--- manual/s_software/text/sarch.tex 2001/10/09 10:33:17 1.1 +++ manual/s_software/text/sarch.tex 2001/10/25 18:36:55 1.4 @@ -1,3 +1,4 @@ +% $Header: /home/ubuntu/mnt/e9_copy/manual/s_software/text/sarch.tex,v 1.4 2001/10/25 18:36:55 cnh Exp $ In this chapter we describe the software architecture and implementation strategy for the MITgcm code. The first part of this @@ -11,17 +12,13 @@ three-fold \begin{itemize} - \item We wish to be able to study a very broad range of interesting and challenging rotating fluids problems. - \item We wish the model code to be readily targeted to a wide range of platforms - \item On any given platform we would like to be able to achieve performance comparable to an implementation developed and specialized specifically for that platform. - \end{itemize} These points are summarized in figure \ref{fig:mitgcm_architecture_goals} @@ -30,26 +27,21 @@ of \begin{enumerate} - \item A core set of numerical and support code. This is discussed in detail in section \ref{sec:partII}. - \item A scheme for supporting optional "pluggable" {\bf packages} (containing for example mixed-layer schemes, biogeochemical schemes, atmospheric physics). These packages are used both to overlay alternate dynamics and to introduce specialized physical content onto the core numerical code. An overview of the {\bf package} scheme is given at the start of part \ref{part:packages}. - - \item A support framework called {\bf WRAPPER} (Wrappable Application Parallel Programming Environment Resource), within which the core numerics and pluggable packages operate. - \end{enumerate} This chapter focuses on describing the {\bf WRAPPER} environment under which both the core numerics and the pluggable packages function. The description -presented here is intended to be a detailed exposistion and contains significant +presented here is intended to be a detailed exposition and contains significant background material, as well as advanced details on working with the WRAPPER. The examples section of this manual (part \ref{part:example}) contains more succinct, step-by-step instructions on running basic numerical @@ -57,19 +49,20 @@ starting from an example code and adapting it to suit a particular situation will be all that is required. + \begin{figure} \begin{center} - \resizebox{!}{2.5in}{ - \includegraphics*[1.5in,2.4in][9.5in,6.3in]{part4/mitgcm_goals.eps} - } +\resizebox{!}{2.5in}{\includegraphics{part4/mitgcm_goals.eps}} \end{center} -\caption{The MITgcm architecture is designed to allow simulation of a wide +\caption{ +The MITgcm architecture is designed to allow simulation of a wide range of physical problems on a wide range of hardware. The computational resource requirements of the applications targeted range from around $10^7$ bytes ( $\approx 10$ megabytes ) of memory to $10^{11}$ bytes ( $\approx 100$ gigabytes). Arithmetic operation counts for the applications of interest range from $10^{9}$ floating point operations to more than $10^{17}$ -floating point operations.} \label{fig:mitgcm_architecture_goals} +floating point operations.} +\label{fig:mitgcm_architecture_goals} \end{figure} \section{WRAPPER} @@ -87,20 +80,21 @@ \ref{fig:fit_in_wrapper} which shows how the WRAPPER serves to insulate code that fits within it from architectural differences between hardware platforms and operating systems. This allows numerical code to be easily retargetted. + + \begin{figure} \begin{center} - \resizebox{6in}{4.5in}{ - \includegraphics*[0.6in,0.7in][9.0in,8.5in]{part4/fit_in_wrapper.eps} - } +\resizebox{!}{4.5in}{\includegraphics{part4/fit_in_wrapper.eps}} \end{center} -\caption{ Numerical code is written too fit within a software support +\caption{ +Numerical code is written too fit within a software support infrastructure called WRAPPER. The WRAPPER is portable and -can be sepcialized for a wide range of specific target hardware and +can be specialized for a wide range of specific target hardware and programming environments, without impacting numerical code that fits within the WRAPPER. Codes that fit within the WRAPPER can generally be made to run as fast on a particular platform as codes specially -optimized for that platform. -} \label{fig:fit_in_wrapper} +optimized for that platform.