Sun Fire / Sun SPARC Enterprise T1000 Server

Appium announces key performance results on Sun Fire Servers with CoolThreads technology.

For telecoms operators worldwide these results translate into the ability to offer high capacity revenue generating services to its customers. They also offer operators significant capital and operational cost savings due to performance and power optimization alongside reduced data centre rack space and
support costs

Review the test results and FAQs.

• Press Release

The Sun Fire T1000 server running Solaris 10 OS beat both 2 & 4-way Intel Xeon / Linux servers from Dell and IBM a head-to-head test of web server performance. These results were achieved in half the space of the Xeon servers, and with at least 3.5X higher performance per watt. The Sun Fire T1000 server also outperformed a 4-Way IBM Power 5 server running AIX, while consuming 4X less space, and delivering an incredible 22X higher server efficiency (SWaP) rating.

These results confirm the Sun Fire T1000 server as the most efficient and cost-effective platform on the planet for deployment of secure web servers.

Benchmark Description

SPECweb2005 is the industry standard benchmark for evaluating Web Server performance, developed by SPEC. The benchmark simulates multiple user sessions accessing a Web Server and generating static and dynamic HTTP requests. The major features of SPECweb2005 are:

• Measures simultaneous user sessions
• Dynamic content: currently PHP and JSP implementations
• Page images requested using 2 parallel HTTP connections
• Multiple, standardized workloads: Banking (HTTPS), E-commerce (HTTP and HTTPS), and Support (HTTP)
• Simulates browser caching effects
• File accesses more accurately simulate today's disk access patterns

Benchmark Outcome (4)

• The Sun Fire T1000 obtained 10466 SPECweb2005 with 1 UltraSPARC T1 processor and 16 GB memory running Solaris 10 6/06 with Sun Java[TM] System Web Server 6.1 SP5 64-bit web server.
• This world record benchmark result clearly demonstrates that the Sun Fire T1000 running the Solaris 10 OS and Java System Webserver 6.1 SP5 can support thousands of concurrent web server sessions while allowing larger and more complex Java applications to be run in-process for better performance and scalability. Combined with the unique 8-core architecture of the UltraSPARC T1 processor, the Java System Web Server provides the fastest web serving environment available.
• The Sun Fire T1000 (US-T1 1.0GHz) SPECweb2005's performance is second only to the industry leading Sun Fire T2000 (US-T1 1.2GHz). This is a remarkable result considering the platform's small footprint and low power consumption.
• The Sun Fire T1000 server delivers over 2x greater performance than the 2 Processor, 4 Core Dell PowerEdge 2850 server with 2 x Dual Core 2.8GHz Xeon processors, all while consuming nearly 2.5X less power and delivering nearly over 5X better performance per watt. This result is achieved in half the space of the Dell / SuSE Linux solution
• The Sun Fire T1000 server delivers 14% greater performance at over 3x less power than the 2-socket, 4 core IBM System x3650 with the latest dual core 3.0GHz Xeon 5160 "WoodCrest" processors, thereby delivering 3.5X better performance per watt. This result is achieved in half the space of the IBM / Red Hat Linux solution
• The Sun Fire T1000 server delivers 1.3x greater performance than the 4-way IBM p5 550 with 1.9 GHz POWER5+ processors, while consuming over 4X less power and occupying 1/4 the space. The Sun Fire T1000 delivers nearly 5.5X higher performance per watt than IBM's Power 5 / AIX solution
• Based on the SWaP metric, the Sun Fire T1000 delivers up to 10X higher server efficiency than the Xeon based solutions and 22X higher efficiency than the IBM p550.
• The Sun Fire T1000 results adds to the overall performance world record result achieved by the Sun Fire T2000 in December 2005 - a result that has never been beaten. These results demonstrate that the Sun Fire T2000 delivers the best overall performance, while the Sun Fire T1000 achieves the best overall levels of efficiency (space, power consumption and SWaP) which is critical for customers operating with data centers constrained by space or thermal limits
• Customers prefer to use secured web server technology by default. However this typically means double the hardware. Consequently, due to cost concerns, customers only use SSL for the most critical functions. With the Sun Fire T1000 unique hardware design and Solaris 10's software features limiting the performance penalty of secured web services to only 10%, users can use SSL enabled web service by default.
• This world record result shows that Solaris 10/JSW 6.1 SP5 delivers superior performance over Linux/Zeus/Tomcat performance.

(4) Sun Fire T1000 (8 cores, 1 chip) 10466 SPECweb2005. IBM p5 550 (4 cores, 2 chips) 7881 SPECweb2005. IBM System x3650 (4 cores, 2 chips) 9182 SPECweb2005. Dell PowerEdge 2850 (4 cores, 2 chips) 4,850 SPECweb2005SPEC, SPECweb reg tm of Standard Performance Evaluation Corporation. Results from www.spec.org as of May 30th 2006. IBM x3650 results from http://www.intel.com/performance/server/xeon/web.htm. IBM x3650 Specifications from brochure, 06/01/06: ftp://ftp.software.ibm.com/common/ssi/rep_sp/n/XSD02333USEN/XSD02333USEN.PDF IBM x3650 power rating estimated by calculating 70% of the power supply data reported in the product brochure
IBM p550 specifications, 10/06/05, from http://www-03.ibm.com/servers/eserver/pseries/hardware/entry/550.html
IBM 550 power ratings calculated by applying 70% of the power supply data published in "Facts and Features Report", 10/06/05, posted at http://www-03.ibm.com/servers/eserver/pseries/hardware/factsfeatures.html
Dell Power Measurements taken from the Dell Power Calculator, 03/06/06, posted: http://www1.us.dell.com/content/topics/topic.aspx/global/products/pedge/topics/en/config_calculator?c=us&cs=555&l=en&s=biz. System configured with 2 x Dual Core 2.8GHz processors, 16GB RAM, 2 x USCSI disks
Sun Fire T1000 server power consumption taken from measurements made during the benchmark run.

The Sun Fire T1000 server running Solaris 10 OS extended its world record leadership on the SPECjbb2005 Java benchmark by beating both 2 & 4-way Intel Xeon / Windows servers from Dell and IBM by 1.5X - 2.5X in a head-to-head test of Java Virtual Machine performance. These results were achieved in half the space of the IBM Xeon server, and with 5X higher performance per watt. The Sun Fire T1000 server also outperformed a 2-Way IBM Power 5 server running AIX by 1.8X, in 4X less space, and delivered an incredible 17X higher server efficiency (SWaP) rating.

These results confirm the Sun Fire CoolThreads servers as the most efficient and cost-effective platforms on the planet for deployment of JVM components, providing the foundation for the delivery of highly scalable and secure web services.

Benchmark Description

SPECjbb2005 (Java Business Benchmark) measures the performance of a Java implemented application tier (server-side Java). The benchmark is based on the order processing in a wholesale supplier application. The metrics given are number of SPECjbb2005 bops (Business Operations per Second) and SPECjbb2005 bops/JVM (bops per JVM instance).

Benchmark Outcome (3)

• Java(TM) Platform, Standard Edition 5.0_08 with Java HotSpot (TM) Virtual Machine technology extends Sun's best in class Java server performance with new SPECjbb2005 World records running the open-source Ultra SPARC T1 processor, demonstrating performance gains exceeding 17% over previous Sun World Record publications using Java SE 5.0_06.
• A Sun Fire T1000 Server set a new world record by delivering 60,323 SPECjbb2005 bops for the SPECjbb2005 Benchmark, beating all other 1 Rack Unit / 1 Processor servers.
• The Sun Fire T1000 server beat IBM's 2-way, 4 core x346 server configured with 2.8GHz Intel Xeon dual core processors by 1.5X, while consuming nearly 2.5X less power in 50% of the space and delivering over 3.5X higher performance per watt.
• The Sun Fire T1000 server beat IBM's 2-way, 2 core p520 server configured with 1.9GHz Power 5+ processors by 1.8x, while consuming over 2X less power in 4X less space and delivering over 4X better performance per watt.
• The Sun Fire T1000 server beat Dells 2-way PowerEdge SC1425 server configured with 2 x 3.6GHz Intel Xeon processor by 2.5X in performance, while consuming 2X less power and delivering over 5X better performance per watt.
• Based on the SWaP metric, the Sun Fire T1000 delivers 17X higher server efficiency than the IBM p520 and over 7X higher efficiency than the IBM x346.

