time : 3 hours Maximum Marks : 70 Note : (i) There are two sections

More information

SHSAT Review Class Week 3-10/21/2016

SHSAT Review Class Week 3-10/21/2016 SHSAT Review Class Week 3-10/21/2016 Week Two Agenda 1. Going over HW (Test 2) 2. Review of Geometry - Practice set 3. Questions before we leave Test 2 Questions? Ask about any questions you were confused

More information

The NE010 iwarp Adapter

The NE010 iwarp Adapter The NE010 iwarp Adapter Gary Montry Senior Scientist +1-512-493-3241 GMontry@NetEffect.com Today s Data Center Users Applications networking adapter LAN Ethernet NAS block storage clustering adapter adapter

More information

Contents. Introduction to Networking. Preface...i. Introduction... xix

Contents. Introduction to Networking. Preface...i. Introduction... xix Contents Introduction to Networking Introduction to Networking Preface...i Key Topics...i Course Objectives...ii Pedagogical Features...ii Learning Objectives, Unit Summaries, Discussion Questions, and

More information

Summary of Contents LIST OF FIGURES LIST OF TABLES

Summary of Contents LIST OF FIGURES LIST OF TABLES Summary of Contents LIST OF FIGURES LIST OF TABLES PREFACE xvii xix xxi PART 1 BACKGROUND Chapter 1. Introduction 3 Chapter 2. Standards-Makers 21 Chapter 3. Principles of the S2ESC Collection 45 Chapter

More information

Configuring BGP on Cisco Routers Volume 1

Configuring BGP on Cisco Routers Volume 1 Volume 1 I. Course Introduction A. Overview/Learner Skills and Knowledge B. Course Flow C. Additional References 1. Cisco Glossary of Terms D. Your Training Curriculum II. BGP Overview III. Introducing

More information

Implementation and Analysis of Large Receive Offload in a Virtualized System

Implementation and Analysis of Large Receive Offload in a Virtualized System Implementation and Analysis of Large Receive Offload in a Virtualized System Takayuki Hatori and Hitoshi Oi The University of Aizu, Aizu Wakamatsu, JAPAN {s1110173,hitoshi}@u-aizu.ac.jp Abstract System

More information

CA AppLogic and Dell PowerEdge R820 Equipment Validation

CA AppLogic and Dell PowerEdge R820 Equipment Validation CA AppLogic and Dell PowerEdge R820 Equipment Validation Opening Statement Last Updated May 14, 2013 The Dell PowerEdge R820 servers described below are compatible and validated with both CA AppLogic 3.1.14

More information

IBM Emulex 16Gb Fibre Channel HBA Evaluation

IBM Emulex 16Gb Fibre Channel HBA Evaluation IBM Emulex 16Gb Fibre Channel HBA Evaluation Evaluation report prepared under contract with Emulex Executive Summary The computing industry is experiencing an increasing demand for storage performance

More information