Sun Blade 6000 Modular System

The new World Record 4-thread performance on the SPEC OMPM2001 benchmark, which is used to compare the performance of shared memory servers executing compute-intensive scientific applications, places the Sun Blade X6220 server module ahead of competition in the High Performance Computing (HPC) market.

Sun's x64 systems have been dominant on HPC compute-intensive workloads using superior compiler software that delivers the best performance by providing optimizations tailored to a specific hardware platform or microprocessor architecture.

The combination of the Sun Studio 12 software (that consists of performance tools, Integrated Development Environment (IDE) and C, C++, and Fortran compilers) and Solaris 10 Operating system (OS), running on two Dual-Core AMD Opteron Model 2222 processors (2 chips, 4 cores) resulted in SPECompM2001 score of 13868 on the medium size problem set of the benchmark.

• Press Release
• Benchmark Results

The Sun Blade X6250 server module, equipped with two dual-core Intel Xeon X5260 CPUs and running 64-bit SuSE Linux (SLES10) OS, posted record-breaking scores on a Standard suite of the ANSYS 11 benchmark, improving upon previously posted results by 25% on average.

ANSYS 11 software is a comprehensive multidisciplinary tool combining structural, thermal, fluid, acoustic and electromagnetic simulation capabilities in a single engineering software solution that is applicable to a broad range of industries and applications. To provide customers with a basis for comparison across multiple architectures, operating systems and interconnects, ANSYS has developed a benchmark that has two suites: Standard and Distributed. These suites consist of multiple workloads and cover a representative set of structural analysis solvers and analysis types for stand-alone SMP systems, as well as clustered configurations.

The Standard suite of the benchmark was run on a Sun Blade X6250 server module at different core levels: 1, 2, and the maximum of 4. It is worth noting that the 7 test cases from ANSYS Standard test suite do not scale very well past 8 cores and the memory requirements for these test cases are less than 3GB, making Sun's dual-socket server module with dual or quad core processors the perfect building block for custom engineering solutions. Additionally, as the below table suggests, the same system has also produced stellar results when outfitted with other recent Intel Xeon processors, thus allowing customers more choice and flexibility in their purchasing decisions.

• Solution Brief
• Benchmark Results

Four Sun Blade X6250 server modules, each equipped with two dual-core Intel Xeon X5260 CPUs and running 64-bit SuSE Linux (SLES10) OS, have formed an eight CPU cluster that demonstrated world record performance on ABAQUS/Explicit, a finite element analysis package that simulates brief transient dynamic events (a.k.a. “crashes” and high velocity impacts).

The ABAQUS/Explicit benchmark suite includes a combination of six high-speed dynamic impact events like projectile penetration or a car crash, and quasi-static events with complicated contact conditions like forming a part out of a blank metal sheet. The benchmark is intended to provide an estimate of the performance that can be expected when running representative ABAQUS analysis jobs on different computer platforms.

In all six cases the Sun Blade X6250-based cluster using InfiniBand PCIe ExpressModules interconnected via a Voltaire Grid Director ISR 9096 switch, performs better than competing clusters at the same core count. Additionally, as the below table suggests, the same cluster has also produced stellar results when outfitted with other recent Intel Xeon processors, thus allowing customers more choice and flexibility in their purchasing decisions.

• Solution Brief
• Benchmark Results

Four Sun Blade X6250 server modules, each equipped with two dual-core Intel Xeon X5260 CPUs and running 64-bit SuSE Linux (SLES10) OS, have formed an eight CPU cluster that outperforms all results published by other hardware vendors on EXA PowerFLOW benchmark at comparable cluster sizes.

EXA's PowerFLOW CFD (Computational Fluid Dynamics) application is an MCAE application used by the major engineering companies in a variety of disciplines such as aerospace, automotive and chemical. The EXA PowerFLOW benchmark consists of two compute and memory intensive suites with small and large data sets that model the air flow over a body of a car.

