source: docs/PACT2011/03-research.tex @ 1047

Last change on this file since 1047 was 1046, checked in by ksherdy, 9 years ago

Update Table 4,5

File size: 10.4 KB
3%Describe key technology behind Parabix
4%Introduce SIMD;
5%Talk about SSE
6%Highlight which SSE instructions are important
7%TAlk about each pass in the parser; How SSE is used in every phase...
8%Benefits of SSE in each phase.
11% Extract section 2.2 and merge into 3.   Add a new subsection
12% in section 2 saying a bit about SIMD.   Say a bit about pure SIMD vertical
13% operations and then mention the pack operations that allow
14% us to implement transposition efficiently in parallel. 
15% Also note that the SIMD registers support bitwise logic across
16% their full width and that this is extensively used in our work.
18% Also, it could be good to have a small excerpt of a byte-at-a-time
19% scanning loop for XML, e.g., extracted from Xerces in section 2.1. 
20% Just a few lines showing the while loop - Linda can tell you the file.
23% This section focuses on the
26% With this method, byte-oriented character data is first transposed to eight parallel bit streams, one for each bit position within the character code units (bytes). These bit streams are then loaded into SIMD registers of width $W$ (e.g., 64-bit, 128-bit, 256-bit, etc). This allows $W$ consecutive code units to be represented and processed at once. Bitwise logic and shift operations, bit scans, population counts and other bit-based operations are then used to carry out the work in parallel \cite{CameronLin2009}.
28% The results of \cite{CameronHerdyLin2008} showed that Parabix, the predecessor of Parabix2, was dramatically faster than both Expat 2.0.1 and Xerces-C++ 2.8.0.
29% It is our expectation is that Parabix2 will outperform both Expat 2.0.1 and Xerces-C++ 3.1.1 in terms of energy consumption per source XML byte.
30% This expectation is based on the relatively-branchless code composition of Parabix2 and the more-efficient utilization of last-level cache resources.
31% The authors of \cite {bellosa2001, bircher2007, bertran2010} indicate that such factors have a considerable effect on overall energy consumption.
32% Hence, one of the foci in our study is the manner in which straight line SIMD code influences energy usage.
34This section provides an overview of the SIMD-based parallel bitstream XML parsers, Parabix1 and Parabix2. A comprehensive study of Parabix2 can be found in the technical report ``Parallel Parsing with Bitstream Addition: An XML Case Study'' \cite{Cameron2010}.
38% Our first generation parallel bitstream XML parser---Parabix1---uses employs a less conventional approach of SIMD technology to represent text in parallel bitstreams. Bits of each stream are in one-to-one-correspondence with the bytes of a character stream. A transposition step first transforms sequential byte stream data into eight basis bitstreams for the bits of each byte.
40At a high level, Parabix1 processes source XML in a functionally equivalent manner as a traditional processor. That is, Parabix1 moves sequentially through the source document, maintaining a single cursor position throughout the parsing process. Where Parabix1 differs from the traditional parser is that it scans for key markup characters using a series of basis bitstreams.
41A bitstream is simply a sequence of $0$s and $1$s, where there is one such bit in the bitstream for each character in a source data stream. A basis bitstream is a bitstream that consists of only transposed textual XML data.
42In other words, a source character consisting of $M$ bits can be represented with $M$ bitstreams and
43by utilizing $M$ SIMD registers of width $W$, it is possible to scan through $W$ characters in parallel.
44The register width $W$ varies between 64-bit for MMX, 128-bit for SSE, and 256-bit for AVX.
45Figure \ref{fig:BitstreamsExample} presents an example of how we represent 8-bit ASCII characters using eight bitstreams. $B_0 \ldots B_7$ are the individual bitstreams. The $0$ bits in the bitstreams are represented by periods, so that the $1$ bits stand out.
50source data & \verb`<t1>abc</t1><tag2/>`\\
51$B_0$ & \verb`..`\\
52$B_1$ & \verb`...1.11.1..1...1111`\\
53$B_2$ & \verb`11.1...`\\
54$B_3$ & \verb`1..1...11..11....11`\\
55$B_4$ & \verb`1111...1.11111..1.1`\\
56$B_5$ & \verb`1111111111111111111`\\
57$B_6$ & \verb`.1..111..1...111...`\\
58$B_7$ & \verb`...................`\\
61\caption{Parallel Bitstream Example}
65In order to represent the byte-oriented character data as parallel bitstreams, the source data is first loaded in sequential order and converted into its transposed representation through a series of packs, shifts, and bitwise operations.
66Using the SIMD capabilities of current commodity processors, this transposition of source data to bitstreams incurs an amortized overhead of about 1 CPU cycle per byte for transposition \cite{CameronHerdyLin2008}. When parsing, we need to consider multiple properties of characters at different stages during the process. Using the basis bitstreams, it is possible to combine them using bitwise logic in order to compute character-class bitstreams; that is, streams that identify the positions at which characters belonging to a specific character class occur. For example, the $j$-th character is an open angle bracket `<' if and only if the $j$-th bit of $B_2, B_3, B_4, B_5 =$ 1 and the $j\th$ bit of $B_0, B_1, B_6, B_7 =$ 0. Once these character-class bitstreams are created, a {\em bit-scan} operation, which is an 1-cycle intrinsic function for commodity processors, can be used for sequential markup scanning and data validation operations. A common operation in all XML parsers is start tag validation. Starts tags begin with `<' and end with either ``/>'' or ``>'' (depending whether the element tag is an empty element tag or not, respectively).
