FDVM_DD (1158349), страница 4

Просмтор этого файла доступен только зарегистрированным пользователям. Но у нас супер быстрая регистрация: достаточно только электронной почты!

Текст из файла (страница 4)

A(I₁,I₂, ..., I_N)

by

N

base(A(N+2) + I₁ +  A(N-j+2) * I_j )

i=2

where A - distributed array name,

base - i0000m , if A is of type integer

r0000m , if A is of type real

d0000m , if A is of type double precision

c0000m , if A is of type complex

l0000m , if A is of type logical

Generates the expression

A(N+2) + I₁ +  A(N-j+2) * I_j

where A - distributed array name.

Traverses the expression e and linearizes each distributed array element reference (calls DistArrayRef( )).

Traverses the expression ele and linearizes distributed array element references, except whole array reference.

The function traverses the expression e and linearizes each distributed array element reference (calling DistArrayRef( )). If compilation mode is set on debugging this function inserts the statements before stmt to check values of the variables during the execution in debugging mode.

The function traverses the expression e and linearizes each distributed array element reference. If compilation mode is set on debugging this function inserts the statements before and after stmt to check values of the variables during the execution in debugging mode.

The function traverses the expression e and linearizes each distributed array element reference except whole array reference. If compilation mode is set on debugging this function inserts the statements before stmt to check values of the variables during the execution in debugging mode.

4.3 Parallel loop

The regular parallel loop:

*DVM$ PARALLEL (I1, ..., In) ON A(…)...

DO label I1 = ...

. . .

DO label In = ...

loop-body

label CONTINUE

is translated into

[ REDUCTION-block-1 ]

* creating parallel loop

dvm000(ipl) = crtpl(n)

[ SHADOW-RENEW-block-1 ]

[ SHADOW-START-block ]

[ SHADOW-WAIT-block ]

[ REMOTE-ACCESS-block ]

* mapping parallel loop (1)

dvm000(j) = mappl(dvm000(ipl),A,...)

[ SHADOW-RENEW-block-2 ]

* inquiry of continuation of parallel loop execution

lab1 if(dopl(dvm000(ipl)) .eq. 0) go to lab2

DO label I1 = ...

. . .

DO label In = ...

loop-body

label CONTINUE

go to lab1

* terminating parallel loop

lab2 dvm000(k) = endpl(dvm000(ipl))

[ REDUCTION-block-2 ]

The function generates and inserts in procedure all the statements preceding the DO nest. In addition, the initial, end and step value of do-variables in parallel DO-nest are changed. The statements followed end statement of parallel loop:

label CONTINUE

are generated by TransFunc( ) when this statement is processing.

If compilation mode is set on debugging, the CALL statements:

call dbegpl(...)

call diter(...)

call dendl(...)

are created and inserted in block (1) before IF statement, before first statement of parallel loop body ,and after the statement

go to lab1

accordingly.

If compilation mode is set on performance analyzing, the CALL statements:

call bploop(...)

call eloop(...)

are created and inserted before the first and after the last statement of the block (1).

The following functions are called for creating blocks implementing SHADOW_RENEW, REDUCTION, and REMOTE_ACCESS specifications:

ShadowList

ReductionList

RemoteVariableList

If compilation mode is set on debugging, the CALL statements:

call dbegpl(...)

call diter(...)

are created and inserted before the DO loop nest and before the first statement of parallel loop body accordingly. This function generates the REDUCTION-block-1 if need be.

If compilation mode is set on performance analyzing, the CALL statement:

call bploop(...)

is created and inserted before the DO loop nest.

Generates and inserts in procedure the statements:

* creating reduction

dvm000(irv) = crtrgf(reduction-function, red-var,...)

* including reduction in reduction group

dvm000(i) = insred(gref,dvm000(irv))

for each reduction in reduction list and the statement

* storing values of reduction variables

dvm000(j) = saverv(gref)

Generates and inserts in procedure the statement:

* including shadow edge in the group

dvm000(i) = inssh(gref,array,...)

for each array in the renewee-list.

The function generates and inserts in procedure the statements for reading remote data in buffer.

1) In case of one element of array is remote variable the following statement is generated:

* copying of element of distributed array

dvm000(i) = rwelmf(array-header,buffer-var,...)

2) In case of n-dimensional section of array is remote variable the statements for creating distributed buffer array and copying distributed arrays are generated.

*

* creating abstract machine representation

dvm000(iamv) = crtamv(am,n,...)

* mapping abstract machine representation

dvm000(i) = distr(dvm000(iamv),...)

* storing lower bounds of array dimensions in header of

* distributed array

dvm000(j+n+3) = lower-bound-of-1-st-section-dimension

. . .

buffer(j+2*N) = lower-bound-of-n-th-section-dimension

Характеристики

Список файлов учебной работы