Performance Report PRIMERGY BX924 S4 - Fujitsu · PDF fileWhite Paper Performance Report PRIMERGY BX924 S4 ... or RAID 1 for up to 2 × 2.5˝ SATA HDDs, ... time which is required

White Paper Performance Report PRIMERGY BX924 S4

http://ts.fujitsu.com/primergy Page 1 (34)

White Paper Fujitsu PRIMERGY Servers Performance Report PRIMERGY BX924 S4

This document contains a summary of the benchmarks executed for the PRIMERGY BX924 S4.

The PRIMERGY BX924 S4 performance data are compared with the data of other PRIMERGY models and discussed. In addition to the benchmark results, an explanation has been included for each benchmark and for the benchmark environment.

Version

1.0

2013-11-14

White Paper Performance Report PRIMERGY BX924 S4 Version: 1.0 2013-11-14

Page 2 (34) http://ts.fujitsu.com/primergy

Contents

Document history

Version 1.0

New:

Technical data SPECcpu2006

Results for Xeon E5-2600 v2 processor series Disk I/O

Measurements with “LSI SW RAID on Intel C600 (Onboard SATA)”, “LSI SW RAID on Intel C600 (Onboard SAS)” and “PY SAS RAID Mezz Card 6Gb” controllers

OLTP-2 Results for Xeon E5-2600 v2 processor series

vServCon Results for Xeon E5-2600 v2 processor series

STREAM Measurements with Xeon E5-2600 v2 processor series

LINPACK Measurements with Xeon E5-2600 v2 processor series

Document history ................................................................................................................................................ 2

Technical data .................................................................................................................................................... 3

SPECcpu2006 .................................................................................................................................................... 5

Disk I/O ............................................................................................................................................................. 11

OLTP-2 ............................................................................................................................................................. 17

vServCon .......................................................................................................................................................... 21

STREAM ........................................................................................................................................................... 28

LINPACK .......................................................................................................................................................... 30

Literature ........................................................................................................................................................... 33

Contact ............................................................................................................................................................. 34



Technical data

Decimal prefixes according to the SI standard are used for measurement units in this white paper (e.g. 1 GB = 10

9 bytes). In contrast, these prefixes should be interpreted as binary prefixes (e.g. 1 GB = 2

30 bytes) for

the capacities of caches and storage modules. Separate reference will be made to any further exceptions where applicable.

Model PRIMERGY BX924 S4

Form factor Server blade

Chipset Intel C600 series

Number of sockets 2

Number of processors orderable 1 or 2

Processor type Intel® Xeon

® series E5-2600 v2

Number of memory slots 24 (12 per processor)

Maximum memory configuration 1536 GB

Onboard LAN controller 2 × 10 Gbit/s CNA

Onboard HDD controller Controller with RAID 0 or RAID 1 for up to 2 × 2.5˝ SATA HDDs, optional: SAS Enabling Key for Onboard Ports for up to 2 × 2.5˝ SAS HDDs

PCI slots 2 × PCI-Express 3.0 x8

Max. number of internal hard disks 2

The processor frequency specified in the following table is always at least achieved given full utilization. Processors with Turbo Boost Technology 2.0 additionally permit automatically regulated, dynamic overclocking. The overclocking rate depends on the utilization of the processor and its ambient conditions. As far as utilization is concerned, the number of cores subject to utilization as well as the type and strength of core utilization play a role. Added to these as influencing factors are the strength of the heating, the level of the ambient temperature and the heat dissipation options. As a result of overclocking it is even possible to exceed the thermal design power of the processor for short periods of time.

How much a processor benefits from the Turbo mode in an individual case depends on the respective application and can in some application scenarios even differ from processor example to processor example.

PRIMERGY BX924 S4



Processors (since system release)

Processor

Co

res

Th

rea

ds Cache

[MB]

QPI Speed

[GT/s]

Processor Frequency

[Ghz]

Max. Turbo

Frequency at full load

[Ghz]

Max. Turbo

Frequency

[Ghz]

Max. Memory

Frequency

[MHz]

TDP

[Watt]

Xeon E5-2603 v2 4 4 10 6.40 1.80 n/a n/a 1333 80

Xeon E5-2609 v2 4 4 10 6.40 2.50 n/a n/a 1333 80

Xeon E5-2637 v2 4 8 15 8.00 3.50 3.60 3.80 1866 130

Xeon E5-2620 v2 6 12 15 7.20 2.10 2.40 2.60 1600 80

Xeon E5-2630Lv2 6 12 15 7.20 2.40 2.60 2.80 1600 60

Xeon E5-2630 v2 6 12 15 7.20 2.60 2.90 3.10 1600 80

Xeon E5-2643 v2 6 12 25 8.00 3.50 3.60 3.80 1866 130

Xeon E5-2640 v2 8 16 20 7.20 2.00 2.30 2.50 1600 95

Xeon E5-2650 v2 8 16 20 8.00 2.60 3.00 3.40 1866 95

Xeon E5-2667 v2 8 16 25 8.00 3.30 3.60 4.00 1866 130

Xeon E5-2650Lv2 10 20 25 7.20 1.70 1.90 2.10 1600 70

Xeon E5-2660 v2 10 20 25 8.00 2.20 2.60 3.00 1866 95

Xeon E5-2670 v2 10 20 25 8.00 2.50 2.90 3.30 1866 115

Xeon E5-2680 v2 10 20 25 8.00 2.80 3.10 3.60 1866 115

Xeon E5-2690 v2 10 20 25 8.00 3.00 3.30 3.60 1866 130

Xeon E5-2695 v2 12 24 30 8.00 2.40 2.80 3.20 1866 115

Xeon E5-2697 v2 12 24 30 8.00 2.70 3.00 3.50 1866 130

Memory modules (since system release)

Memory module

Cap

ac

ity [

GB

]

Ran

ks

Bit

wid

th o

f th

e

me

mo

ry c

hip

s

Fre

qu

en

cy

[M

Hz]

Lo

w v

olt

ag

e

Lo

ad

red

uc

ed

Reg

iste

red

EC

C

4GB (1x4GB) 1Rx4 L DDR3-1600 R ECC (4 GB 1Rx4 PC3L-12800R)

4 1 4 1600


8 1 4 1600

8GB (1x8GB) 2Rx8 DDR3-1866 R ECC (8 GB 2Rx8 PC3-14900R)

8 2 8 1866


16 2 4 1600

16GB (1x16GB) 2Rx4 DDR3-1866 R ECC (16 GB 2Rx4 PC3-14900R)

16 2 4 1866

32GB (1x32GB) 4Rx4 L DDR3-1600 LR ECC (32 GB 4Rx4 PC3L-12800L)

32 4 4 1600

64GB (1x64GB) 8Rx4 L DDR3-1333 LR ECC (64 GB 8Rx4 PC3L-10600L)

64 8 4 1333

Some components may not be available in all countries or sales regions.

Detailed technical information is available in the data sheet PRIMERGY BX924 S4.

http://docs.ts.fujitsu.com/dl.aspx?id=94ad5e5b-e23c-445a-8bc8-792313ebd7e4



SPECcpu2006

Benchmark description

SPECcpu2006 is a benchmark which measures the system efficiency with integer and floating-point operations. It consists of an integer test suite (SPECint2006) containing 12 applications and a floating-point test suite (SPECfp2006) containing 17 applications. Both test suites are extremely computing-intensive and concentrate on the CPU and the memory. Other components, such as Disk I/O and network, are not measured by this benchmark.

SPECcpu2006 is not tied to a special operating system. The benchmark is available as source code and is compiled before the actual measurement. The used compiler version and their optimization settings also affect the measurement result.

SPECcpu2006 contains two different performance measurement methods: the first method (SPECint2006 or SPECfp2006) determines the time which is required to process single task. The second method (SPECint_rate2006 or SPECfp_rate2006) determines the throughput, i.e. the number of tasks that can be handled in parallel. Both methods are also divided into two measurement runs, “base” and “peak” which differ in the use of compiler optimization. When publishing the results the base values are always used; the peak values are optional.