} +\label{fig:fit_in_wrapper} \end{figure} \subsection{Target hardware} @@ -172,7 +166,7 @@ space allocated to a particular logical processor, there will be data structures (arrays, scalar variables etc...) that hold the simulated state of that region. We refer to these data structures as being {\bf owned} by the -pprocessor to which their +processor to which their associated region of physical space has been allocated. Individual regions that are allocated to processors are called {\bf tiles}. A processor can own more @@ -186,8 +180,8 @@ \begin{figure} \begin{center} - \resizebox{7in}{3in}{ - \includegraphics*[0.5in,2.7in][12.5in,6.4in]{part4/domain_decomp.eps} + \resizebox{5in}{!}{ + \includegraphics{part4/domain_decomp.eps} } \end{center} \caption{ The WRAPPER provides support for one and two dimensional @@ -222,8 +216,8 @@ \begin{figure} \begin{center} - \resizebox{7in}{3in}{ - \includegraphics*[4.5in,3.7in][12.5in,6.7in]{part4/tiled-world.eps} + \resizebox{5in}{!}{ + \includegraphics{part4/tiled-world.eps} } \end{center} \caption{ A global grid subdivided into tiles. @@ -404,8 +398,8 @@ \begin{figure} \begin{center} - \resizebox{5in}{3in}{ - \includegraphics*[1.5in,0.7in][7.9in,4.4in]{part4/comm-primm.eps} + \resizebox{5in}{!}{ + \includegraphics{part4/comm-primm.eps} } \end{center} \caption{Three performance critical parallel primititives are provided @@ -485,8 +479,8 @@ \begin{figure} \begin{center} - \resizebox{5in}{3in}{ - \includegraphics*[0.5in,1.3in][7.9in,5.7in]{part4/tiling_detail.eps} + \resizebox{5in}{!}{ + \includegraphics{part4/tiling_detail.eps} } \end{center} \caption{The tiling strategy that the WRAPPER supports allows tiles @@ -589,8 +583,8 @@ \begin{figure} \begin{center} - \resizebox{5in}{7in}{ - \includegraphics*[0.5in,0.3in][7.9in,10.7in]{part4/size_h.eps} + \resizebox{5in}{!}{ + \includegraphics{part4/size_h.eps} } \end{center} \caption{ The three level domain decomposition hierarchy employed by the @@ -795,7 +789,7 @@ There are six tiles allocated to six separate logical processors ({\em nSx=6}). This set of values can be used for a cube sphere calculation. Each tile of size $32 \times 32$ represents a face of the -cube. Initialising the tile connectivity correctly ( see section +cube. Initializing the tile connectivity correctly ( see section \ref{sec:cube_sphere_communication}. allows the rotations associated with moving between the six cube faces to be embedded within the tile-tile communication code. @@ -812,7 +806,6 @@ by the application code. The startup calling sequence followed by the WRAPPER is shown in figure \ref{fig:wrapper_startup}. - \begin{figure} \begin{verbatim} @@ -849,6 +842,7 @@ \end{figure} \subsubsection{Multi-threaded execution} +\label{sec:multi-threaded-execution} Prior to transferring control to the procedure {\em THE\_MODEL\_MAIN()} the WRAPPER may cause several coarse grain threads to be initialized. The routine {\em THE\_MODEL\_MAIN()} is called once for each thread and is passed a single @@ -904,25 +898,6 @@ \end{enumerate} -\paragraph{Environment variables} -On most systems multi-threaded execution also requires the setting -of a special environment variable. On many machines this variable -is called PARALLEL and its values should be set to the number -of parallel threads required. Generally the help pages associated -with the multi-threaded compiler on a machine will explain -how to set the required environment variables for that machines. - -\paragraph{Runtime input parameters} -Finally the file {\em eedata} needs to be configured to indicate -the number of threads to be used in the x and y directions. -The