alexey-bataev
commented
9 years ago Hi, thanks for the report. I'll fix it ASAP.
Best regards,
Alexey Bataev
Software Engineer Intel Compiler Team Intel Corp.
21.11.2014 17:28, Mihail Popov пишет:
Hi,
OpenMP 4.0 specification explains in chapter 2.7.1, Loop Construct, page 54, that there is an implicit barrier at the end of a loop construct unless a nowait clause is specified. Also, chapter 2.14.3.6 Reduction clause, page 170, line 7, specifies that if nowait is not used with a reduction clause, the reduction computation will be complete at the end of the construct.
I tested the source code below:
include
include
int main() { int sum =0;
#pragma omp parallel { int i; #pragma omp for reduction(+:sum) for (i=1;i<5;i++) { sum += i; } printf("before barrier =%d\n",sum); #pragma omp barrier printf("after barrier =%d\n",sum); }}
From my understanding of the specification, I expect to see no difference on the sum value before or after the barrier because there should be an implicit barrier at the end of the loop. However I observe that the sum has a different values before and after.
This issue was observed only with specific thread configurations: the problem appears only if the number of threads is between 2 and 4. We could observe this problem with LLVM 3.4 and 3.5.
Logs of our tests follow:
|Intel(R) Core(TM) i7-3770 CPU @ 3.40GHz (Ivy Bridge) 3.2.0-4-amd64
Using LLVM 3.5 clang version 3.5.0 (https://github.com/clang-omp/clang bf475046fda9b3d9984fabebd3bcbfcf54c2d907) (https://github.com/clang-omp/llvm e45b045553e027cbe400cbb8ac8c264abbbfaf83) Target: x86_64-unknown-linux-gnu Thread model: posix Selected GCC installation: /usr/lib/gcc/x86_64-linux-gnu/4.7 OUTPUT: * 4 threads OMP: Info #204: KMP_AFFINITY: decoding x2APIC ids. OMP: Info #202: KMP_AFFINITY: Affinity capable, using global cpuid leaf 11 info OMP: Info #154: KMP_AFFINITY: Initial OS proc set respected: {0,1,2,3,4,5,6,7} OMP: Info #156: KMP_AFFINITY: 8 available OS procs OMP: Info #157: KMP_AFFINITY: Uniform topology OMP: Info #179: KMP_AFFINITY: 1 packages x 4 cores/pkg x 2 threads/core (4 total cores) OMP: Info #206: KMP_AFFINITY: OS proc to physical thread map: OMP: Info #171: KMP_AFFINITY: OS proc 0 maps to package 0 core 0 thread 0 OMP: Info #171: KMP_AFFINITY: OS proc 4 maps to package 0 core 0 thread 1 OMP: Info #171: KMP_AFFINITY: OS proc 1 maps to package 0 core 1 thread 0 OMP: Info #171: KMP_AFFINITY: OS proc 5 maps to package 0 core 1 thread 1 OMP: Info #171: KMP_AFFINITY: OS proc 2 maps to package 0 core 2 thread 0 OMP: Info #171: KMP_AFFINITY: OS proc 6 maps to package 0 core 2 thread 1 OMP: Info #171: KMP_AFFINITY: OS proc 3 maps to package 0 core 3 thread 0 OMP: Info #171: KMP_AFFINITY: OS proc 7 maps to package 0 core 3 thread 1 OMP: Info #144: KMP_AFFINITY: Threads may migrate across 1 innermost levels of machine OMP: Info #147: KMP_AFFINITY: Internal thread 0 bound to OS proc set {0,4} OMP: Info #147: KMP_AFFINITY: Internal thread 3 bound to OS proc set {3,7} OMP: Info #147: KMP_AFFINITY: Internal thread 1 bound to OS proc set {1,5} OMP: Info #147: KMP_AFFINITY: Internal thread 2 bound to OS proc set {2,6} before barrier = 3 before barrier = 3 before barrier = 6 before barrier = 10 after barrier = 10 after barrier = 10 after barrier = 10 after barrier = 10 * 8 threads OMP: Info #204: KMP_AFFINITY: decoding x2APIC ids. OMP: Info #202: KMP_AFFINITY: Affinity capable, using global cpuid leaf 11 info OMP: Info #154: KMP_AFFINITY: Initial OS proc set respected: {0,1,2,3,4,5,6,7} OMP: Info #156: KMP_AFFINITY: 8 available OS procs OMP: Info #157: KMP_AFFINITY: Uniform topology OMP: Info #179: KMP_AFFINITY: 1 packages x 4 cores/pkg x 2 threads/core (4 total cores) OMP: Info #206: KMP_AFFINITY: OS proc to physical thread map: OMP: Info #171: KMP_AFFINITY: OS proc 0 maps to package 