| 1 |
/* Test OpenACC 'kernels' construct decomposition. */ |
| 2 |
|
| 3 |
/* { dg-additional-options "-fopt-info-omp-all" } */ |
| 4 |
|
| 5 |
/* { dg-additional-options "-fdump-tree-gimple" } */ |
| 6 |
|
| 7 |
/* { dg-additional-options "--param=openacc-kernels=decompose" } |
| 8 |
{ dg-additional-options "-fdump-tree-omp_oacc_kernels_decompose" } */ |
| 9 |
|
| 10 |
/* { dg-additional-options "--param=openacc-privatization=noisy" } |
| 11 |
Prune a few: uninteresting, and potentially varying depending on GCC configuration (data types): |
| 12 |
{ dg-prune-output {note: variable 'D\.[0-9]+' declared in block isn't candidate for adjusting OpenACC privatization level: not addressable} } */ |
| 13 |
|
| 14 |
/* { dg-additional-options "-Wopenacc-parallelism" } for testing/documenting |
| 15 |
aspects of that functionality. */ |
| 16 |
|
| 17 |
/* See also '../../gfortran.dg/goacc/kernels-decompose-1.f95'. */ |
| 18 |
|
| 19 |
/* It's only with Tcl 8.5 (released in 2007) that "the variable 'varName' |
| 20 |
passed to 'incr' may be unset, and in that case, it will be set to [...]", |
| 21 |
so to maintain compatibility with earlier Tcl releases, we manually |
| 22 |
initialize counter variables: |
| 23 |
{ dg-line l_dummy[variable c_compute 0 c_loop_i 0] } |
| 24 |
{ dg-message "dummy" "" { target iN-VAl-Id } l_dummy } to avoid |
| 25 |
"WARNING: dg-line var l_dummy defined, but not used". */ |
| 26 |
|
| 27 |
#define N 1024 |
| 28 |
|
| 29 |
unsigned int a[N]; |
| 30 |
|
| 31 |
int |
| 32 |
main (void) |
| 33 |
{ |
| 34 |
int i; |
| 35 |
unsigned int sum = 1; |
| 36 |
|
| 37 |
#pragma acc kernels copyin(a[0:N]) copy(sum) /* { dg-line l_compute[incr c_compute] } */ |
| 38 |
/* { dg-note {variable 'sum\.[0-9]+' declared in block isn't candidate for adjusting OpenACC privatization level: not addressable} {} { target *-*-* } l_compute$c_compute } */ |
| 39 |
/* { dg-bogus {optimized: assigned OpenACC seq loop parallelism} TODO { xfail *-*-* } l_compute$c_compute } |
| 40 |
TODO Is this maybe the report that belongs to the XFAILed report further down? */ |
| 41 |
{ |
| 42 |
#pragma acc loop /* { dg-line l_loop_i[incr c_loop_i] } */ |
| 43 |
/* { dg-note {forwarded loop nest in OpenACC 'kernels' region to 'parloops' for analysis} {} { target *-*-* } 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 |
/* { dg-optimized "assigned OpenACC seq loop parallelism" "" { target *-*-* } l_loop_i$c_loop_i } */ |
| 46 |
for (i = 0; i < N; ++i) |
| 47 |
sum += a[i]; |
| 48 |
|
| 49 |
/* { dg-note {beginning 'gang-single' part in OpenACC 'kernels' region} {} { target *-*-* } .+1 } */ |
| 50 |
sum++; |
| 51 |
a[0]++; |
| 52 |
|
| 53 |
#pragma acc loop independent /* { dg-line l_loop_i[incr c_loop_i] } */ |
| 54 |
/* { dg-note {parallelized loop nest in OpenACC 'kernels' region} {} { target *-*-* } l_loop_i$c_loop_i } */ |
| 55 |
/* { dg-note {variable 'i' in 'private' clause isn't candidate for adjusting OpenACC privatization level: not addressable} {} { target *-*-* } l_loop_i$c_loop_i } */ |
| 56 |
/* { dg-optimized "assigned OpenACC gang vector loop parallelism" "" { target *-*-* } l_loop_i$c_loop_i } */ |
| 57 |
for (i = 0; i < N; ++i) |
| 58 |
sum += a[i]; |
| 59 |
|
| 60 |
/* { dg-note {beginning 'parloops' part in OpenACC 'kernels' region} {} { target *-*-* } .+1 } */ |
| 61 |
if (sum > 10) |
| 62 |
{ |
| 63 |
#pragma acc loop /* { dg-line l_loop_i[incr c_loop_i] } */ |
| 64 |
/* { dg-missed "unparallelized loop nest in OpenACC 'kernels' region: it's executed conditionally" "" { target *-*-* } l_loop_i$c_loop_i } */ |
| 65 |