Benchmark Arithmetics Type Compiler optimization

Measurement result

Application

SPECint2006 integer peak aggressive Speed single-threaded

SPECint_base2006 integer base conservative

SPECint_rate2006 integer peak aggressive Throughput multi-threaded

SPECint_rate_base2006 integer base conservative

SPECfp2006 floating point peak aggressive Speed single-threaded

SPECfp_base2006 floating point base conservative

SPECfp_rate2006 floating point peak aggressive Throughput multi-threaded

SPECfp_rate_base2006 floating point base conservative

The measurement results are the geometric average from normalized ratio values which have been determined for individual benchmarks. The geometric average - in contrast to the arithmetic average - means that there is a weighting in favour of the lower individual results. Normalized means that the measurement is how fast is the test system compared to a reference system. Value “1” was defined for the SPECint_base2006-, SPECint_rate_base2006, SPECfp_base2006 and SPECfp_rate_base2006 results of the reference system. For example, a SPECint_base2006 value of 2 means that the measuring system has handled this benchmark twice as fast as the reference system. A SPECfp_rate_base2006 value of 4 means that the measuring system has handled this benchmark some 4/[# base copies] times faster than the reference system. “# base copies” specify how many parallel instances of the benchmark have been executed.

Not every SPECcpu2006 measurement is submitted by us for publication at SPEC. This is why the SPEC web pages do not have every result. As we archive the log files for all measurements, we can prove the correct implementation of the measurements at any time.



Benchmark environment

System Under Test (SUT)

Hardware

Enclosure PRIMERGY BX900 S2


Processor 2 processors of Xeon E5-2600 v2 processor series

Memory 16 × 16GB (1x16GB) 2Rx4 DDR3-1866 R ECC

Software

BIOS settings Energy Performance = Performance SPECint_base2006, SPECint2006, SPECfp_base2006, SPECfp2006:

Utilization Profile = Unbalanced

Operating system Red Hat Enterprise Linux Server release 6.4

Operating system settings

echo always > /sys/kernel/mm/redhat_transparent_hugepage/enabled

Compiler Intel C++/Fortran Compiler 14.0




Benchmark results

In terms of processors the benchmark result depends primarily on the size of the processor cache, the support for Hyper-Threading, the number of processor cores and on the processor frequency. In the case of processors with Turbo mode the number of cores, which are loaded by the benchmark, determines the maximum processor frequency that can be achieved. In the case of single-threaded benchmarks, which largely load one core only, the maximum processor frequency that can be achieved is higher than with multi-threaded benchmarks (see the processor table in the section "Technical Data").

The results marked (est.) are estimates which are based on PRIMERGY RX300 S8 measurement results.

Processor

Nu

mb

er

of

pro

ce

sso

rs

SP

EC

int_

ba

se2

006

SP

EC

int2

00

6

Nu

mb

er

of

pro

ce

sso

rs

SP

EC

int_

rate

_b

as

e20

06

SP

EC

int_

rate

200

6

Nu

mb

er

of

pro

ce

sso

rs

SP

EC

int_

rate

_b

as

e20

06

SP

EC

int_

rate

200

6

Xeon E5-2603 v2 2 27.9 29.0 1 91.6 (est.) 95.2 (est.) 2 179 185

Xeon E5-2609 v2 2 37.9 39.8 1 123 (est.) 128 (est.) 2 240 248

Xeon E5-2637 v2 2 56.1 59.9 1 212 (est.) 220 (est.) 2 411 428

Xeon E5-2620 v2 2 40.3 42.7 1 212 (est.) 221 (est.) 2 413 428

Xeon E5-2630Lv2 2 43.0 45.5 1 228 (est.) 237 (est.) 2 442 459

Xeon E5-2630 v2 2 47.3 50.5 1 249 (est.) 260 (est.) 2 484 502

Xeon E5-2643 v2 2 59.3 63.3 1 314 (est.) 326 (est.) 2 610 635

Xeon E5-2640 v2 2 40.1 42.7 1 268 (est.) 278 (est.) 2 520 540

Xeon E5-2650 v2 2 52.4 56.4 1 338 (est.) 351 (est.) 2 658 682

Xeon E5-2667 v2 2 62.6 67.5 1 399 (est.) 413 (est.) 2 776 806

Xeon E5-2650Lv2 2 34.4 36.7 1 278 (est.) 289 (est.) 2 542 563

Xeon E5-2660 v2 2 48.1 51.8 1 368 (est.) 380 (est.) 2 716 743

Xeon E5-2670 v2 2 53.0 56.9 1 401 (est.) 415 (est.) 2 782 810

Xeon E5-2680 v2 2 56.7 61.5 1 422 (est.) 436 (est.) 2 823 852

Xeon E5-2690 v2 2 57.6 61.8 1 442 (est.) 457 (est.) 2 863 892

Xeon E5-2695 v2 2 51.0 55.1 1 453 (est.) 468 (est.) 2 883 914

Xeon E5-2697 v2 2 55.3 60.0 1 477 (est.) 493 (est.) 2 931 962



Processor

Nu

mb

er

of

pro

ce

sso

rs

SP

EC

fp_

ba

se2

00

6

SP

EC

fp2

00

6

Nu

mb

er

of

pro

ce

sso

rs

SP

EC

fp_

rate

_b

as

e20

06

SP

EC

fp_

rate

20

06

Nu

mb

er

of

pro

ce

sso

rs

SP

EC

fp_

rate

_b

as

e20

06

SP

EC

fp_

rate

20

06

Xeon E5-2603 v2 2 51.0 52.5 1 103 (est.) 105 (est.) 2 199 203

Xeon E5-2609 v2 2 64.4 66.3 1 131 (est.) 133 (est.) 2 254 259

Xeon E5-2637 v2 2 94.1 97.5 1 196 (est.) 201 (est.) 2 383 394

Xeon E5-2620 v2 2 72.8 75.7 1 193 (est.) 197 (est.) 2 378 386

Xeon E5-2630Lv2 2 77.1 80.0 1 202 (est.) 206 (est.) 2 397 406

Xeon E5-2630 v2 2 83.0 85.9 1 215 (est.) 220 (est.) 2 423 432

Xeon E5-2643 v2 2 101 105 1 163 (est.) 270 (est.) 2 517 530

Xeon E5-2640 v2 2 74.4 77.3 1 228 (est.) 233 (est.) 2 448 458

Xeon E5-2650 v2 2 95.2 99.5 1 276 (est.) 283 (est.) 2 543 556

Xeon E5-2667 v2 2 108 112 1 302 (est.) 311 (est.) 2 594 610

Xeon E5-2650Lv2 2 66.9 69.7 1 234 (est.) 239 (est.) 2 459 470

Xeon E5-2660 v2 2 88.3 92.4 1 290 (est.) 298 (est.) 2 570 585

Xeon E5-2670 v2 2 94.8 99.1 1 306 (est.) 314 (est.) 2 602 618

Xeon E5-2680 v2 2 100 105 1 315 (est.) 324 (est.) 2 621 638

Xeon E5-2690 v2 2 102 106 1 323 (est.) 332 (est.) 2 632 649

Xeon E5-2695 v2 2 92.4 96.8 1 333 (est.) 342 (est.) 2 656 674

Xeon E5-2697 v2 2 98.0 103 1 343 (est.) 353 (est.) 2 678 696



The following four diagrams illustrate the throughput of the PRIMERGY BX924 S4 in comparison to its predecessor PRIMERGY BX924 S3, in their respective most performant configuration.

SPECint_rate_base2006

SPECint_rate2006

0

100

200

300

400

500

600

700

800

900

1000

PRIMERGY BX924 S32 x Xeon E5-2690

PRIMERGY BX924 S42 x Xeon E5-2697 v2

669

931

697

962

SPECint_base2006

SPECint2006

0

10

20

30

40

50

60

70



56.0

62.6

60.8

67.5

SPECcpu2006: integer performance PRIMERGY BX924 S4 vs. PRIMERGY BX924 S3

SPECcpu2006: integer performance PRIMERGY BX924 S4 vs. PRIMERGY BX924 S3



SPECfp_rate_base2006

SPECfp_rate2006

0

100

200

300

400

500

600

700



493

678

508

696

SPECfp_base2006

SPECfp2006

0

20

40

60

80

100

120



88.3

10893.2

112

SPECcpu2006: floating-point performance PRIMERGY BX924 S4 vs. PRIMERGY BX924 S3

SPECcpu2006: floating-point performance PRIMERGY BX924 S4 vs. PRIMERGY BX924 S3



Disk I/O


Performance measurements of disk subsystems for PRIMERGY servers are used to assess their performance and enable a comparison of the different storage connections for PRIMERGY servers. As standard, these performance measurements are carried out with a defined measurement method, which models the hard disk accesses of real application scenarios on the basis of specifications.