0 core 0 thread 0 OMP: Info #171: KMP_AFFINITY: OS proc 4 maps to package 0 core 0 thread 1 OMP: Info #171: KMP_AFFINITY: OS proc 1 maps to package 0 core 1 thread 0 OMP: Info #171: KMP_AFFINITY: OS proc 5 maps to package 0 core 1 thread 1 OMP: Info #171: KMP_AFFINITY: OS proc 2 maps to package 0 core 2 thread 0 OMP: Info #171: KMP_AFFINITY: OS proc 6 maps to package 0 core 2 thread 1 OMP: Info #171: KMP_AFFINITY: OS proc 3 maps to package 0 core 3 thread 0 OMP: Info #171: KMP_AFFINITY: OS proc 7 maps to package 0 core 3 thread 1 OMP: Info #144: KMP_AFFINITY: Threads may migrate across 1 innermost levels of machine OMP: Info #147: KMP_AFFINITY: Internal thread 0 bound to OS proc set {0,4} OMP: Info #147: KMP_AFFINITY: Internal thread 1 bound to OS proc set {1,5} OMP: Info #147: KMP_AFFINITY: Internal thread 2 bound to OS proc set {2,6} OMP: Info #147: KMP_AFFINITY: Internal thread 3 bound to OS proc set {3,7} OMP: Info #147: KMP_AFFINITY: Internal thread 4 bound to OS proc set {0,4} OMP: Info #147: KMP_AFFINITY: Internal thread 5 bound to OS proc set {1,5} OMP: Info #147: KMP_AFFINITY: Internal thread 6 bound to OS proc set {2,6} OMP: Info #147: KMP_AFFINITY: Internal thread 7 bound to OS proc set {3,7} before barrier = 10 before barrier = 10 before barrier = 10 before barrier = 10 before barrier = 10 before barrier = 10 before barrier = 10 before barrier = 10 after barrier = 10 after barrier = 10 after barrier = 10 after barrier = 10 after barrier = 10 after barrier = 10 after barrier = 10 after barrier = 10 It works as expected Using LLVM 3.4 clang version 3.4 (https://github.com/clang-omp/clang bdc8bcdb94f2b954059d4ce1b5762f701ecc0809) (https://github.com/clang-omp/llvm 92414b1167a33c0ab9187c72b098a54ecbffc15c) Target: x86_64-unknown-linux-gnu Thread model: posix Selected GCC installation: /usr/lib/gcc/x86_64-linux-gnu/4.7 OUTPUT: * 4 threads OMP: Info #204: KMP_AFFINITY: decoding x2APIC ids. OMP: Info #202: KMP_AFFINITY: Affinity capable, using global cpuid leaf 11 info OMP: Info #154: KMP_AFFINITY: Initial OS proc set respected: {0,1,2,3,4,5,6,7} OMP: Info #156: KMP_AFFINITY: 8 available OS procs OMP: Info #157: KMP_AFFINITY: Uniform topology OMP: Info #179: KMP_AFFINITY: 1 packages x 4 cores/pkg x 2 threads/core (4 total cores) OMP: Info #206: KMP_AFFINITY: OS proc to physical thread map: OMP: Info #171: KMP_AFFINITY: OS proc 0 maps to package 0 core 0 thread 0 OMP: Info #171: KMP_AFFINITY: OS proc 4 maps to package 0 core 0 thread 1 OMP: Info #171: KMP_AFFINITY: OS proc 1 maps to package 0 core 1 thread 0 OMP: Info #171: KMP_AFFINITY: OS proc 5 maps to package 0 core 1 thread 1 OMP: Info #171: KMP_AFFINITY: OS proc 2 maps to package 0 core 2 thread 0 OMP: Info #171: KMP_AFFINITY: OS proc 6 maps to package 0 core 2 thread 1 OMP: Info #171: KMP_AFFINITY: OS proc 3 maps to package 0 core 3 thread 0 OMP: Info #171: KMP_AFFINITY: OS proc 7 maps to package 0 core 3 thread 1 OMP: Info #144: KMP_AFFINITY: Threads may migrate across 1 innermost levels of machine OMP: Info #147: KMP_AFFINITY: Internal thread 0 bound to OS proc set {0,4} OMP: Info #147: KMP_AFFINITY: Internal thread 2 bound to OS proc set {2,6} OMP: Info #147: KMP_AFFINITY: Internal thread 1 bound to OS proc set {1,5} OMP: Info #147: KMP_AFFINITY: Internal thread 3 bound to OS proc set {3,7} before barrier = 4 before barrier = 10 before barrier = 5 before barrier = 10 after barrier = 10 after barrier = 10 after barrier = 10 after barrier = 10 * 8 threads OMP: Info #204: KMP_AFFINITY: decoding x2APIC ids. OMP: Info #202: KMP_AFFINITY: Affinity capable, using global cpuid leaf 11 info OMP: Info #154: KMP_AFFINITY: Initial OS proc set respected: {0,1,2,3,4,5,6,7} OMP: Info #156: KMP_AFFINITY: 8 available OS procs OMP: Info #157: KMP_AFFINITY: Uniform topology OMP: Info #179: KMP_AFFINITY: 