variables {\em nTx} and {\em nTy} in this file are used to -specify the information required. The product of {\em nTx} and -{\em nTy} must be equal to the number of threads spawned i.e. -the setting of the environment variable PARALLEL. -The value of {\em nTx} must subdivide the number of sub-domains -in x ({\em nSx}) exactly. The value of {\em nTy} must subdivide the -number of sub-domains in y ({\em nSy}) exactly. - An example of valid settings for the {\em eedata} file for a domain with two subdomains in y and running with two threads is shown below @@ -955,6 +930,7 @@ } \\ \subsubsection{Multi-process execution} +\label{sec:multi-process-execution} Despite its appealing programming model, multi-threaded execution remains less common then multi-process execution. One major reason for this @@ -966,7 +942,8 @@ Multi-process execution is more ubiquitous. In order to run code in a multi-process configuration a decomposition -specification is given ( in which the at least one of the +specification ( see section \ref{sec:specifying_a_decomposition}) +is given ( in which the at least one of the parameters {\em nPx} or {\em nPy} will be greater than one) and then, as for multi-threaded operation, appropriate compile time and run time steps must be taken. @@ -1046,6 +1023,25 @@ \end{minipage} } \\ + +\paragraph{Environment variables} +On most systems multi-threaded execution also requires the setting +of a special environment variable. On many machines this variable +is called PARALLEL and its values should be set to the number +of parallel threads required. Generally the help pages associated +with the multi-threaded compiler on a machine will explain +how to set the required environment variables for that machines. + +\paragraph{Runtime input parameters} +Finally the file {\em eedata} needs to be configured to indicate +the number of threads to be used in the x and y directions. +The variables {\em nTx} and {\em nTy} in this file are used to +specify the information required. The product of {\em nTx} and +{\em nTy} must be equal to the number of threads spawned i.e. +the setting of the environment variable PARALLEL. +The value of {\em nTx} must subdivide the number of sub-domains +in x ({\em nSx}) exactly. The value of {\em nTy} must subdivide the +number of sub-domains in y ({\em nSy}) exactly. The multiprocess startup of the MITgcm executable {\em mitgcmuv} is controlled by the routines {\em EEBOOT\_MINIMAL()} and {\em INI\_PROCS()}. The first routine performs basic steps required @@ -1058,7 +1054,7 @@ output files {\bf STDOUT.0001} and {\bf STDERR.0001} etc... These files are used for reporting status and configuration information and for reporting error conditions on a process by process basis. -The {{\em EEBOOT\_MINIMAL()} procedure also sets the variables +The {\em EEBOOT\_MINIMAL()} procedure also sets the variables {\em myProcId} and {\em MPI\_COMM\_MODEL}. These variables are related to processor identification are are used later in the routine @@ -1099,6 +1095,7 @@ The WRAPPER maintains internal information that is used for communication operations and that can be customized for different platforms. This section describes the information that is held and used. + \begin{enumerate} \item {\bf Tile-tile connectivity information} For each tile the WRAPPER sets a flag that sets the tile number to the north, south, east and @@ -1112,7 +1109,7 @@ This latter set of variables can take one of the following values {\em COMM\_NONE}, {\em COMM\_MSG}, {\em COMM\_PUT} and {\em COMM\_GET}. A value of {\em COMM\_NONE} is used to indicate that a tile has no -neighbor to cummnicate with on a particular face. A value +neighbor to communicate with on a particular face. A value of {\em COMM\_MSG} is used to indicated that some form of distributed memory communication is required to communicate between these tile faces ( see section \ref{sec:distributed_memory_communication}). @@ -1169,7 +1166,7 @@ are read from the file {\em eedata}. If the value of {\em nThreads} is inconsistent with the number of threads requested from the operating system (for example by using an environment -varialble as described in section \ref{sec:multi_threaded_execution}) +variable as described in section \ref{sec:multi_threaded_execution}) then usually an error will be reported by the routine {\em CHECK\_THREADS}.