The essential specifications are:

Share of random accesses / sequential accesses Share of read / write access types Block size (kB) Number of parallel accesses (# of outstanding I/Os)

A given value combination of these specifications is known as “load profile”. The following five standard load profiles can be allocated to typical application scenarios:

In order to model applications that access in parallel with a different load intensity, the “# of Outstanding I/Os” is increased, starting with 1, 3, 8 and going up to 512 (from 8 onwards in increments to the power of two).

The measurements of this document are based on these standard load profiles.

The main results of a measurement are:

Throughput [MB/s] Throughput in megabytes per second Transactions [IO/s] Transaction rate in I/O operations per second Latency [ms] Average response time in ms

The data throughput has established itself as the normal measurement variable for sequential load profiles, whereas the measurement variable “transaction rate” is mostly used for random load profiles with their small block sizes. Data throughput and transaction rate are directly proportional to each other and can be transferred to each other according to the formula

Data throughput [MB/s] = Transaction rate [IO/s] × Block size [MB]

Transaction rate [IO/s] = Data throughput [MB/s] / Block size [MB]

This section specifies hard disk capacities on a basis of 10 (1 TB = 1012

bytes) while all other capacities, file sizes, block sizes and throughputs are specified on a basis of 2 (1 MB/s = 2

20 bytes/s).

All the details of the measurement method and the basics of disk I/O performance are described in the white paper “Basics of Disk I/O Performance”.

Standard load profile

Access Type of access Block size [kB]

Application

read write

File copy random 50% 50% 64 Copying of files

File server random 67% 33% 64 File server

Database random 67% 33% 8 Database (data transfer) Mail server

Streaming sequential 100% 0% 64 Database (log file), Data backup; Video streaming (partial)

Restore sequential 0% 100% 64 Restoring of files

http://docs.ts.fujitsu.com/dl.aspx?id=65781a00-556f-4a98-90a7-7022feacc602




All the measurement results discussed in this chapter were determined using the hardware and software components listed below:


Hardware

Controller 1 × “LSI SW RAID on Intel C600 (Onboard SATA)” 1 × “LSI SW RAID on Intel C600 (Onboard SAS)” 1 × “PY SAS RAID Mezz Card 6Gb”

Drive 2 × EP SSD SATA 6 Gbit/s 2.5 200 GB MLC

2 × EP SSD SAS 6 Gbit/s 2.5 200 GB MLC

Software

Operating system Microsoft Windows Server 2008 Enterprise x64 Edition SP2 Microsoft Windows Server 2012 Standard

Administration software

ServerView RAID Manager 5.7.2

Initialization of RAID arrays

RAID arrays are initialized before the measurement with an elementary block size of 64 kB (“stripe size”)

File system NTFS

Measuring tool Iometer 2006.07.27

Measurement data Measurement files of 32 GB with 1 – 8 hard disks; 64 GB with 9 – 16 hard disks; 128 GB with 17 or more hard disks

Some components may not be available in all countries / sales regions.

Benchmark results

The results presented here are designed to help you choose the right solution from the various configuration options of the PRIMERGY BX924 S4 in the light of disk-I/O performance. The selection of suitable components and the right settings of their parameters is important here. These two aspects should therefore be dealt with as preparation for the discussion of the performance values.

Components

The hard disks are the first essential component. If there is a reference below to “hard disks”, this is meant as the generic term for HDDs (“hard disk drives”, in other words conventional hard disks) and SSDs (“solid state drives”, i.e. non-volatile electronic storage media). When selecting the type of hard disk and number of hard disks you can move the weighting in the direction of storage capacity, performance, security or price. In order to enable a pre-selection of the hard disk types – depending on the required weighting – the hard disk types for PRIMERGY servers are divided into three classes:

“Economic” (ECO): low-priced hard disks “Business Critical” (BC): very failsafe hard disks “Enterprise” (EP): very failsafe and very high-performance hard disks

The following table is a list of the hard disk types that have been available for the PRIMERGY BX924 S4 since system release.

Drive class

Data medium type

Interface Form factor

krpm

Economic *) HDD SATA 6G 2.5" 5.4

Enterprise SSD SATA 6G 2.5" -

Enterprise SSD SAS 6G 2.5" -

*) “Advanced Format” technology with 512-byte sector emulation (512e)



Mixed drive configurations of SAS and SATA hard disks are not possible.

Of all the hard disk types SSDs offer on the one hand by far the highest transaction rates for random load profiles, and on the other hand the shortest access times. In return, however, the price per gigabyte of storage capacity is substantially higher.

More detailed performance statements about hard disk types are available in the white paper “Single Disk Performance”.

More information about 512e HDDs is available in the white paper “512e HDDs: Technology, Performance, Configurations”.

The maximum number of hard disks in the system depends on the system configuration. The following table lists the essential cases.

Form factor

Interface Connection

type Number of PCIe

controllers Maximum number

of hard disks

2.5" SATA 3G, SAS 3G direct 0 2

2.5" SATA 3G/6G, SAS 6G direct 1 2

After the hard disks the RAID controller is the second performance-determining key component. In the case of these controllers the “modular RAID” concept of the PRIMERGY servers offers a plethora of options to meet the various requirements of a wide range of different application scenarios.

The following table summarizes the most important features of the available RAID controllers of the PRIMERGY BX924 S4. A short alias is specified here for each controller, which is used in the subsequent list of the performance values.

Controller name Alias Cache Supported interfaces

Max. # disks in the system

RAID levels in the system

BBU/ FBU

LSI SW RAID on Intel C600 (Onboard SATA)

Patsburg A - SATA 3G - 2 × 2.5" 0, 1 -/-

LSI SW RAID on Intel C600 (Onboard SAS)

Patsburg B - SATA 3G SAS 3G

- 2 × 2.5" 0, 1 -/-

PY SAS RAID Mezz Card 6Gb

LSI2108 512 MB SATA 3G/6G SAS 3G/6G

PCIe 2.0 x8

2 × 2.5" 0, 1 /-

The onboard RAID controller is implemented in the chipset Intel C600 on the motherboard of the server and uses the CPU of the server for the RAID functionality. This controller is a simple solution that does not require a PCIe slot. In addition to the invariably available connection option of SATA hard disks, the additional SAS functionality can be activated via an “SAS enabling key”.

http://docs.ts.fujitsu.com/dl.aspx?id=0e30cb69-44db-4cd5-92a7-d38bacec6a99


http://docs.ts.fujitsu.com/dl.aspx?id=f5550c48-d4db-47f6-ab9d-ce135eaacf81




System-specific interfaces

The interfaces of a controller to the motherboard and to the hard disks have in each case specific limits for data throughput. These limits are listed in the following table. The minimum of these two values is a definite limit, which cannot be exceeded. This value is highlighted in bold in the following table.

Controller alias

Effective in the configuration Connection via expander # Disk

channels Limit for throughput of disk interface

PCIe version

PCIe width

Limit for throughput of PCIe interface

Patsburg A 2 × SATA 3G 487 MB/s - - - -

Patsburg B 2 × SAS 3G 487 MB/s - - - -

LSI2108 2 × SAS 6G 973 MB/s 2.0 x8 3433 MB/s -

More details about the RAID controllers of the PRIMERGY systems are available in the white paper “RAID Controller Performance”.

Settings

In most cases, the cache of the hard disks has a great influence on disk-I/O performance. This is particular valid for HDDs. It is frequently regarded as a security problem in case of power failure and is thus switched off. On the other hand, it was integrated by hard disk manufacturers for the good reason of increasing the write performance. For performance reasons it is therefore advisable to enable the hard disk cache. This is particular valid for SATA-HDDs. The performance can as a result increase more than tenfold for specific access patterns and hard disk types. More information about the performance impact of the hard disk cache is available in the document “Single Disk Performance”. To prevent data loss in case of power failure you are recommended to equip the system with a UPS.

In the case of controllers with a cache there are several parameters that can be set. The optimal settings can depend on the RAID level, the application scenario and the type of data medium. If the controller cache is enabled, the data temporarily stored in the cache should be safeguarded against loss in case of power failure. Suitable accessories are available for this purpose (e.g. a BBU or FBU).

For the purpose of easy and reliable handling of the settings for RAID controllers and hard disks it is advisable to use the RAID-Manager software “ServerView RAID” that is supplied for PRIMERGY servers. All the cache settings for controllers and hard disks can usually be made en bloc – specifically for the application – by using the pre-defined modi “Performance” or “Data Protection”. The “Performance” mode ensures the best possible performance settings for the majority of the application scenarios.

More information about the setting options of the controller cache is available in the white paper “RAID Controller Performance”.

http://docs.ts.fujitsu.com/dl.aspx?id=e2489893-cab7-44f6-bff2-7aeea97c5aef







Performance values

In general, disk-I/O performance of a RAID array depends on the type and number of hard disks, on the RAID level and on the RAID controller. If the limits of the system-specific interfaces are not exceeded, the statements on disk-I/O performance are therefore valid for all PRIMERGY systems. This is why all the performance statements of the document “RAID Controller Performance” also apply for the PRIMERGY BX924 S4 if the configurations measured there are also supported by this system.

The performance values of the PRIMERGY BX924 S4 are listed in table form below, specifically for different RAID levels, access types and block sizes. Substantially different configuration versions are dealt with separately.

The performance values in the following tables use the established measurement variables, as already mentioned in the subsection Benchmark description. Thus, transaction rate is specified for random accesses and data throughput for sequential accesses. To avoid any confusion among the measurement units the tables have been separated for the two access types.

The table cells contain the maximum achievable values. This has three implications: On the one hand hard disks with optimal performance were used (the components used are described in more detail in the subsection Benchmark environment). Furthermore, cache settings of controllers and hard disks, which are optimal for the respective access scenario and the RAID level, are used as a basis. And ultimately each value is the maximum value for the entire load intensity range (# of outstanding I/Os).

In order to also visualize the numerical values each table cell is highlighted with a horizontal bar, the length of which is proportional to the numerical value in the table cell. All bars shown in the same scale of length have the same color. In other words, a visual comparison only makes sense for table cells with the same colored bars.

Since the horizontal bars in the table cells depict the maximum achievable performance values, they are shown by the color getting lighter as you move from left to right. The light shade of color at the right end of the bar tells you that the value is a maximum value and can only be achieved under optimal prerequisites. The darker the shade becomes as you move to the left, the more frequently it will be possible to achieve the corresponding value in practice.




Random accesses (performance values in IO/s):

Sequential accesses (performance values in MB/s):

At full configuration with powerful hard disks (configured as RAID 1) the PRIMERGY BX924 S4 achieves a throughput of up to 684 MB/s for sequential load profiles and a transaction rate of up to 19002 IO/s for typical, random application scenarios.

RA

ID

Co

ntr

olle

r

Ha

rd d

isk

typ

e

Fo

rm f

ac

tor

#D

isk

s

Configuration

version

RA

ID le

ve

l

SS

Ds

ra

nd

om

8 k

B b

loc

ks

67

% r

ea

d

[IO

/s]

SS

Ds

ra

nd

om

64

kB

blo

ck

s

67

% r

ea

d

[IO

/s]

Patsburg A EP SATA SSD 2.5" 2 1 17760 3951

Patsburg B EP SAS SSD 2.5" 2 1 17736 3916

LSI2108 EP SAS SSD 2.5" 2 1 19002 4400

RA

ID

Co

ntr

olle

r

Ha

rd d

isk

typ

e

Fo

rm f

ac

tor

#D

isk

s

Configuration

version

RA

ID le

ve

l

SS

Ds

se

qu

en

tia

l

64

kB

blo

ck

s

10

0%

re

ad

[MB

/s]

SS

Ds

se

qu

en

tia

l

64

kB

blo

ck

s

10

0%

wri

te

[MB

/s]

Patsburg A EP SATA SSD 2.5" 2 1 506 175

Patsburg B EP SAS SSD 2.5" 2 1 504 180

LSI2108 EP SAS SSD 2.5" 2 1 684 176



OLTP-2


OLTP stands for Online Transaction Processing. The OLTP-2 benchmark is based on the typical application scenario of a database solution. In OLTP-2 database access is simulated and the number of transactions achieved per second (tps) determined as the unit of measurement for the system.

In contrast to benchmarks such as SPECint and TPC-E, which were standardized by independent bodies and for which adherence to the respective rules and regulations are monitored, OLTP-2 is an internal benchmark of Fujitsu. OLTP-2 is based on the well-known database benchmark TPC-E. OLTP-2 was designed in such a way that a wide range of configurations can be measured to present the scaling of a system with regard to the CPU and memory configuration.

Even if the two benchmarks OLTP-2 and TPC-E simulate similar application scenarios using the same load profiles, the results cannot be compared or even treated as equal, as the two benchmarks use different methods to simulate user load. OLTP-2 values are typically similar to TPC-E values. A direct comparison, or even referring to the OLTP-2 result as TPC-E, is not permitted, especially because there is no price-performance calculation.

Further information can be found in the document Benchmark Overview OLTP-2.


The measurement set-up is symbolically illustrated below:

All results were determined by way of example on a PRIMERGY RX300 S8.

Application Server

Tier A Tier B

Clients

Database Server Disk

subsystem


Driver

Network

Network

http://docs.ts.fujitsu.com/dl.aspx?id=e6f7a4c9-aff6-4598-b199-836053214d3f



Database Server (Tier B)

Hardware

Model PRIMERGY RX300 S8

Processor Xeon E5-2600 v2 processor series

Memory 1 processor: 8 × 32GB (1x32GB) 4Rx4 L DDR3-1600 LR ECC 2 processors: 16 × 32GB (1x32GB) 4Rx4 L DDR3-1600 LR ECC

Network interface 2 × onboard LAN 1 Gb/s

Disk subsystem RX300 S8: Onboard RAID Ctrl SAS 6G 5/6 1024MB (D3116C)

2 × 146 GB 15k rpm SAS Drive, RAID1 (OS),

6 × 300 GB 15k rpm SAS Drive, RAID10 (LOG)

5 × LSI MegaRAID SAS 9286CV-8e

5 × JX40: 16 × 200 GB SSD Drive each, RAID5 (data)

Software

BIOS Version R0.91.0

Operating system Microsoft Windows Server 2012 Standard

Database Microsoft SQL Server 2012 Enterprise SP1

Application Server (Tier A)

Hardware

Model 1 × PRIMERGY RX200 S8

Processor 2 × Xeon E5-2640 v2

Memory 32 GB, 1600 MHz registered ECC DDR3

Network interface 2 × onboard LAN 1 Gb/s 1 × Dual Port LAN 1Gb/s

Disk subsystem 1 × 250 GB 7.2k rpm SATA Drive

Software

Operating system Microsoft Windows Server 2012 Standard

Client

Hardware

Model 1 × PRIMERGY RX200 S5

Processor 2 × Xeon X5570

Memory 24 GB, 1333 MHz registered ECC DDR3

Network interface 2 × onboard LAN 1 Gb/s

Disk subsystem 1 × 73 GB 15k rpm SAS Drive

Software

Operating system Microsoft Windows Server 2008 R2 Standard

Benchmark OLTP-2 Software EGen version 1.12.0

Some components may not be available in all countries / sales regions.



Benchmark results

Database performance greatly depends on the configuration options with CPU, memory and on the connectivity of an adequate disk subsystem for the database. In the following scaling considerations for the processors we assume that both the memory and the disk subsystem has been adequately chosen and is not a bottleneck.

A guideline in the database environment for selecting main memory is that sufficient quantity is more important than the speed of the memory accesses. This why a configuration with a total memory of 512 GB was considered for the measurements with two processors and a configuration with a total memory of 256 GB for the measurements with one processor. Both memory configurations have memory access of 1600 MHz. Further information about memory performance can be found in the White Paper Memory performance of Xeon E5-2600 v2 (Ivy Bridge-EP)-based systems.

The following diagram shows the OLTP-2 transaction rates that can be achieved with one and two processors of the Intel Xeon E5-2600 v2 series.