71source data                     & \verb`<t1>abc</t1><tag2/>`\\
72$M_0 = 1$                       & \verb`1..................`\\
73$M_1 = advance(M_0)$            & \verb`.1.................`\\
74$M_2 = bitscan('>')$            & \verb`...1...............`\\
75$M_3 = advance(M_2)$            & \verb`....1..............`\\
76$M_4 = bitscan('<')$            & \verb`.......1...........`\\
77$M_5 = advance(M_4)$            & \verb`........1..........`\\
78$M_6 = advance(M_5)$            & \verb`.........1.........`\\
79$M_7 = bitscan('<')$            & \verb`............1......`\\
80$M_{8} = advance(M_7)$  & \verb`.............1.....`\\
81$M_{9} = bitscan('/')$  & \verb`.................1.`\\
82$M_{10} = advance(M_{9})$       & \verb`..................1`\\
85\caption{Parabix1 Start Tag Validation (Conceptual)}
89Figure \ref{fig:Parabix1StarttagExample} demonstrates the concept of start tag validation as performed in Parabix1. The first marker stream $M_0$ is created and the parser begins scanning the source data for an open angle bracket `<', starting at position 1. Since the source data begins with `<', $M_0$ is assigned a cursor position of 1. The $advance$ operation then then shifts the $M_0$'s cursor position by 1, resulting in the creation of a new marker stream, $M_1$, with the cursor position at 2. The following $bitscan$ operation takes the cursor position from $M_1$ and sequentially scans every position until it locates either an `>'. It finds a `>' at position 4 and returns that as the new cursor position for $M_2$. Calculating $M_3$ advances the cursor again, and the $bitscan$ used to create $M_4$ locates the new opening angle bracket. This process continues until in this manner until all start tags are validated.
91Unlike traditional parsers, these sequential operations are accelerated significantly since the bit-scan operation can skip up to $w$ positions, where $w$ is the processor word width in bits. This approach has recently been applied to Unicode transcoding and XML parsing to good effect, with research prototypes showing substantial speed-ups over even the best of byte-at-a-time alternatives \cite{CameronHerdyLin2008, Herdy2008, Cameron2009}.
95In Parabix2, we replaced the sequential single-cursor parsing using bit scan instructions with a parallel parsing method using bitstream addition.
96Unlike the single-cursor approach of Parabix1 (and conceptually of all sequential XML parsers),
97Parabix2 processes multiple cursors in parallel. For example, using the source data from
98Figure \ref{fig:Parabix1StarttagExample}, Figure \ref{fig:Parabix2StarttagExample} shows how Parabix2 identifies and moves each of the start tag markers forwards to the corresponding end tag. Unlike Parabix1, Parabix2 begins scanning by creating two character-class marker streams, $N$, denoting the position of every alpha numeric character within the basis stream, and $M_0$, marking the position of every potential start tag in the bitstream. $M_0$ is then advanced to create $M_1$, which is fed into the first $scanto$ operation along with $N$.  To handle variable length tag names, the $scanto$ operation effectively locates the cursor positions of the end tags in parallel by adding $M_1$ to $N$, and using the bitwise AND operation of the negation of $N$ to find only the true end tags of $M_1$. Because and end tag may end on an `/' or '>', $scanto$ is called again to advance any cursor from `/' to `>'. For additional details, see the technical report \cite{Cameron2010}.
104source data                     & \verb`<t1>abc</t1><tag2/>`\\
105$N = $ alphanumerics            & \verb`.11......11..1111..`\\
106$M_0 = \texttt{[<]}$            & \verb`1...........1......`\\
107$M_1 = advance(M_0)$            & \verb`.1...........1.....`\\
108$M_2 = scanto(M_1, N)$          & \verb`...1.............1.`\\
109$M_3 = scanto(M_2, N)$          & \verb`...1..............1`
112\caption{Parabix2 Start Tag Validation(Conceptual)}
116In general, the set of bit positions in a marker bitstream may be considered to be the current parsing
117positions of multiple parses taking place in parallel throughout the source data stream.
118A further aspect of the parallel method is that conditional branch statements used to identify
119syntax error at each each parsing position are eliminated. Although we do not show it in the prior examples,
120error bitstreams can be used to identify any well-formedness errors found during the parsing process. Error positions are gathered and
121processed in as a final post processing step. Hence, Parabix2 offers additional parallelism over Parabix1 in the form of multiple cursor
122parsing as well as further reducing branch misprediction penalties.
Note: See TracBrowser for help on using the repository browser.