1 packages x 4 cores/pkg x 2 threads/core (4 total cores) OMP: Info #206: KMP_AFFINITY: OS proc to physical thread map: OMP: Info #171: KMP_AFFINITY: OS proc 0 maps to package 0 core 0 thread 0 OMP: Info #171: KMP_AFFINITY: OS proc 4 maps to package 0 core 0 thread 1 OMP: Info #171: KMP_AFFINITY: OS proc 1 maps to package 0 core 1 thread 0 OMP: Info #171: KMP_AFFINITY: OS proc 5 maps to package 0 core 1 thread 1 OMP: Info #171: KMP_AFFINITY: OS proc 2 maps to package 0 core 2 thread 0 OMP: Info #171: KMP_AFFINITY: OS proc 6 maps to package 0 core 2 thread 1 OMP: Info #171: KMP_AFFINITY: OS proc 3 maps to package 0 core 3 thread 0 OMP: Info #171: KMP_AFFINITY: OS proc 7 maps to package 0 core 3 thread 1 OMP: Info #144: KMP_AFFINITY: Threads may migrate across 1 innermost levels of machine OMP: Info #147: KMP_AFFINITY: Internal thread 0 bound to OS proc set {0,4} OMP: Info #147: KMP_AFFINITY: Internal thread 1 bound to OS proc set {1,5} OMP: Info #147: KMP_AFFINITY: Internal thread 2 bound to OS proc set {2,6} OMP: Info #147: KMP_AFFINITY: Internal thread 3 bound to OS proc set {3,7} OMP: Info #147: KMP_AFFINITY: Internal thread 4 bound to OS proc set {0,4} OMP: Info #147: KMP_AFFINITY: Internal thread 5 bound to OS proc set {1,5} OMP: Info #147: KMP_AFFINITY: Internal thread 6 bound to OS proc set {2,6} OMP: Info #147: KMP_AFFINITY: Internal thread 7 bound to OS proc set {3,7} before barrier = 10 before barrier = 10 before barrier = 10 before barrier = 10 before barrier = 10 before barrier = 10 before barrier = 10 before barrier = 10 after barrier = 10 after barrier = 10 after barrier = 10 after barrier = 10 after barrier = 10 after barrier = 10 after barrier = 10 after barrier = 10 It works as expectedIntel(R) Xeon(R) CPU E5620 @ 2.40GHz (Nehalem) 3.2.0-4-amd64 Using LLVM 3.4 clang version 3.4 (https://github.com/clang-omp/clang a65de2d19de69dbd544d748cf31372f1cdceac8b) (https://github.com/clang-omp/llvm 233b1e3f0347e8946e336812763849b896cc2300) Target: x86_64-unknown-linux-gnu Thread model: posix Selected GCC installation: /usr/lib/gcc/x86_64-linux-gnu/4.7
Genuine Intel(R) CPU @ 2.70GHz (Sandy Bridge) 3.14.0 Using LLVM 3.4 clang version 3.4 (https://github.com/clang-omp/clang b340ac07c932acc9d99ccc2b291561cc04b8554f) (https://github.com/clang-omp/llvm 233b1e3f0347e8946e336812763849b896cc2300) Target: x86_64-unknown-linux-gnu Thread model: posix Selected GCC installation: /usr/lib/gcc/x86_64-linux-gnu/4.7
Intel(R) Core(TM)2 Duo CPU E7500 @ 2.93GHz (Core2) 3.2.0-4-amd64 Using LLVM 3.4 clang version 3.4 (https://github.com/clang-omp/clang bdc8bcdb94f2b954059d4ce1b5762f701ecc0809) (https://github.com/clang-omp/llvm 92414b1167a33c0ab9187c72b098a54ecbffc15c) Target: x86_64-unknown-linux-gnu Thread model: posix Selected GCC installation: /usr/lib/gcc/x86_64-linux-gnu/4.7 |
— Reply to this email directly or view it on GitHub https://github.com/clang-omp/clang/issues/52.
mihailpopov
Hi,
OpenMP 4.0 specification explains in chapter 2.7.1, Loop Construct, page 54, that there is an implicit barrier at the end of a loop construct unless a nowait clause is specified. Also, chapter 2.14.3.6 Reduction clause, page 170, line 7, specifies that if nowait is not used with a reduction clause, the reduction computation will be complete at the end of the construct.
I tested the source code below:
From my understanding of the specification, I expect to see no difference on the sum value before or after the barrier because there should be an implicit barrier at the end of the loop. However I observe that the sum has a different values before and after.
This issue was observed only with specific thread configurations: the problem appears only if the number of threads is between 2 and 4. We could observe this problem with LLVM 3.4 and 3.5.
Logs of our tests follow: