Better Performance Per Watt
89% better
SPECpower_ssj2008 v1.1.3
Testing conducted by Apple in March 2009 using a preproduction 2.93GHz 8-core Xserve (SPECpower_ssj™2008 result of 464 overall ssj_ops/watt; 173W at Active Idle; 227,974 ssj_ops and 334W at 100% target load) and a shipping 3.0GHz 8-core Xserve (SPECpower_ssj™2008 result of 245 overall ssj_ops/watt; 213W at Active Idle; 141,739 ssj_ops and 353W at 100% target load). All units were configured for optimal memory performance (18GB for 2.93GHz 8-core units and 16GB for 3.0GHz 8-core units). SPEC™ and the benchmark name SPECpower_ssj™ are trademarks of the Standard Performance Evaluation Corporation (SPEC); see www.spec.org for more information. Competitive benchmark results stated here reflect internal Apple testing and were submitted to SPEC in February 2009. For the latest SPECpower_ssj2008 benchmark results, visit www.spec.org/power_ssj2008. Performance tests are conducted using specific computer systems and reflect the approximate performance of Xserve.
Lower Active Idle
19% lower
SPECpower_ssj2008 v1.1.3
Testing conducted by Apple in February 2009 using a preproduction 2.93GHz 8-core Xserve (SPECpower_ssj™2008 result of 464 overall ssj_ops/watt; 173W at Active Idle; 227,974 ssj_ops and 334W at 100% target load) and a shipping 3.0GHz 8-core Xserve (SPECpower_ssj™2008 result of 245 overall ssj_ops/watt; 213W at Active Idle; 141,739 ssj_ops and 353W at 100% target load). All units were configured for optimal memory performance (18GB for 2.93GHz 8-core units and 16GB for 3.0GHz 8-core units). SPEC® and the benchmark name SPECpower_ssj™ are trademarks of the Standard Performance Evaluation Corporation (SPEC); see www.spec.org for more information. Competitive benchmark results stated here reflect internal Apple testing and were submitted to SPEC in February 2009. For the latest SPECpower_ssj2008 benchmark results, visit www.spec.org/power_ssj2008. Performance tests are conducted using specific computer systems and reflect the approximate performance of Xserve.
- SPEC CPU2006 Up to 80% faster
- Massive Memory Bandwidth Up to 140% faster
- Server Performance Up to 100% faster
- Application Performance Up to 50% faster
CFP2006 and CINT2006 Rates
8-core 2.93GHz Xserve up to 80% faster than 8-core 3.0GHz Xserve (previous generation)
Testing conducted by Apple in February 2009 using preproduction 2.93GHz 8-core Xserve units and shipping 3.0GHz 8-core Xserve units. SPEC® is a registered trademark of the Standard Performance Evaluation Corporation (SPEC); see www.spec.org for more information. Performance tests are conducted using specific computer systems and reflect the approximate performance of Xserve.
Massive Memory Bandwidth
8-core 2.93GHz Xserve up to 140% faster than 8-core 3.0GHz Xserve (previous generation)
Testing conducted by Apple in February 2009 using preproduction 2.93GHz 8-core Xserve units, preproduction 2.26GHz quad-core Xserve units, and shipping 3.0GHz 8-core Xserve units. All units were configured for optimal memory performance (18GB for 8-core 2.93GHz units, 12GB for quad-core 2.26GHz units, and 16GB for 8-core 3.0GHz units). Results are based on the STREAM v.5.8 benchmark (www.cs.virginia.edu/stream/ref.html) using OMP support for multiprocessor-compiled builds. Performance tests are conducted using specific computer systems and reflect the approximate performance of Xserve.