\\ @@ -1186,31 +1183,6 @@ \end{minipage} } -\begin{figure} -\begin{verbatim} -C-- -C-- Parallel directives for MIPS Pro Fortran compiler -C-- -C Parallel compiler directives for SGI with IRIX -C$PAR PARALLEL DO -C$PAR& CHUNK=1,MP_SCHEDTYPE=INTERLEAVE, -C$PAR& SHARE(nThreads),LOCAL(myThid,I) -C - DO I=1,nThreads - myThid = I - -C-- Invoke nThreads instances of the numerical model - CALL THE_MODEL_MAIN(myThid) - - ENDDO -\end{verbatim} -\caption{Prior to transferring control to -the procedure {\em THE\_MODEL\_MAIN()} the WRAPPER may use -MP directives to spawn multiple threads. -} \label{fig:mp_directives} -\end{figure} - - \item {\bf memsync flags} As discussed in section \ref{sec:memory_consistency}, when using shared memory, a low-level system function may be need to force memory consistency. @@ -1228,7 +1200,7 @@ \begin{verbatim} asm("membar #LoadStore|#StoreStore"); \end{verbatim} -for an Alpha based sytem the euivalent code reads +for an Alpha based system the equivalent code reads \begin{verbatim} asm("mb"); \end{verbatim} @@ -1344,9 +1316,33 @@ because there has to be enough storage allocated for all tiles. However, the technique can sometimes be a useful scheme for reducing memory requirements in complex physical paramterisations. - \end{enumerate} +\begin{figure} +\begin{verbatim} +C-- +C-- Parallel directives for MIPS Pro Fortran compiler +C-- +C Parallel compiler directives for SGI with IRIX +C$PAR PARALLEL DO +C$PAR& CHUNK=1,MP_SCHEDTYPE=INTERLEAVE, +C$PAR& SHARE(nThreads),LOCAL(myThid,I) +C + DO I=1,nThreads + myThid = I + +C-- Invoke nThreads instances of the numerical model + CALL THE_MODEL_MAIN(myThid) + + ENDDO +\end{verbatim} +\caption{Prior to transferring control to +the procedure {\em THE\_MODEL\_MAIN()} the WRAPPER may use +MP directives to spawn multiple threads. +} \label{fig:mp_directives} +\end{figure} + + \subsubsection{Specializing the Communication Code} The isolation of performance critical communication primitives and the @@ -1371,7 +1367,7 @@ \item The {\em \_GSUM} and {\em \_EXCH} macro definitions are replaced with calls to custom routines ( see {\em gsum\_jam.F} and {\em exch\_jam.F}) \item a highly specialized form of the exchange operator (optimized -for overlap regions of width one) is substitued into the elliptic +for overlap regions of width one) is substituted into the elliptic solver routine {\em cg2d.F}. \end{itemize} Developing specialized code for other libraries follows a similar @@ -1383,16 +1379,16 @@ a series of template files, for example {\em exch\_rx.template}. This is done to allow a large number of variations on the exchange process to be maintained. One set of variations supports the -cube sphere grid. Support for a cube sphere gris in MITgcm is based +cube sphere grid. Support for a cube sphere grid in MITgcm is based on having each face of the cube as a separate tile (or tiles). -The exchage routines are then able to absorb much of the +The exchange routines are then able to absorb much of the detailed rotation and reorientation required when moving around the cube grid. The set of {\em \_EXCH} routines that contain the word cube in their name perform these transformations. They are invoked when the run-time logical parameter {\em useCubedSphereExchange} is set true. To facilitate the transformations on a staggered C-grid, exchange operations are defined -separately for both vector and scalar quantitities and for +separately for both vector and scalar quantities and for grid-centered and for grid-face and corner quantities. Three