a2dbfb6b72
This revision simplifies Clang codegen for parallel regions in OpenMP GPU target offloading and corresponding changes in libomptarget: SPMD/non-SPMD parallel calls are unified under a single `kmpc_parallel_51` runtime entry point for parallel regions (which will be commonized between target, host-side parallel regions), data sharing is internalized to the runtime. Tests have been auto-generated using `update_cc_test_checks.py`. Also, the revision contains changes to OpenMPOpt for remark creation on target offloading regions. Reviewed By: jdoerfert, Meinersbur Differential Revision: https://reviews.llvm.org/D95976
32 lines
944 B
Python
Executable File
32 lines
944 B
Python
Executable File
#!/usr/bin/env python3
|
|
|
|
import argparse
|
|
|
|
def main():
|
|
parser = argparse.ArgumentParser()
|
|
parser.add_argument('--max_args', type=int, help='Max number of arguments to generate case statements for', required=True)
|
|
parser.add_argument('--output', help='Output header file to include', required=True)
|
|
args = parser.parse_args()
|
|
|
|
output=''
|
|
for i in range(args.max_args+1):
|
|
output += 'case %d:\n'%(i)
|
|
output += '((void (*)(kmp_int32 *, kmp_int32 *\n'
|
|
for j in range(i):
|
|
output += ', void *'
|
|
if (j+1)%4 == 0:
|
|
output += '\n'
|
|
output += '))fn)(&global_tid, &bound_tid\n'
|
|
for j in range(i):
|
|
output += ', args[%d]'%(j)
|
|
if (j+1)%4 == 0:
|
|
output += '\n'
|
|
output += ');\n'
|
|
output += 'break;\n'
|
|
|
|
with open(args.output, 'w') as f:
|
|
print(output, file=f)
|
|
|
|
if __name__ == "__main__":
|
|
main()
|