266.95

342.54

634.96

655.53

696.33

758.76

977.61

850.54

1051.93

1217.04

894.87

1136.79

1240.47

1309.59

1376.26

1376.33

1442.64

482.41

619.01

1147.46

1184.63

1258.36

1371.18

1681.30

1462.76

1809.11

2093.08

1539.01

1955.07

2133.38

2252.25

2366.91

2367.02

2479.88

0 500 1000 1500 2000 2500

E5-2603 v2 - 4C

E5-2609 v2 - 4C

E5-2637 v2 - 4C, HT

E5-2620 v2 - 6C, HT

E5-2630Lv2 - 6C, HT

E5-2630 v2 - 6C, HT

E5-2643 v2 - 6C, HT

E5-2640 v2 - 8C, HT

E5-2650 v2 - 8C, HT

E5-2667 v2 - 8C, HT

E5-2650Lv2 - 10C, HT

E5-2660 v2 - 10C, HT

E5-2670 v2 - 10C, HT

E5-2680 v2 - 10C, HT

E5-2690 v2 - 10C, HT

E5-2695 v2 - 12C, HT

E5-2697 v2 - 12C, HT

OLTP-2 tps

2CPUs 512GB RAM

1CPU 256GB RAM

tps

bold: measured cursive: calculated

HT: Hyper-Threading

http://docs.ts.fujitsu.com/dl.aspx?id=a344b05e-2e9d-481b-8c9b-c6542defd839




It is evident that a wide performance range is covered by the variety of released processors. If you compare the OLTP-2 value of the processor with the lowest performance (Xeon E5-2603 v2) with the value of the processor with the highest performance (Xeon E5-2697 v2), the result is a 5.1-fold increase in performance.

Based on the results achieved the processors can be divided into different performance groups:

The start is made with Xeon E5-2603 v2 and E5-2609 v2 as processors with four cores, but without Hyper-Threading and without turbo mode. Due to its high clock frequency and the high QPI speed of 8.00 GT/s the throughput rates of the 6-core processors with the lowest frequencies (Xeon E5-2620 v2 and E5-2630Lv2) are almost achieved with the performance-optimized 4-core processor Xeon E5-2637 v2. However, the processors with 80 Watt and 60 Watt respectively also have distinctly lower power consumption than the Xeon E5-2637 v2 with 130 Watt.

The processors with six, eight, ten and twelve cores are all Hyper-Threading-capable and have with 7.20 GT/s or 8.00 GT/s a high QPI speed. Within a group of processors performance scales via the CPU clock frequency; leaps in performance for some processor types are a result of a higher QPI speed or a larger L3 cache per processor core.

At the upper end of the performance scale of the 6-core processors and also the 8-core processors are the E5-2643 v2 and E5-2667 v2 with their especially high frequency, which on the other hand achieve an OLTP performance that is above the processor with the lowest performance in the previous group.

The groups of processors with ten or twelve cores and a QPI speed of 8.00 GT/s (except the 10-core low-voltage CPU E5-2650Lv2) are to be found at the upper end of the performance scale. Due to the graduated CPU clock frequencies an OLTP performance of between 1539.01 tps (2 × Xeon E5-2650Lv2) and 2366.91 tps (2 × Xeon E5-2690 v2) is achieved within the group of the 10-core processors, while the 12-core processors with up to 2479.88 tps (2 × Xeon E5-2697 v2) provide the best performance.

If you compare the maximum achievable OLTP-2 values of the current system generation with the values that were achieved on the predecessor systems, the result is an increase of about 33%.

Current System TX300 S8 RX200 S8 RX300 S8 RX350 S8 BX924 S4

Predecessor System TX300 S7 RX200 S7 RX300 S7 RX350 S7 BX924 S3

0

500

1000

1500

2000

2500

+ ~33%

tps

Current System Predecessor System

Maximum OLTP-2 tps

Comparison of system generations

2 × E5-2690 512 GB

SQL 2012

2 × E5-2697 v2 512 GB

SQL 2012



vServCon


vServCon is a benchmark used by Fujitsu Technology Solutions to compare server configurations with hypervisor with regard to their suitability for server consolidation. This allows both the comparison of systems, processors and I/O technologies as well as the comparison of hypervisors, virtualization forms and additional drivers for virtual machines.

vServCon is not a new benchmark in the true sense of the word. It is more a framework that combines already established benchmarks (or in modified form) as workloads in order to reproduce the load of a consolidated and virtualized server environment. Three proven benchmarks are used which cover the application scenarios database, application server and web server.

Each of the three application scenarios is allocated to a dedicated virtual machine (VM). Add to these a fourth machine, the so-called idle VM. These four VMs make up a “tile”. Depending on the performance capability of the underlying server hardware, you may as part of a measurement also have to start several identical tiles in parallel in order to achieve a maximum performance score.

Each of the three vServCon application scenarios provides a specific benchmark result in the form of application-specific transaction rates for the respective VM. In order to derive a normalized score, the individual benchmark results for one tile are put in relation to the respective results of a reference system. The resulting relative performance values are then suitably weighted and finally added up for all VMs and tiles. The outcome is a score for this tile number.

Starting as a rule with one tile, this procedure is performed for an increasing number of tiles until no further significant increase in this vServCon score occurs. The final vServCon score is then the maximum of the vServCon scores for all tile numbers. This score thus reflects the maximum total throughput that can be achieved by running the mix defined in vServCon that consists of numerous VMs up to the possible full utilization of CPU resources. This is why the measurement environment for vServCon measurements is designed in such a way that only the CPU is the limiting factor and that no limitations occur as a result of other resources.

The progression of the vServCon scores for the tile numbers provides useful information about the scaling behavior of the “System under Test”.

Moreover, vServCon also documents the total CPU load of the host (VMs and all other CPU activities) and, if possible, electrical power consumption.

A detailed description of vServCon is in the document: Benchmark Overview vServCon.

Application scenario Benchmark No. of logical CPU cores Memory

Database Sysbench (adapted) 2 1.5 GB

Java application server SPECjbb (adapted, with 50% - 60% load) 2 2 GB

Web server WebBench 1 1.5 GB

System Under Test

… …

Tile n

Tile 3

Tile 2

Tile 1

Database VM

Web VM

Idle VM

Java VM

Database VM

Web VM

Idle VM

Java VM

Database VM

Web VM

Idle VM

Java VM

Database VM

Web VM

Idle VM

Java VM

http://docs.ts.fujitsu.com/dl.aspx?id=b953d1f3-6f98-4b93-95f5-8c8ba3db4e59




The measurement set-up is symbolically illustrated below:


Hardware

Processor Xeon E5-2600 v2 processor series

Memory 1 processor: 8 × 8GB (1x8GB) 2Rx4 L DDR3-1600 R ECC 2 processors: 16 × 8GB (1x8GB) 2Rx4 L DDR3-1600 R ECC

Network interface 1 × dual port 1GbE adapter 1 × dual port 10GbE server adapter

Disk subsystem 1 × dual-channel FC controller Emulex LPe12002

ETERNUS DX80 storage systems:

Each Tile: 50 GB LUN

Each LUN: RAID 0 with 2 × Seagate ST3300657SS-Disks (15 krpm)

Software

Operating system VMware ESX 5.1.0 U1 Build 1065491

Load generator (incl. Framework controller)

Hardware (Shared)

Enclosure PRIMERGY BX900

Hardware

Model 18 × PRIMERGY BX920 S1 server blades

Processor 2 × Xeon X5570

Memory 12 GB

Network interface 3 × 1 Gbit/s LAN

Software

Operating system Microsoft Windows Server 2003 R2 Enterprise with Hyper-V

Multiple 1Gb or 10Gb

networks

Load generators

Server Disk subsystem


Framework

controller



Load generator VM (per tile 3 load generator VMs on various server blades)

Hardware

Processor 1 × logical CPU

Memory 512 MB

Network interface 2 × 1 Gbit/s LAN

Software

Operating system Microsoft Windows Server 2003 R2 Enterprise Edition


Benchmark results

The PRIMERGY dual-socket systems dealt with here are based on Intel Xeon series E5-2600 v2 processors. The features of the processors are summarized in the section “Technical data”.

The available processors of these systems with their results can be seen in the following table.