sets of exchange routines are defined. Routines with names of the form {\em exch\_rx} are used to exchange @@ -1457,7 +1453,7 @@ C | C |-THE_MODEL_MAIN :: Primary driver for the MITgcm algorithm C | :: Called from WRAPPER level numerical -C | :: code innvocation routine. On entry +C | :: code invocation routine. On entry C | :: to THE_MODEL_MAIN separate thread and C | :: separate processes will have been established. C | :: Each thread and process will have a unique ID @@ -1471,22 +1467,22 @@ C | | :: By default kernel parameters are read from file C | | :: "data" in directory in which code executes. C | | -C | |-MON_INIT :: Initialises monitor pacakge ( see pkg/monitor ) +C | |-MON_INIT :: Initializes monitor package ( see pkg/monitor ) C | | -C | |-INI_GRID :: Control grid array (vert. and hori.) initialisation. +C | |-INI_GRID :: Control grid array (vert. and hori.) initialization. C | | | :: Grid arrays are held and described in GRID.h. C | | | -C | | |-INI_VERTICAL_GRID :: Initialise vertical grid arrays. +C | | |-INI_VERTICAL_GRID :: Initialize vertical grid arrays. C | | | -C | | |-INI_CARTESIAN_GRID :: Cartesian horiz. grid initialisation +C | | |-INI_CARTESIAN_GRID :: Cartesian horiz. grid initialization C | | | :: (calculate grid from kernel parameters). C | | | C | | |-INI_SPHERICAL_POLAR_GRID :: Spherical polar horiz. grid -C | | | :: initialisation (calculate grid from +C | | | :: initialization (calculate grid from C | | | :: kernel parameters). C | | | C | | |-INI_CURVILINEAR_GRID :: General orthogonal, structured horiz. -C | | :: grid initialisations. ( input from raw +C | | :: grid initializations. ( input from raw C | | :: grid files, LONC.bin, DXF.bin etc... ) C | | C | |-INI_DEPTHS :: Read (from "bathyFile") or set bathymetry/orgography. @@ -1497,7 +1493,7 @@ C | |-INI_LINEAR_PHSURF :: Set ref. surface Bo_surf C | | C | |-INI_CORI :: Set coriolis term. zero, f-plane, beta-plane, -C | | :: sphere optins are coded. +C | | :: sphere options are coded. C | | C | |-PACAKGES_BOOT :: Start up the optional package environment. C | | :: Runtime selection of active packages. @@ -1518,7 +1514,7 @@ C | |-PACKAGES_CHECK C | | | C | | |-KPP_CHECK :: KPP Package. pkg/kpp -C | | |-OBCS_CHECK :: Open bndy Pacakge. pkg/obcs +C | | |-OBCS_CHECK :: Open bndy Package. pkg/obcs C | | |-GMREDI_CHECK :: GM Package. pkg/gmredi C | | C | |-PACKAGES_INIT_FIXED @@ -1538,7 +1534,7 @@ C |-CTRL_UNPACK :: Control vector support package. see pkg/ctrl C | C |-ADTHE_MAIN_LOOP :: Derivative evaluating form of main time stepping loop -C ! :: Auotmatically gerenrated by TAMC/TAF. +C ! :: Auotmatically generated by TAMC/TAF. C | C |-CTRL_PACK :: Control vector support package. see pkg/ctrl C | @@ -1552,7 +1548,7 @@ C | | |-INI_LINEAR_PHISURF :: Set ref. surface Bo_surf C | | | C | | |-INI_CORI :: Set coriolis term. zero, f-plane, beta-plane, -C | | | :: sphere optins are coded. +C | | | :: sphere options are coded. C | | | C | | |-INI_CG2D :: 2d con. grad solver initialisation. C | | |-INI_CG3D :: 3d con. grad solver initialisation. @@ -1560,7 +1556,7 @@ C | | |-INI_DYNVARS :: Initialise to zero all DYNVARS.h arrays (dynamical C | | | :: fields). C | | | -C | | |-INI_FIELDS :: Control initialising model fields to non-zero +C | | |-INI_FIELDS :: Control initializing model fields to non-zero C | | | |-INI_VEL :: Initialize 3D flow field. C | | | |-INI_THETA :: Set model initial temperature field. C | | | |-INI_SALT :: Set model initial salinity field. @@ -1638,7 +1634,7 @@ C/\ | | |-CALC_SURF_DR :: Calculate the new surface level thickness. C/\ | | |-EXF_GETFORCING :: External forcing package. ( pkg/exf ) C/\ | | |-EXTERNAL_FIELDS_LOAD :: Control loading time dep. external data. -C/\ | | | | :: Simple interpolcation between end-points +C/\ | | | | :: Simple interpolation between end-points C/\ | | | | :: for forcing datasets. C/\ | | | | C/\ | | | |-EXCH :: Sync forcing. in overlap regions.