# source:docs/HPCA2011/00-abstract.tex@1320

Last change on this file since 1320 was 1302, checked in by lindanl, 8 years ago

Create a directory for HPCA

File size: 2.0 KB
Line
1XML is a set of rules for the encoding of documents in machine-readable form.
2The simplicity and generality of the rules make it widely used in web services and database
3systems.  Traditional XML parsers are built around a
4byte-at-a-time processing model where each character token
5of an XML document is examined in sequence.  Unfortunately, the byte-at-a-time
6sequential model is a performance barrier in more demanding applications,
7is energy-inefficient, and makes poor use of the wide SIMD registers
8and other parallelism features of modern processors.
9
10This paper assesses the energy and performance of a new approach
11to XML parsing, based on parallel bit stream technology, and as implemented on successive
12software generations of the Parabix XML parser.
13In Parabix, we first convert character streams into sets of parallel
14bit streams. We then exploit the SIMD operations prevalent on commodity-level hardware for performance.
15The first generation Parabix1 parser exploits the processor built-in $bitscan$ instructions
16over these streams to make multibyte moves but follows an otherwise sequential
17approach.  The second generation Parabix2 technology adds further
18parallelism by replacing much of the sequential
19bit scanning with a parallel scanning approach based on bit stream
20addition.  We evaluate Parabix1 and Parabix2
21against two widely used XML parsers, James Clark's Expat and Apache's Xerces, and
22across three generations of x86 machines, including the new Intel
23\SB{}.  We show that Parabix2's speedup is 2$\times$--7$\times$
24over Expat and Xerces.  In stark contrast to the energy expenditures necessary
25to realize performance gains through multicore parallelism, we also show
26that our Parabix parsers deliver energy savings in direct proportion
27to the gains in performance.  In addition, we assess the scalability advantages
28of SIMD processor improvements across Intel processor generations,
29culminating with an evaluation of the 256-bit AVX technology in
30\SB{} versus the now legacy 128-bit SSE technology.
31
Note: See TracBrowser for help on using the repository browser.