Processor

RX

20

0 S

8

RX

30

0 S

8

RX

35

0 S

8

TX

30

0 S

8

BX

92

4 S

4

CX

25

0 S

2

CX

27

0 S

2

#Tiles Score

Xe

on

E5

-26

00

v2

Se

rie

s

4 Cores E5-2603 v2 2 3.29

E5-2609 v2 3 4.52

4 Cores, HT, TM E5-2637 v2 6 7.57

6 Cores HT, TM

E5-2620 v2 6 7.69

E5-2630Lv2 6 8.25

E5-2630 v2 6 9.03

E5-2643 v2 6 11.2

8 Cores HT, TM

E5-2640 v2 6 9.70

E5-2650 v2 7 12.3

E5-2667 v2 8 14.5

10 Cores HT, TM

E5-2650Lv2 6 9.99

E5-2660 v2 8 13.2

E5-2670 v2 9 14.6

E5-2680 v2 10 15.4

E5-2690 v2 10 16.3

12 Cores HT, TM

E5-2695 v2 11 16.3

E5-2697 v2 11 17.1

HT = Hyper-Threading, TM = Turbo Mode bold: measured, cursive: calculated

These PRIMERGY dual-socket systems are very suitable for application virtualization thanks to the progress made in processor technology. Compared with a system based on the previous processor generation an approximate 26% higher virtualization performance can be achieved (measured in vServCon score in their maximum configuration).



The first diagram compares the virtualization performance values that can be achieved with the processors reviewed here.

The relatively large performance differences between the processors can be explained by their features. The values scale on the basis of the number of cores, the size of the L3 cache and the CPU clock frequency and as a result of the features of Hyper-Threading and turbo mode, which are available in most processor types. Furthermore, the data transfer rate between processors (“QPI Speed”) also determines performance.

A low performance can be seen in the Xeon E5-2603 v2 and E5-2609 v2 processors, as they have to manage without Hyper-Threading (HT) and turbo mode (TM). In principle, these weakest processors are only to a limited extent suitable for the virtualization environment.

Within a group of processors with the same number of cores scaling can be seen via the CPU clock frequency.

As a matter of principle, the memory access speed also influences performance. A guideline in the virtualization environment for selecting main memory is that sufficient quantity is more important than the speed of the memory accesses. The vServCon scaling measurements presented here were all performed with a memory access speed – depending on the processor type – of at most 1600 MHz. More information about the topic “Memory Performance” and QPI architecture can be found in the White Paper Memory performance of Xeon E5-2600 v2 (Ivy Bridge-EP)-based systems.

E5-2

603 v

2

E5-2

609 v

2

E5-2

637v2

E5-2

620 v

2

E5-2

630Lv2

E5-2

630 v

2

E5-2

643 v

2

E5-2

640 v

2

E5-2

650 v

2

E5-2

667 v

2

E5-2

650Lv2

E5-2

660 v

2

E5-2

670 v

2

E5-2

680 v

2

E5-2

690 v

2

E5-2

695 v

2

E5-2

697 v

2

2 3 6 6 6 6 6 6 7 8 6 8 9 10 10 11 11

0

2

4

6

8

10

12

14

16

18

Fin

al vS

erv

Co

n S

co

re

Xeon E5-2600 v2 Processor Series #Tiles

12 core 8 core 10 core 6 core 4 core





Until now we have looked at the virtualization performance of a fully configured system. However, with a server with two sockets the question also arises as to how good performance scaling is from one to two processors. The better the scaling, the lower the overhead usually caused by the shared use of resources within a server. The scaling factor also depends on the application. If the server is used as a virtualization platform for server consolidation, the system scales with a factor of 1.86 or better. When operated with two processors, the system thus achieves a significantly better performance than with one processor, as is illustrated in the diagram opposite using the processor version Xeon E5-2697 v2 as an example. In this case, the scaling of the processor versions with a lower overall performance is somewhat better than for the CPU reviewed here with the largest number of cores.

The next diagram illustrates the virtualization performance for increasing numbers of VMs based on the Xeon E5-2667 v2 (8 core) and E5-2697 v2 (12 core) processors.

In addition to the increased number of physical cores, Hyper-Threading, which is supported by almost all Xeon processors of the E5-2600 v2 series, is an additional reason for the high number of VMs that can be operated. As is known, a physical processor core is consequently divided into two logical cores so that the number of cores available for the hypervisor is doubled. This standard feature thus generally increases the virtualization performance of a system.

The previous diagram examined the total performance of all application VMs of a host. However, studying the performance from an individual application VM viewpoint is also interesting. This information is in the previous diagram. For example, the total optimum is reached in the above Xeon E5-2667 v2 situation with 24 application VMs (eight tiles, not including the idle VMs); the low load case is represented by three application VMs (one tile, not including the idle VM). Remember: the vServCon score for one tile is an average value across the three application scenarios in vServCon. This average performance of one tile drops when changing from the low load case to the total optimum of the vServCon score - from 2.85 to 14.5/8=1.81, i.e. to 64%. The individual types of application VMs can react very differently in the high load situation. It is thus clear that in a specific situation the performance requirements of an individual application must be balanced against the overall requirements regarding the numbers of VMs on a virtualization host.

2.8

5

5.7

3

8.3

4

10.0

11.8

13.0

14.0

14.5

2.4

2

4.7

7

7.0

6

9.3

0

11.0

12.5

13.9

14.9

15.9

16.5

17.1

E5-2667 v2 E5-2697 v2

0

2

4

6

8

10

12

14

16

18

1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8 9 10 11

vS

erv

Con S

core

#Tiles

9.1

7@

6 t

iles

17.1

0@

11 tile

s

0

5

10

15

20

1 x E5-2697 v2 2 x E5-2697 v2

× 1.86

Fin

al vS

erv

Co

n S

co

re



The virtualization-relevant progress in processor technology since 2008 has an effect on the one hand on an individual VM and, on the other hand, on the possible maximum number of VMs up to CPU full utilization. The following comparison shows the proportions for both types of improvements. Five systems are compared: a system from 2008, a system from 2009, a system from 2011, a system from 2012 and a current system with the best processors each (see table opposite) for few VMs and for highest maximum performance.

2013 TX300 S8 RX200 S8 RX300 S8 RX350 S8 - - BX924 S4 CX250 S2 CX270 S2

2012 TX300 S7 RX200 S7 RX300 S7 RX350 S7 - - BX924 S3 CX250 S1 CX270 S1

2011 TX300 S6 RX200 S6 RX300 S6 TX300 S6 BX620 S6 BX922 S2 BX924 S2 - -

2009 TX300 S5 RX200 S5 RX300 S5 - BX620 S5 - - - -

2008 TX300 S4 RX200 S4 RX300 S4 - BX620 S4 - - - -

The clearest performance improvements arose from 2008 to 2009 with the introduction of the Xeon 5500 processor generation (e. g. via the feature “Extended Page Tables” (EPT)

1). One sees an increase of the

vServCon score by a factor of 1.28 with a few VMs (one tile).

With full utilization of the systems with VMs there was an increase by a factor of 2.07. The one reason was the performance increase that could be achieved for an individual VM (see score for a few VMs). The other reason was that more VMs were possible with total optimum (via Hyper-Threading). However, it can be seen that the optimum was “bought” with a triple number of VMs with a reduced performance of the individual VM.

Where exactly is the technology progress between 2009 and 2013?

The performance for an individual VM in low-load situations has only slightly increased for the processors compared here with the highest clock frequency per core. We must explicitly point out that the increased virtualization performance as seen in the score cannot be completely deemed as an improvement for one individual VM.

The decisive progress is in the higher number of physical cores and – associated with it – in the increased values of maximum performance (factor 1.58, 1.40 and 1.27 in the diagram).

Up to and including 2011 the best processor type of a processor generation had both the highest clock frequency and the highest number of cores. From 2012 there have been differently optimized processors on

1 EPT accelerates memory virtualization via hardware support for the mapping between host and guest memory addresses.

Best

Performance Few VMs

vServCon Score 1 Tile

Best Maximum

Performance

vServCon Score max.

2008 X5460 1.91 X5460 2.94

2009 X5570 2.45 X5570 6.08

2011 X5690 2.63 X5690 9.61

2012 E5-2643 2.73 E5-2690 13.5

2013 E5-2667 v2 2.85 E5-2697 v2 17.1

0

2

4

6

8

10

12

14

16

18

2008X5460

3.17 GHz4C

2009X5570

2.93 GHz4C

2011X5690

2.93 GHz6C

2012E5-26433.3 GHz

4C

2013E5-2667 v2

3.3 GHz8C

2008X5460

3.17 GHz4C

2009X5570

2.93 GHz4C

2011X5690

2.93 GHz6C

2012E5-26902.9 GHz

8C

2013E5-2697 v2

2.7 GHz12C

vS

erv

Co

n S

co

re

YYear CPUFreq.

#Cores

× 2.07

× 1.58

× 1.40

× 1.28

× 1.27Few VMs (1 Tile)

Virtualization relevant improvements

Score at optimum Tile count



offer: Versions with a high clock frequency per core for few cores and versions with a high number of cores, but with a lower clock frequency per core. The features of the processors are summarized in the section “Technical data”.