(3) Sun Fire T1000 Server (1 chip, 8 cores, 1-way) 60,323 SPECjbb2005 bops, 15,081 SPECjbb2005 bops/JVM
IBM x346 (2 chip, 4 cores, 4-way) 39,585 SPECjbb2005 bops, 39,585 SPECjbb2005 bops/JVM
IBM p520 (1 chip, 2 cores, 2-way) 32,820 SPECjbb2005 bops, 32,820 SPECjbb2005 bops/JVM
Dell SC1425 (2 chip, 2 cores, 2-way) 24,208 SPECjbb2005 bops, 24,208 SPECjbb2005 bops/JVM
SPEC, SPECjbb reg tm of Standard Performance Evaluation Corporation. Sun Fire T1000 results submitted to SPEC. Other results as of 05/06/2006 on www.spec.org
IBM x346 Specifications from brochure, 09/05/05: http://www-03.ibm.com/servers/eserver/xseries/x346.html
IBM x346 power rating estimated by calculating 70% of the power supply data reported in the product brochure
IBM p520 specifications, 10/06/05, from http://www-03.ibm.com/servers/eserver/pseries/hardware/entry/520.html
IBM p520 power ratings calculated by applying 70% of the power supply data published in "Facts and Features Report", 10/06/05, posted at http://www-03.ibm.com/servers/eserver/pseries/hardware/factsfeatures.html
Dell SC1425 power rating taken from Input Power reported on Dell Product Power Configuration Calculator, 11/24/05. System configured with 2 x Xeon processors, 8GB DDR2 memory and 1 x Hard Disk: http://www1.us.dell.com/content/topics/topic.aspx/global/products/pedge/topics/en/config_calculator?c=us&;cs=555&l=en&s=biz
Sun Fire T1000 server power consumption taken from measurements made during the benchmark run.

Benchmark Description

The Web Tier Consolidation benchmark was designed by Sun to measure maximum throughput of of static web operations that a system can perform.

This benchmark demonstrates that a single Sun Fire T1000 server running multiple Web Server applications each within an individual Solaris Container to provide an isolated run-time environment can replace multiple Intel Web Servers with competitive performance at lower cost and power requirements.

Benchmark Outcome (2)

• One Sun Fire T1000 equipped with the 1.0 GHz UltraSPARC T1 processor, combined With multiple Apache webservers each running in a separate Solaris Container, provided total throughput of 12,597 HTTP ops/s.
• Demonstrates that a Sun Fire T1000 provides price/performance and performance/watts benefits to replace a 2 processor Xeon-based Dell PowerEdge 1850.
• Solaris Containers provide a means of creating virtualized operating system environments within an instance of Solaris, allowing one or more processes to run in isolation from other activity on the system.

• Read the Full Blueprint

(2) Consolidation Benchmark Performance as of 12/05/2005 as measured by Sun on the following platforms running the Apache 1.3.3 webserver: Sun Fire T1000 1.0 GHz (8 cores, 1 chip) 16 GB main memory, Solaris 10, 4,100 users, 12,597 static ops/s, 180 MB/s net throughput, 175w; Dell PowerEdge 1850 3.6 GHz (2 cores, 2 chips) 4 GB main memory, RHEL4 U1, 2,000 users, 6,595 static ops/s, 99 MB/s net throughput, 260w. Sun Fire T1000 and Dell PowerEdge 1850 server power consumption taken from measurements made during the benchmark run.

Benchmark Description

The RSA/DSA Cryptography benchmark was developed by Sun to measure maximum throughput of RSA/DSA private key (sign) operations that a system can perform. This is a critical performance factor as organizations look to implement greater levels of real time security to their on-line properties.

On multi-chip and/or multi-core systems, multiple processes are used to achieve the maximum throughput. Two micro benchmark programs are used, pk11rsaperf/pk11dsaperf on Solaris and OpenSSL speed test on non-Solaris systems. Though each micro benchmark uses different crypto APIs, they both measure the raw throughput of the same crypto operations.

Benchmark Outcome (1)

• The Sun Fire T2000 achieved 12,850 1024-bit RSA signs/s and 18,720 1024-bit DSA signs/s with 1 UltraSPARC T1 processor and 32 GB memory running Solaris 10 3/05 HW2. The Sun Fire T1000 achieved 10,764 1024-bit RSA signs/s and 17,023 1024-bit DSA signs/s with 1 UltraSPARC T1 processor and 16 GB memory running Solaris 10 3/05 HW2.
• The UltraSPARC T1 chip contains Modular Arithmetic Unit (MAU) per core which speeds up the modular exponentiation operation in RSA and DSA significantly. This benchmarking effort measures the maximum bandwidth of RSA/DSA sign (decrypt) operations that a single UltraSPARC T1 chip can provide across all its 8 cores, in ops/sec.
• The Sun Fire T2000 server delivers 10.7 times greater RSA performance and 6.6 times greater DSA performance than the 2-way IBM p5 510 with 1.5 GHz POWER5 processors.
• The Sun Fire T2000 server delivers 6.6 times greater RSA performance and 4.5 times greater DSA performance than the 2 way Dell PowerEdge 1850 with 3.6 GHz Xeon processors.
• The Sun Fire T2000 server delivers 2.1 times greater RSA performance and 2.9 times greater DSA performance than the 16-way Fujitsu PRIMEPOWER850 with 1.9 GHz SPARC64 V processors.
• The Sun Fire T1000 server similarly beat all measured competitors, with performance that scales linearly with frequency from the T2000 server.