/* { dg-note {variable 'i' in 'private' clause isn't candidate for adjusting OpenACC privatization level: not addressable} {} { target *-*-* } l_loop_i$c_loop_i } */ |
| 66 |
/*TODO { dg-optimized "assigned OpenACC seq loop parallelism" "TODO" { xfail *-*-* } l_loop_i$c_loop_i } */ |
| 67 |
for (i = 0; i < N; ++i) |
| 68 |
sum += a[i]; |
| 69 |
} |
| 70 |
|
| 71 |
#pragma acc loop auto /* { dg-line l_loop_i[incr c_loop_i] } */ |
| 72 |
/* { dg-note {forwarded loop nest in OpenACC 'kernels' region to 'parloops' for analysis} {} { target *-*-* } l_loop_i$c_loop_i } */ |
| 73 |
/* { dg-note {variable 'i' in 'private' clause isn't candidate for adjusting OpenACC privatization level: not addressable} {} { target *-*-* } l_loop_i$c_loop_i } */ |
| 74 |
/* { dg-optimized "assigned OpenACC seq loop parallelism" "" { target *-*-* } l_loop_i$c_loop_i } */ |
| 75 |
for (i = 0; i < N; ++i) |
| 76 |
sum += a[i]; |
| 77 |
} |
| 78 |
|
| 79 |
return 0; |
| 80 |
} |
| 81 |
|
| 82 |
/* { dg-final { scan-tree-dump-times {(?n)#pragma omp target oacc_kernels map\(tofrom:sum \[len: [0-9]+\]\) map\(to:a\[0\] \[len: [0-9]+\]\) map\(firstprivate:a \[pointer assign, bias: 0\]\)$} 1 "gimple" } } |
| 83 |
|
| 84 |
{ dg-final { scan-tree-dump-times {(?n)#pragma acc loop private\(i\)$} 2 "gimple" } } |
| 85 |
{ dg-final { scan-tree-dump-times {(?n)#pragma acc loop independent private\(i\)$} 1 "gimple" } } |
| 86 |
{ dg-final { scan-tree-dump-times {(?n)#pragma acc loop auto private\(i\)$} 1 "gimple" } } |
| 87 |
{ dg-final { scan-tree-dump-times {(?n)#pragma acc loop} 4 "gimple" } } */ |
| 88 |
|
| 89 |
/* Check that the OpenACC 'kernels' got decomposed into 'data' and an enclosed |
| 90 |
sequence of compute constructs. |
| 91 |
{ dg-final { scan-tree-dump-times {(?n)#pragma omp target oacc_data_kernels map\(tofrom:sum \[len: [0-9]+\]\) map\(to:a\[0\] \[len: [0-9]+\]\)$} 1 "omp_oacc_kernels_decompose" } } |
| 92 |
As noted above, we get three "old-style" kernel regions, one gang-single region, and one parallelized loop region. |
| 93 |
{ dg-final { scan-tree-dump-times {(?n)#pragma omp target oacc_kernels async\(-1\) map\(force_present:sum \[len: [0-9]+\]\) map\(force_present:a\[0\] \[len: [0-9]+\]\) map\(firstprivate:a \[pointer assign, bias: 0\]\)$} 3 "omp_oacc_kernels_decompose" } } |
| 94 |
{ dg-final { scan-tree-dump-times {(?n)#pragma omp target oacc_parallel_kernels_parallelized async\(-1\) map\(force_present:sum \[len: [0-9]+\]\) map\(force_present:a\[0\] \[len: [0-9]+\]\) map\(firstprivate:a \[pointer assign, bias: 0\]\)$} 1 "omp_oacc_kernels_decompose" } } |
| 95 |
{ dg-final { scan-tree-dump-times {(?n)#pragma omp target oacc_parallel_kernels_gang_single async\(-1\) num_gangs\(1\) map\(force_present:sum \[len: [0-9]+\]\) map\(force_present:a\[0\] \[len: [0-9]+\]\) map\(firstprivate:a \[pointer assign, bias: 0\]\)$} 1 "omp_oacc_kernels_decompose" } } |
| 96 |
|
| 97 |
'data' plus five CCs. |
| 98 |
{ dg-final { scan-tree-dump-times {(?n)#pragma omp target } 6 "omp_oacc_kernels_decompose" } } |
| 99 |
|
| 100 |
{ dg-final { scan-tree-dump-times {(?n)#pragma acc loop private\(i\)$} 2 "omp_oacc_kernels_decompose" } } |
| 101 |
{ dg-final { scan-tree-dump-times {(?n)#pragma acc loop independent private\(i\)$} 1 "omp_oacc_kernels_decompose" } } |
| 102 |
{ dg-final { scan-tree-dump-times {(?n)#pragma acc loop auto private\(i\)$} 1 "omp_oacc_kernels_decompose" } } |
| 103 |
{ dg-final { scan-tree-dump-times {(?n)#pragma acc loop} 4 "omp_oacc_kernels_decompose" } } |
| 104 |
|
| 105 |
Each of the parallel regions is async, and there is a final call to |
| 106 |
__builtin_GOACC_wait. |
| 107 |
{ dg-final { scan-tree-dump-times "__builtin_GOACC_wait" 1 "omp_oacc_kernels_decompose" } } */ |