Performance increases in the virtualization environment since 2009 are mainly achieved by increased VM numbers due to the increased number of available logical or physical cores. However, since 2012 it has been possible - depending on the application scenario in the virtualization environment – to also select a CPU with an optimized clock frequency if a few or individual VMs require maximum computing power.



STREAM


STREAM is a synthetic benchmark that has been used for many years to determine memory throughput and which was developed by John McCalpin during his professorship at the University of Delaware. Today STREAM is supported at the University of Virginia, where the source code can be downloaded in either Fortran or C. STREAM continues to play an important role in the HPC environment in particular. It is for example an integral part of the HPC Challenge benchmark suite.

The benchmark is designed in such a way that it can be used both on PCs and on server systems. The unit of measurement of the benchmark is GB/s, i.e. the number of gigabytes that can be read and written per second.

STREAM measures the memory throughput for sequential accesses. These can generally be performed more efficiently than accesses that are randomly distributed on the memory, because the CPU caches are used for sequential access.

Before execution the source code is adapted to the environment to be measured. Therefore, the size of the data area must be at least four times larger than the total of all CPU caches so that these have as little influence as possible on the result. The OpenMP program library is used to enable selected parts of the program to be executed in parallel during the runtime of the benchmark, consequently achieving optimal load distribution to the available processor cores.

During implementation the defined data area, consisting of 8-byte elements, is successively copied to four types, and arithmetic calculations are also performed to some extent.

Type Execution Bytes per step Floating-point calculation per step

COPY a(i) = b(i) 16 0

SCALE a(i) = q × b(i) 16 1

SUM a(i) = b(i) + c(i) 24 1

TRIAD a(i) = b(i) + q × c(i) 24 2

The throughput is output in GB/s for each type of calculation. The differences between the various values are usually only minor on modern systems. In general, only the determined TRIAD value is used as a comparison.

The measured results primarily depend on the clock frequency of the memory modules; the CPUs influence the arithmetic calculations. The accuracy of the results is approximately 5%.

This chapter specifies throughputs on a basis of 10 (1 GB/s = 109 Byte/s).



Hardware





Software

BIOS settings Processors other than Xeon E5-2603 v2, E5-2609 v2: Hyper-Threading = Disabled


Operating system settings

echo never > /sys/kernel/mm/redhat_transparent_hugepage/enabled

Compiler Intel C Compiler 12.1

Benchmark Stream.c Version 5.9




Benchmark results

Processor Cores Processor Frequency

[Ghz]

Max. Memory Frequency

[MHz]

TRIAD

[GB/s]

2 × Xeon E5-2603 v2 4 1.80 1333 48.5

2 × Xeon E5-2609 v2 4 2.50 1333 59.1

2 × Xeon E5-2637 v2 4 3.50 1866 82.9

2 × Xeon E5-2620 v2 6 2.10 1600 78.9

2 × Xeon E5-2630Lv2 6 2.40 1600 80.7

2 × Xeon E5-2630 v2 6 2.60 1600 82.4

2 × Xeon E5-2643 v2 6 3.50 1866 96.9

2 × Xeon E5-2640 v2 8 2.00 1600 83.4

2 × Xeon E5-2650 v2 8 2.60 1866 96.9

2 × Xeon E5-2667 v2 8 3.30 1866 98.5

2 × Xeon E5-2650Lv2 10 1.70 1600 81.9

2 × Xeon E5-2660 v2 10 2.20 1866 95.9

2 × Xeon E5-2670 v2 10 2.50 1866 97.2

2 × Xeon E5-2680 v2 10 2.80 1866 97.7

2 × Xeon E5-2690 v2 10 3.00 1866 98.1

2 × Xeon E5-2695 v2 12 2.40 1866 101

2 × Xeon E5-2697 v2 12 2.70 1866 101

The results depend primarily on the maximum memory frequency. The processors with only 4 cores, which do not fully utilize their memory controller, are an exception. The smaller differences with processors with the same maximum memory frequency are a result in arithmetic calculation of the different processor frequencies.

The following diagram illustrates the throughput of the PRIMERGY BX924 S4 in comparison to its predecessor, the PRIMERGY BX924 S3, in their most performant configuration.

0

20

40

60

80

100

120

PRIMERGY BX924 S32 × Xeon E5-2667

PRIMERGY BX924 S42 × Xeon E5-2697 v2

81.8

101

GB/s

STREAM TRIAD: PRIMERGY BX924 S4 vs. PRIMERGY BX924 S3



LINPACK


LINPACK was developed in the 1970s by Jack Dongarra and some other people to show the performance of supercomputers. The benchmark consists of a collection of library functions for the analysis and solution of linear system of equations. A description can be found in the document http://www.netlib.org/utk/people/JackDongarra/PAPERS/hplpaper.pdf.

LINPACK can be used to measure the speed of computers when solving a linear equation system. For this purpose, an n × n matrix is set up and filled with random numbers between -2 and +2. The calculation is then performed via LU decomposition with partial pivoting.

A memory of 8n² bytes is required for the matrix. In case of an n × n matrix the number of arithmetic operations required for the solution is

2/3n

3 + 2n

2. Thus, the choice of n determines the duration of the

measurement: a doubling of n results in an approximately eight-fold increase in the duration of the measurement. The size of n also has an influence on the measurement result itself: as n increases, the measured value asymptotically approaches a limit. The size of the matrix is therefore usually adapted to the amount of memory available. Furthermore, the memory bandwidth of the system only plays a minor role for the measurement result, but a role that cannot be fully ignored. The processor performance is the decisive factor for the measurement result. Since the algorithm used permits parallel processing, in particular the number of processors used and their processor cores are - in addition to the clock rate - of outstanding significance.

LINPACK is used to measure how many floating point operations were carried out per second. The result is referred to as Rmax and specified in GFlops (Giga Floating Point Operations per Second).

An upper limit, referred to as Rpeak, for the speed of a computer can be calculated from the maximum number of floating point operations that its processor cores could theoretically carry out in one clock cycle:

Rpeak = Maximum number of floating point operations per clock cycle × Number of processor cores of the computer × Maximum processor frequency[GHz]

LINPACK is classed as one of the leading benchmarks in the field of high performance computing (HPC). LINPACK is one of the seven benchmarks currently included in the HPC Challenge benchmark suite, which takes other performance aspects in the HPC environment into account.

Manufacturer-independent publication of LINPACK results is possible at http://www.top500.org/. The use of a LINPACK version based on HPL is prerequisite for this (see: http://www.netlib.org/benchmark/hpl).

Intel offers a highly optimized LINPACK version (shared memory version) for individual systems with Intel processors. Parallel processes communicate here via "shared memory", i.e. jointly used memory. Another version provided by Intel is based on HPL (High Performance Linpack). Intercommunication of the LINPACK processes here takes place via OpenMP and MPI (Message Passing Interface). This enables communication between the parallel processes - also from one computer to another. Both versions can be downloaded from http://software.intel.com/en-us/articles/intel-math-kernel-library-linpack-download/.