In both cases, the cost-effective solution based on high-performance Sun Blade X6250 servers that were using InfiniBand PCIe ExpressModules interconnected via a Voltaire Grid Director ISR 9096 switch, performs better than competing clusters at the same core count. Additionally, as the below table suggests, the same cluster has also produced stellar results when outfitted with other recent Intel Xeon processors, thus allowing customers more choice and flexibility in their purchasing decisions.

• Solution Brief

Four Sun Blade X6250 server modules, each equipped with two dual-core Intel Xeon X5260 CPUs and running 64-bit SuSE Linux (SLES10) OS, have formed an eight CPU cluster that demonstrated world record performance on one of the most popular high velocity impact (a.k.a. "crash") analysis applications – LS-DYNA. Sun's cluster surpassed all published results, ranging from one to 8 CPUs, that were posted on topcrunch.org web site, while maintaining superb scalability with the increasing size of the cluster.

The TopCrunch benchmark was created with the objective of tracking the aggregate performance trends of high performance computer systems. Instead of relying upon synthetic workloads, it was built upon an LS-DYNA application that reflects the types of calculations performed in the mechanical and aerospace communities and is used for crash simulation and analysis. The benchmark consists of two standardized problems: "Neon_Refined" and a much larger "3 Vehicle Collision". These problems were chosen to reflect current engineering practices in the real world and are not intended to be optimal, because engineers rarely have time to optimize their analysis in real life.

The Sun Blade X6250-based cluster using InfiniBand PCIe ExpressModules interconnected via Voltaire Grid Director ISR 9096 switch, performs better than competing clusters at the same core count. Specifically, the Sun Blade X6250-based cluster beats a similarly equipped IBM BladeCenter HS21 XM-based cluster on "3 Vehicle Collision" problem by an average of 30%. Moreover, as evidenced in the table below, Sun's solution maintained 85% scalability efficiency despite the non-linear scalability nature inherent in these sub-optimal engineering codes.

• Solution Brief
• Benchmark Results

Four Sun Blade X6250 server modules, each equipped with two dual-core Intel Xeon X5260 CPUs and running 64-bit SuSE Linux (SLES10) OS, have formed an eight CPU cluster that demonstrated outstanding performance on one of the most popular MCAE applications - FLUENT. The software solves fluid flow problems, is based on a numerical technique called computational fluid dynamics (CFD) and is heavily used in the automotive, aerospace and consumer products industries.

FLUENT offers two benchmarking suites that have distinctly different sets of problems incorporated into them. The "Standard" FLUENT v6.3 benchmark test suite consists of nine industrial CFD applications that are selected to demonstrate the performance of FLUENT on a variety of hardware platforms. The problems are divided into three varying in size classes: Small, Medium and Large.

The "New" FLUENT v6.3 benchmark test suite consists of seven larger models (both in memory requirements and mesh/model size) that are more suited for multi-node clustered environments and representative of modern engineering CFD clusters. Vendors benchmark their systems with the principal objective of providing comparative performance information for FLUENT software that, among other things, depends on compilers, optimization, interconnect, and the performance characteristics of the hardware.

According to the results that are listed on the above web sites, Sun's solution, using InfiniBand PCIe ExpressModules interconnected via Voltaire Grid Director ISR 9096 switch, performs better than competing clusters at the same core count. Specifically, the Sun Blade X6250-based cluster beats the similarly equipped IBM System x3550-based cluster on the "Standard" suite by as much as 64%, and an Intel quad-core whitebox-based cluster on the "New" suite by as much as 85%. Additionally, the average scalability efficiency of Sun's offering stayed above 90%, as the cluster grew in size up to the maximum of 8 cores, while IBM's average scalability was closer to 80%.

• Solution Brief
• Benchmark Results - "Standard" suite
• Benchmark Results - "New" suite

Operating system (OS) and compiler software often help to deliver the best performance by providing optimizations tailored to a specific hardware platform or microprocessor architecture.

