| 1 |
/* Test of vector-private variables declared on loop directive. Array type. */ |
| 2 |
|
| 3 |
/* { dg-additional-options "--param=openacc-kernels=decompose" } */ |
| 4 |
|
| 5 |
/* { dg-additional-options "-fopt-info-omp-all" } |
| 6 |
{ dg-additional-options "-foffload=-fopt-info-omp-all" } */ |
| 7 |
|
| 8 |
/* { dg-additional-options "--param=openacc-privatization=noisy" } |
| 9 |
{ dg-additional-options "-foffload=--param=openacc-privatization=noisy" } |
| 10 |
Prune a few: uninteresting: |
| 11 |
{ dg-prune-output {note: variable 'D\.[0-9]+' declared in block isn't candidate for adjusting OpenACC privatization level: not addressable} } */ |
| 12 |
|
| 13 |
/* It's only with Tcl 8.5 (released in 2007) that "the variable 'varName' |
| 14 |
passed to 'incr' may be unset, and in that case, it will be set to [...]", |
| 15 |
so to maintain compatibility with earlier Tcl releases, we manually |
| 16 |
initialize counter variables: |
| 17 |
{ dg-line l_dummy[variable c_compute 0 c_loop_i 0 c_loop_j 0 c_loop_k 0] } |
| 18 |
{ dg-message "dummy" "" { target iN-VAl-Id } l_dummy } to avoid |
| 19 |
"WARNING: dg-line var l_dummy defined, but not used". */ |
| 20 |
|
| 21 |
#include <assert.h> |
| 22 |
|
| 23 |
int |
| 24 |
main (int argc, char* argv[]) |
| 25 |
{ |
| 26 |
int pt[2], i, arr[32 * 32 * 32]; |
| 27 |
|
| 28 |
for (i = 0; i < 32 * 32 * 32; i++) |
| 29 |
arr[i] = i; |
| 30 |
|
| 31 |
#pragma acc kernels copy(arr) /* { dg-line l_compute[incr c_compute] } */ |
| 32 |
/* [PR104784] For some reason, for C++, the OpenACC 'kernels' decomposition |
| 33 |
decides that a data region is needed for 'j', and subsequently requests it |
| 34 |
to be made addressable. |
| 35 |
{ dg-note {OpenACC 'kernels' decomposition: variable 'j' declared in block requested to be made addressable} {} { target c++ } l_compute$c_compute } |
| 36 |
{ dg-note {variable 'j' made addressable} {} { target c++ } l_compute$c_compute } |
| 37 |
{ dg-note {variable 'j' declared in block is candidate for adjusting OpenACC privatization level} {} { target c++ } l_compute$c_compute } */ |
| 38 |
{ |
| 39 |
int j; |
| 40 |
|
| 41 |
/* { dg-note {forwarded loop nest in OpenACC 'kernels' region to 'parloops' for analysis} {} { target *-*-* } .+1 } */ |
| 42 |
#pragma acc loop gang(num:32) /* { dg-line l_loop_i[incr c_loop_i] } */ |
| 43 |
/* { dg-note {variable 'j' declared in block isn't candidate for adjusting OpenACC privatization level: not addressable} {} { target c} l_loop_i$c_loop_i } */ |
| 44 |
/* { dg-note {variable 'i' in 'private' clause isn't candidate for adjusting OpenACC privatization level: not addressable} {} { target *-*-* } l_loop_i$c_loop_i } */ |
| 45 |
for (i = 0; i < 32; i++) |
| 46 |
{ |
| 47 |
#pragma acc loop worker(num:32) /* { dg-line l_loop_j[incr c_loop_j] } */ |
| 48 |
/* { dg-note {variable 'j' in 'private' clause isn't candidate for adjusting OpenACC privatization level: not addressable} {} { target *-*-* } l_loop_j$c_loop_j } */ |
| 49 |
/* { dg-note {variable 'k' declared in block isn't candidate for adjusting OpenACC privatization level: not addressable} {} { target *-*-* } l_loop_j$c_loop_j } */ |
| 50 |
for (j = 0; j < 32; j++) |
| 51 |
{ |
| 52 |
int k; |
| 53 |
|
| 54 |
#pragma acc loop vector(length:32) private(pt) /* { dg-line l_loop_k[incr c_loop_k] } */ |
| 55 |
/* { dg-note {variable 'pt' in 'private' clause isn't candidate for adjusting OpenACC privatization level: not addressable} {} { target *-*-* } l_loop_k$c_loop_k } */ |
| 56 |
/* { dg-note {variable 'k' in 'private' clause isn't candidate for adjusting OpenACC privatization level: not addressable} {} { target *-*-* } l_loop_k$c_loop_k } */ |
| 57 |
for (k = 0; k < 32; k++) |
| 58 |
{ |
| 59 |
pt[0] = i ^ j * 3; |
| 60 |
pt[1] = i | j * 5; |
| 61 |
arr[i * 1024 + j * 32 + k] += pt[0] * k; |
| 62 |
arr[i * 1024 + j * 32 + k] += pt[1] * k; |
| 63 |
} |
| 64 |
} |
| 65 |
} |
| 66 |
/* { dg-optimized {assigned OpenACC seq loop parallelism} {} { target *-*-* } l_loop_i$c_loop_i } */ |
| 67 |
} |
| 68 |
|
| 69 |
for (i = 0; i < 32; i++) |
| 70 |
for (int j = 0; j < 32; j++) |
| 71 |
for (int k = 0; k < 32; k++) |
| 72 |
{ |
| 73 |
int idx = i * 1024 + j * 32 + k; |
| 74 |
assert (arr[idx] == idx + (i ^ j * 3) * k + (i | j * 5) * k); |
| 75 |
} |
| 76 |
|
| 77 |
return 0; |
| 78 |
} |