Manufacturer-specific LINPACK versions also come into play when graphics cards for General Purpose Computation on Graphics Processing Unit (GPGPU) are used. These are based on HPL and include extensions which are needed for communication with the graphics cards.

http://www.netlib.org/utk/people/JackDongarra/PAPERS/hplpaper.pdf

http://www.top500.org/

http://www.netlib.org/benchmark/hpl

http://software.intel.com/en-us/articles/intel-math-kernel-library-linpack-download/





Hardware





Software

BIOS settings All processors apart from Xeon E5-2603 v2, E5-2609 v2: Hyper Threading = Disabled

All processors apart from Xeon E5-2603 v2, E5-2609 v2: Turbo Mode = Enabled (default) = Disabled


Benchmark HPL version: Intel Optimized MP LINPACK Benchmark for Clusters 11.0 Update 5


Benchmark results

Pro

ce

ss

or

Co

res

Pro

ce

ss

or

freq

ue

nc

y [

Gh

z]

Ma

xim

um

tu

rbo

fre

qu

en

cy

at

full

lo

ad

[G

hz]

Nu

mb

er

of

pro

ce

sso

rs

Without Turbo Mode With Turbo Mode

Rpeak [GFlops]

Rmax [GFlops]

Rpeak [GFlops]

Rmax [GFlops]

Xeon E5-2603 v2 4 1.80 n/a 2 115 110

Xeon E5-2609 v2 4 2.50 n/a 2 160 152

Xeon E5-2637 v2 4 3.50 3.60 2 224 213 230 220

Xeon E5-2620 v2 6 2.10 2.40 2 202 192 230 219

Xeon E5-2630Lv2 6 2.40 2.60 2 230 219 250 238

Xeon E5-2630 v2 6 2.60 2.90 2 250 238 278 265

Xeon E5-2643 v2 6 3.50 3.60 2 336 320 346 329

Xeon E5-2640 v2 8 2.00 2.30 2 256 244 294 280

Xeon E5-2650 v2 8 2.60 3.00 2 333 317 384 363

Xeon E5-2667 v2 8 3.30 3.60 2 422 402 461 420

Xeon E5-2650Lv2 10 1.70 1.90 2 272 259 304 289

Xeon E5-2660 v2 10 2.20 2.60 2 352 336 416 397

Xeon E5-2670 v2 10 2.50 2.90 2 400 381 464 441

Xeon E5-2680 v2 10 2.80 3.10 2 448 426 496 463

Xeon E5-2690 v2 10 3.00 3.30 2 480 456 528 486

Xeon E5-2695 v2 12 2.40 2.80 2 461 437 538 492

Xeon E5-2697 v2 12 2.70 3.00 2 518 492 576 547

Rmax = Measurement result

SPECcpu2006: floating-point performance PRIMERGY TX200 S6 vs. predecessor



Rpeak = Maximum number of floating point operations per clock cycle × Number of processor cores of the computer × Maximum processor frequency[GHz]

The following applies for processors without Turbo mode and for those with Turbo mode disabled:

Maximum processor frequency[GHz] = Nominal processor frequency[GHz]

Processors with Turbo mode enabled are not limited by the nominal processor frequency and therefore do not provide a constant processor frequency. Instead the actual processor frequency swings - depending on temperature and power consumption - between the nominal processor frequency and maximum turbo frequency at full load. Therefore, the following applies for these processors:

Maximum processor frequency[GHz] = Maximum turbo frequency at full load[GHz]

System comparison

The following diagram illustrates the throughput of the PRIMERGY BX924 S4 in comparison to its predecessor, the PRIMERGY BX924 S3, in their most performant configuration.

0

100

200

300

400

500

600

PRIMERGY BX924 S32 × Xeon E5-2690

PRIMERGY BX924 S42 × Xeon E5-2697 v2

351

547

GFlops

LINPACK: PRIMERGY BX924 S4 vs. PRIMERGY BX924 S3



Literature

PRIMERGY Systems

http://primergy.com/

PRIMERGY BX924 S4

This White Paper: http://docs.ts.fujitsu.com/dl.aspx?id=fd0b3cde-e195-41ae-80bb-4465333ecbe5 http://docs.ts.fujitsu.com/dl.aspx?id=2eeb8971-c835-444f-8d12-1e82ce80363b http://docs.ts.fujitsu.com/dl.aspx?id=30785baa-ed79-4b55-b3e8-10f55adef98c

Data sheet http://docs.ts.fujitsu.com/dl.aspx?id=94ad5e5b-e23c-445a-8bc8-792313ebd7e4

BIOS optimizations for Xeon E5-2600 v2 based systems http://docs.ts.fujitsu.com/dl.aspx?id=84dc1adf-adb8-419f-8d08-b226eb077e46

Memory performance of Xeon E5-2600 v2 (Ivy Bridge-EP)-based systems http://docs.ts.fujitsu.com/dl.aspx?id=a344b05e-2e9d-481b-8c9b-c6542defd839

PRIMERGY Performance

http://www.fujitsu.com/fts/products/computing/servers/primergy/benchmarks/

Disk I/O

Basics of Disk I/O Performance http://docs.ts.fujitsu.com/dl.aspx?id=65781a00-556f-4a98-90a7-7022feacc602

Single Disk Performance http://docs.ts.fujitsu.com/dl.aspx?id=0e30cb69-44db-4cd5-92a7-d38bacec6a99

512e HDDs: Technology, Performance, Configurations http://docs.ts.fujitsu.com/dl.aspx?id=f5550c48-d4db-47f6-ab9d-ce135eaacf81

RAID Controller Performance http://docs.ts.fujitsu.com/dl.aspx?id=e2489893-cab7-44f6-bff2-7aeea97c5aef

Information about Iometer http://www.iometer.org

LINPACK

The LINPACK Benchmark: Past, Present, and Future http://www.netlib.org/utk/people/JackDongarra/PAPERS/hplpaper.pdf.

TOP500 http://www.top500.org/

HPL - A Portable Implementation of the High-Performance Linpack Benchmark for Distributed-Memory Computers http://www.netlib.org/benchmark/hpl

Intel Math Kernel Library – LINPACK Download http://software.intel.com/en-us/articles/intel-math-kernel-library-linpack-download/

OLTP-2

Benchmark Overview OLTP-2 http://docs.ts.fujitsu.com/dl.aspx?id=e6f7a4c9-aff6-4598-b199-836053214d3f

SPECcpu2006

http://www.spec.org/osg/cpu2006

Benchmark overview SPECcpu2006 http://docs.ts.fujitsu.com/dl.aspx?id=1a427c16-12bf-41b0-9ca3-4cc360ef14ce

STREAM

http://www.cs.virginia.edu/stream/

http://primergy.com/

http://docs.ts.fujitsu.com/dl.aspx?id=fd0b3cde-e195-41ae-80bb-4465333ecbe5

http://docs.ts.fujitsu.com/dl.aspx?id=2eeb8971-c835-444f-8d12-1e82ce80363b

http://docs.ts.fujitsu.com/dl.aspx?id=30785baa-ed79-4b55-b3e8-10f55adef98c

http://docs.ts.fujitsu.com/dl.aspx?id=94ad5e5b-e23c-445a-8bc8-792313ebd7e4

http://docs.ts.fujitsu.com/dl.aspx?id=84dc1adf-adb8-419f-8d08-b226eb077e46


http://www.fujitsu.com/fts/products/computing/servers/primergy/benchmarks/

http://docs.ts.fujitsu.com/dl.aspx?id=65781a00-556f-4a98-90a7-7022feacc602




http://www.iometer.org/

http://www.netlib.org/utk/people/JackDongarra/PAPERS/hplpaper.pdf

http://www.top500.org/

http://www.netlib.org/benchmark/hpl

http://software.intel.com/en-us/articles/intel-math-kernel-library-linpack-download/

http://docs.ts.fujitsu.com/dl.aspx?id=e6f7a4c9-aff6-4598-b199-836053214d3f

http://www.spec.org/osg/cpu2006

http://docs.ts.fujitsu.com/dl.aspx?id=1a427c16-12bf-41b0-9ca3-4cc360ef14ce

http://www.cs.virginia.edu/stream/



vServCon

Benchmark Overview vServCon http://docs.ts.fujitsu.com/dl.aspx?id=b953d1f3-6f98-4b93-95f5-8c8ba3db4e59

Contact

FUJITSU

Website: http://www.fujitsu.com/

PRIMERGY Product Marketing

mailto:[email protected]

PRIMERGY Performance and Benchmarks


© Copyright 2013 Fujitsu Technology Solutions. Fujitsu and the Fujitsu logo are trademarks or registered trademarks of Fujitsu Limited in Japan and other countries. Other company, product and service names may be trademarks or registered trademarks of their respective owners. Technical data subject to modification and delivery subject to availability. Any liability that the data and illustrations are complete, actual or correct is excluded. Designations may be trademarks and/or copyrights of the respective manufacturer, the use of which by third parties for their own purposes may infringe the rights of such owner. For further information see http://www.fujitsu.com/fts/resources/navigation/terms-of-use.html

2013-11-14 WW EN

http://docs.ts.fujitsu.com/dl.aspx?id=b953d1f3-6f98-4b93-95f5-8c8ba3db4e59

http://www.fujitsu.com/



http://www.fujitsu.com/fts/resources/navigation/terms-of-use.html

Documents

Performance Report PRIMERGY BX924 S4 - Fujitsu · PDF fileWhite Paper Performance Report PRIMERGY BX924 S4 ... or RAID 1 for up to 2 × 2.5˝ SATA HDDs, ... time which is required