Solaris 10 OS and Sun Studio 12 complier software, use many innovative optimization techniques targeted at x86 platforms. Sun's software running on the top of the Intel-based system architecture, propelled the Sun Blade 6250 server module to the top with the best 4-core result on integer-intensive throughput suite of the SPECcpu2006 benchmark. Equipped with two Dual-Core Intel Xeon 5160 processors, the Sun Blade X6250 server module, posted the SPECint_rate2006 result of 65.0.

The SPEC CPU2006 Benchmark exercises a computer's processor, memory architecture, and compilers on a variety of real-world compute intensive workloads and consists of two benchmark suites. One suite measures and compares compute-intensive integer performance and the other measures and compares floating point performance. For each of these suites, two metrics are collected, one measures how fast the computer completes a single task and the other measures how many tasks a computer can accomplish in a certain amount of time (the throughput).

• Press Release
• Benchmark Results

Sun's new Intel-based system, the Sun Blade X6250 server module enters the market with an attention-grabbing SPECint2006 result of 21.0.

The SPEC CPU2006 Benchmark exercises a computer's processor, memory architecture, and compilers on a variety of real-world compute intensive workloads and consists of two benchmark suites. One suite measures and compares compute-intensive integer performance and the other measures and compares floating point performance. For each of these suites, two metrics are collected, one measures how fast the computer completes a single task and the other measures how many tasks a computer can accomplish in a certain amount of time.

Operating system (OS) and compiler software often help to deliver the best performance by providing optimizations tailored to a specific hardware platform or microprocessor architecture. The advanced features of freely available Sun Studio 12 complier software allowed Sun Blade 6250 to successfully claim the top spot on a single integer task of the SPECcpu2006 benchmark. Equipped with the Dual-Core Intel Xeon 5160 processor, the Sun Blade X6250 server module uses many innovative optimization techniques targeted at x86 platforms that are inherent in Solaris 10, the most advanced OS on the planet.

• Press Release
• Benchmark Results

The Sun Blade 6000 chassis, outfitted with ten Sun Blade T6320 server modules, achieves unprecedented levels of performance by delivering the required computing capacity in an innovative way. By combining the integrated design advantages of the modular blade architecture with advanced features of Solaris 10 Operating System and Sun Studio 12 software, Sun's modular system claims the top spot among all blades on Integer and Floating Point throughput suites of SPEC CPU2006 benchmark.

The Sun Blade 6000 modular blade system with ten CoolThreads server modules, powered by UltraSPARC T2 processor at 1.4GHz, produce record-setting SPECint_rate2006 score of 838 and SPECfp_rate2006 result of 571.

Benchmark Description

The SPEC CPU2006 benchmark provides a broader variety of workloads and better real-world applicability of the results. It consists of two benchmark suites in which one suite measures and compares compute-intensive integer performance and the other measures and compares floating point performance. The benchmark exercises a computer's processor, memory architecture, and compilers on a variety of compute intensive workloads, including protein sequencing, MPEG-4 decoding, XML processing, fluid dynamics, structural mechanics, and speech recognition.