8-Core Server Performance
8-core 2.93GHz Xserve up to 100% faster than 8-core 3.0GHz Xserve (previous generation)
- Testing conducted by Apple in February 2009 using a preproduction 2.93GHz 8-core Xserve (2 chips, 8 cores, 4 cores per chip, 2.93GHz; SPECjbb2005 bops = 203,439, SPECjbb2005 bops/JVM = 50,860) and a shipping 3.0GHz 8-core Xserve (2 chips, 8 cores, 4 cores per chip, 3.0GHz; SPECjbb2005 bops = 103,387, SPECjbb2005 bops/JVM = 25,847). SPEC® and SPECjbb2005® are trademarks of the Standard Performance Evaluation Corporation (SPEC); see www.spec.org for more information. Competitive benchmark results stated above reflect internal Apple testing and were submitted to SPEC in February 2009. For the latest SPECjbb2005 benchmark results, visit www.spec.org/osg/jbb2005. Performance tests are conducted using specific computer systems and reflect the approximate performance of Xserve.
- Testing conducted by Apple in February 2009 using preproduction 2.93GHz 8-core Xserve units and shipping 3.0GHz 8-core Xserve units. All units were configured for optimal memory performance (24GB for 8-core 2.93GHz units and 16GB for 8-core 3.0GHz units). Units were connected via Fibre Channel to a Promise VTrak E-Class RAID subsystem, with 16 SAS drives and 2 controllers, equally divided into 4 RAID 5 arrays. Results are based on the AFPBench benchmark, using up to 40 physical client systems and making requests simulating typical home directory usage. Performance tests are conducted using specific computer systems and reflect the approximate performance of Xserve.
8-Core Application Performance
8-core 2.93GHz Xserve up to 50% faster than 8-core 3.0GHz Xserve (previous generation)
Testing conducted by Apple in February 2009 using preproduction 2.93GHz 8-core Xserve units and shipping 3.0GHz 8-core Xserve units. Content was encoded using the H.264 for iPod Video 640-by-480 preset in Compressor and Qmaster with the maximum number of instances available on each system. Performance tests are conducted using specific computer systems and reflect the approximate performance of Xserve.
- SAS and SATA Throughput Outstanding storage performance
- Three-Drive SATA Throughput Xserve RAID Card performance
- Three-Drive SAS Throughput Xserve RAID Card performance
SAS and SATA Throughput
Outstanding storage performance
Iometer megabytes per second
Average seek time in milliseconds (shorter is better)
Testing conducted by Apple in February 2009 using preproduction 2.93GHz 8-core Xeon-based Xserve units configured with Xserve RAID Card. Testing conducted using Iometer 2006.07.27 with a 30-second ramp-up, 5-minute run duration, 512KB request size, 4 outstanding IOs, and using non-OS-cached reads and writes. System configured with the operating system on 128GB solid-state drive. Performance tests are conducted using specific computer systems and reflect the approximate performance of Xserve RAID Card. Drive average read latency data from manufacturer’s published data sheets.
Three-Drive SATA Throughput
Xserve RAID Card performance
Iometer megabytes per second
Testing conducted by Apple in February 2009 using preproduction 2.93GHz 8-core Xeon-based Xserve units configured with Xserve RAID Card. Testing conducted using Iometer 2006.07.27 with a 30-second ramp-up, 5-minute run duration, 512KB request size, 4 outstanding IOs, and using non-OS-cached reads and writes. System configured with the operating system on Xserve solid-state drive. Performance tests are conducted using specific computer systems and reflect the approximate performance of Xserve RAID Card.
Three-Drive SAS Throughput
Xserve RAID Card performance
Iometer megabytes per second
Testing conducted by Apple in February 2009 using preproduction 2.93GHz 8-core Xeon-based Xserve units configured with Xserve RAID Card. Testing conducted using Iometer 2006.07.27 with a 30-second ramp-up, 5-minute run duration, 512KB request size, 4 outstanding IOs, and using non-OS-cached reads and writes. System configured with the operating system on Xserve solid-state drive. Performance tests are conducted using specific computer systems and reflect the approximate performance of Xserve RAID Card.



