Specifications

10
Technology Overview
Xserve
HMMER
HMMER is a set of popular bioinformatics applications that implement Hidden Markov
Models—statistical models of multiple sequence alignments used in computational
biology to search for patterns in DNA sequences. To test processor and memory
performance in the new Xserve, hmmcalibrate was run on Xserve G5 and Xserve
against the same sample data set, then runtimes were compared.
hmmcalibrate performance: Almost three times the performance of Xserve G5
Xserve
Quad 3.0GHz Xeon
Xserve G5
Dual 2.3GHz G5
2.9x
Baseline
Testing conducted by Apple in October 2006 using preproduction quad 3.0GHz Xeon-based Xserve units; Xserve G5 systems were shipping units. For
HMMER tests, Xserve G5 was tested with the binary distribution available at hmmer.janelia.org; Xeon-based Xserve was tested with source compiled
for Xserve using the patch available from www.scalableinformatics.com and the following compile-time options: -O3 --enable-threads. Performance
tests are conducted using specific computer systems and reflect the approximate performance of Xserve.
Video compression
Compressing video into popular delivery formats takes a tremendous amount of
processor performance. Organizations looking to produce content in a variety of
delivery formats can leverage batch processing software such as Apple’s Compressor
application, part of Final Cut Studio, on an Xserve to simplify and automate content
compression and delivery. Apple tested a variety of video formats being encoded
into MPEG-2 on Xserve and Xserve G5.
Compressor 2.3 results: MPEG-2 encoding performance
Xserve performs MPEG-2 compression up to 2.3 times faster than Xserve G5,
depending on the source footage format.
Xserve—DV content
Quad 3.0GHz Xeon
Xserve—HDV content
Quad 3.0GHz Xeon
2.3x
1.8x
Xserve—SD content
Quad 3.0GHz Xeon
Xserve G5
Dual 2.3GHz G5
1.9x
Baseline
Testing conducted by Apple in October 2006 using preproduction quad 3.0GHz Xeon-based Xserve units; Xserve G5 systems were shipping units.
Compressor MPEG-2 6.2-Mbps two-pass encode settings were used. Performance tests are conducted using specific computer systems and reflect
the approximate performance of Xserve.
Memory Performance
Xserve incorporates a new 256-bit-wide memory architecture that allows for higher
speed, greater capacity, and better reliability. Using four channels of 667MHz DDR2
ECC fully buffered DIMMs, Xserve has a maximum memory bandwidth of 21.3GB/s. This
added bandwidth gives an immediate boost to performance. With Xserve, enormous
files and data sets can be loaded into RAM for rapid access by the two dual-core Xeon
processors. A wide range of server services as well as graphics, video, and scientific
applications benefit in particular, running dramatically faster as a direct result of more
and faster RAM.