Benchmark Outcome

• Surpassing the competing dual processor systems from HP and IBM, the Sun Blade 6000 modular system with ten Sun Blade T6320 server modules each equipped with 1.4GHz UltraSPARC T2 processor posted World Records for all blade systems on both integer and floating-point throughput suites of the SPEC CPU2006 benchmark.
• The Sun Blade 6000 modular system fully populated with ten Sun Blade T6320 server modules delivered SPECint_rate_base2006 score of 752.
• The Sun Blade 6000 modular system powered by 10 Sun UltraSPARC T2 1.4 GHz processors provides 73% more integer throughput than the IBM p 570 equipped with 8 POWER6 4.7 GHz processors, as measured by SPECint_rate2006.
• The Sun Blade 6000 modular system powered by 10 Sun UltraSPARC T2 1.4 GHz processors provides 79% more integer throughput than the IBM p 570 equipped with 8 POWER6 4.7 GHz processors, as measured by SPECint_rate_base2006.
• The Sun Blade 6000 modular system fully populated with 10 Sun Blade T6320 server modules delivered a blade system world record SPECfp_rate2006 score of 571.
• In addition, the Sun Blade 6000 modular system fully populated with ten Sun Blade T6320 server modules delivered blade system World Record scores of 752 on SPECint_rate_base2006 suite and 571 on SPECfp_rate_base2006 suite of the benchmark. The “base” scores are produced using less optimized benchmark binaries that were compiled under stricter guidelines. Although "base" scores may not reflect the the ultimate performance of the system, some end users, who may prefer to spend less time tuning their compiler optimizations, may find them more appealing.
• The Sun Blade 6000 modular system powered by 10 Sun UltraSPARC T2 1.4 GHz processors provides 55% more floating-point throughput than the IBM p 570 equipped with 8 POWER6 4.7 GHz processors, as measured by SPECfp_rate_base2006.
• Sun is the only vendor to demonstrate the true capabilities of a fully populated modular blade chassis with an industry-standard benchmark such as SPEC CPU2006.
• The Sun Blade 6000 modular system with ten Sun Blade T6320 server modules demonstrates that innovative CMT processors can scale horizontally while providing unprecedented SMP-like computing capacity previously unavailable in this segment of the market.

SPEC, SPECint, SPECfp registered trademarks of Standard Performance Evaluation Corporation. Sun Blade T6320 (UltraSPARC T2, 10 chips, 80 cores), 838 SPECint_rate2006, 752 SPECint_rate_base2006. IBM p 570 (POWER6, 8 chips, 16 cores), 484 SPECint_rate2006, 420 SPECint_rate_base2006. HP Superdome (IA2 9050, 32 chips, 64 cores), 824 SPECint_rate2006, 770 SPECint_rate_base2006

The Sun Blade T6320 server module, based on the innovative UltraSPARC T2 processor, achieved the highest ever single-chip integer throughput result on the SPEC CPU2006 benchmark. This new record-breaking result outperforms, on a per-processor basis, IBM systems with the fastest 4.7GHz POWER6 CPU by over 40% and beats the best published result for Intel Xeon-based systems by 17%.

Benchmark Description

SPEC CPU2006 is a CPU-intensive benchmark suite, stressing a system's processor, memory subsystem and compiler. SPEC designed CPU2006 to provide a comparative measure of compute-intensive performance across the widest practical range of hardware.

The gccfs compiler combines the power of Sun's optimization tools, developed for SPARC architectures, with the simplicity of the popular gcc coding conventions and delivers the performance levels that were not previously possible without extensive and time-consuming code changes.

Benchmark Outcome (9)

• The Sun Blade T6320 server module, equipped with the 1.4GHz UltraSPARC T2 processor, achieved single-chip world record integer throughput performance on the SPEC CPU2006 benchmark.
• On a per-processor basis using integer-intensive tests, the Sun Blade T6320 server module delivered 40% higher performance than the 4.7GHz dual core POWER6-based IBM System p570 server and 17% more performance than the 3.16 GHz quad-core Intel Xeon-based HP Proliant DL360 G5 server.
• These results demonstrate that the UltraSPARC T2 processor provides more integer computing capacity than any other chip on the market.
• The Operating System (OS) and compiler software often help to deliver the best performance by providing optimizations tailored to a specific hardware platform or microprocessor architecture. The advanced optimization techniques used in freely available Sun Studio 12 compiler software in combination with the popular gcc coding conventions, allowed the Sun Blade T6320 server module to successfully claim the top spot on the industry-standard SPECint_rate2006 benchmark suite.

(9) SPEC, SPECint registered trademarks of Standard Performance Evaluation Corporation. Sun Blade T6320 (1xUltraSPARC T2 1.4GHz) (1 chip, 8 cores, 64 threads) 85.5 SPECint_rate2006. Competitive results from www.spec.org as of February 19, 2008. IBM System p570 (4.7GHz POWER6 processor, 1 chip, 2 cores, 4 threads) 60.9 SPECint_rate2006. HP Proliant DL360 G5 (3.16 GHz, Intel Xeon processor X5460,1 chip, 4 cores, 4 threads) 73.0 SPECint_rate2006.

The Sun Blade 6000 modular system, configured with 10 Sun Blade T6300 server modules (10 chips total) delivered the highest ever result of 8253.21 SPECjAppServer2004 JOPS@Standard on the industry-standard SPECjAppServer2004 benchmark.

SPECjAppServer2004 (Java Application Server) is a multi-tier benchmark for measuring the performance of Java 2 Enterprise Edition (J2EE) technology-based application servers. SPECjAppServer2004 is designed to exercise all major J2EE technologies implemented by compliant application servers, as well as all parts of the underlying infrastructure that make up the application environment, including hardware, JVM software, database software, JDBC drivers, and the system network.

Sun's World Record-setting multi-tier solution, which can support over 50,000 concurrent users accessing J2EE applications, demonstrates the "best of breed" approach to solving customer's problems and emphasizes the enterprise capabilities of Sun's new blade server in the areas of business transaction processing and server side Java.

At the application tier, the Sun Blade 6000 modular system, wrapped into a neat 10-rack units (RU) package, enables customers to overcome the traditional power and space limitations in their data centers by building on the unparalleled efficiency of Sun Blade T6300 server modules, equipped with the eco-friendly 1.4 GHz UltraSPARC T1 processor. Combining the Solaris 10 Operating System (OS) and BEA WebLogic 10.0 application server software allows organizations to process more application-tier transactions and serve more users than ever before, all while reducing software licensing costs (as a result of system's single socket design) and increasing the system's utilization by partitioning the hardware with Solaris Containers technology that enables multiple applications to each run in their own isolated execution environment. Operational costs are also reduced through lower power, cooling and space consumption, enabling organizations to reduce their carbon footprint.

At the database tier, the Sun Fire E6900 server outfitted with 24 1.95GHz UltraSPARC IV+ processors (24 chips), ran IBM DB2 9.1 Enterprise Edition database software on top of the award-winning Solaris 10 OS.

The following table summarizes the advantage of Sun's solution against competing systems using metrics that are often employed in TCO studies: power consumption, space utilization and performance.

* HP rx6660 power calculated as 70% of maximum input power reported on HP's web site as of 07/25/07.
** IBM HS20 power calculated as 70% of maximum input power using xSeries Power Configurator - version 4.4.2.28. Sun Blade 6000 power was measured during the actual benchmark test.

• Press Release
• Benchmark Results

The SPECjbb2005 benchmark emulates the design of a real-world server-side Java application typical for a wholesale company with multiple warehouses that serve a number of districts. The increased number of warehouses during the benchmark run directly translates into the number of Java threads that are running on the system, and thus can be used to stress the scalability of the system's Java Virtual Machine (JVM), processors and memory.

The Sun Blade T6300 server module, equipped with the 1.4 GHz UltraSPARC T1 processor, is the fastest single processor system that was running multiple JVM instances with scores of 96,523 SPECjbb2005 bops and 24,131 SPECjbb2005 bops/JVM. Moreover, the Sun Blade T6300 server module is also the most scalable single socket system on SPECjbb2005 benchmark with the highest number of supported warehouses. Unique features of eco-friendly UltraSPARC T1 processors in combination with reliability of Solaris 10 operating system and scalability of Java HotSpot Server Virtual Machine ascertain the enterprise capabilities of Sun's new blade in the areas of business transaction processing and server side Java.

• Press Release
• Benchmark Results