6,728 questions
-1
votes
0
answers
38
views
OpenMP schedule(dynamic) fails in RcppEigen on macOS Sequoia 15.6.1 (symbol not found in flat namespace)
I am running a straightforward parallelized computation using RcppEigen and OpenMP.
Here is a minimal reproducible example of the code:
// [[Rcpp::depends(RcppEigen)]]
// [[Rcpp::plugins(openmp)]]
#...
- 469
2
votes
1
answer
147
views
How is a non-parallelized for loop inside an OpenMP parallel section executed?
Consider the following code:
#pragma omp parallel
for (int run = 0; run < 10; run++)
{
std::vector<int> out;
#pragma omp for
for (int i = 0; i < 1'000'000; i++)
{
...
}
}
...
- 7,515
Best practices
0
votes
3
replies
96
views
Code design conundrum: runtime polymorphism, templates, compile times, and OpenMP
I'm struggling to finalise the design of my C++17 library.
One of the primary goals is to use runtime polymorphism to allow users to extend or rewrite default features of the library for their own use ...
- 196
2
votes
1
answer
117
views
Does the construct `workshare` work with `do-concurrent`?
I'm one of the developers of the Lumen code: https://www.lumen-code.org/.
That is computational code for condensed matter physics simulations.
We are replacing FORALL with DO CONCURRENT, since FORALL ...
- 21
0
votes
1
answer
95
views
C++ segmentation fault when throwing in ordered OMP parallel for
the code below crashes with
terminate called after throwing an instance of 'std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> >'
Aborted ...
- 431
1
vote
1
answer
114
views
OpenMP in C | How to keep private iterable after loop
So i'm working on some homework related to fixing som buggy code when i ran into an interesting problem. I don't think it was one of the indended bugs because the lecturer was confused by it as well. ...
-1
votes
0
answers
34
views
Session aborted in stochastic simulation using Rcpp and OpenMP [duplicate]
Overview - simple actuarial model to perform a stochastic simulation on a group of individuals.
How it works - In a nutshell, the sequential version works as follows: in each of the nRip iterations, ...
- 1
3
votes
2
answers
176
views
OpenMP tasks not working in a Fortran code with asymmetric loads
I am optimising a code already parallelised but not very optimised because of the very different duration of operations among the threads, even though all threads have a similar task to do. So I ...
- 41
0
votes
0
answers
79
views
Segmentation fault in Fortran OpenMP+OpenACC hybrid parallelization
I'm working on porting the CAMB (Cosmological Boltzmann code) to run with hybrid CPU+GPU parallelization using OpenMP and OpenACC with NVIDIA HPC SDK compilers. The code works perfectly when compiled ...
0
votes
0
answers
82
views
Android NDK Crash in libomp.so at __kmpc_barrier with FluidSynth and OpenMP
I'm developing an Android app using the NDK that plays MIDI files with FluidSynth, integrated with a Java frontend.
The app crashes with a native crash in libomp.so at __kmpc_barrier, and I suspect it'...
3
votes
2
answers
130
views
How to do *user defined reduction* on *allocatable array* and *user reduction functions* with openMP in C?
I have written some programs with OMP reduction directive in Fortran and in C. But the data types were simple (int, float, arrays with fixed size, ...) and reduction-identifiers used were implicitly ...
- 2,525
2
votes
3
answers
110
views
Open-MP Parallel for (three-dimensional array)
We are working with the following code:
int i, j, k;
for (i = 2; i < n; i++){ // S1
for (j = 3; j < n - 3; j++){ // S2
for (k = 4; k < n - 4; k++){ // S3
A[...
- 21
1
vote
1
answer
138
views
Fortran OpenMP offloading painfully slow on NVIDIA architectures
I am currently trying to porting a big portion of a Fortran code to GPU devices with OpenMP. I have a working version for AMD, specifically for the MI300A which features unified shared memory. I ...
1
vote
1
answer
166
views
Is there a way to express dependency on the first iteration of the parallelized loop?
My program needs to perform some heavy calculations on all widgets in the box. The calculations are repeated an appreciable number of times processing multiple variations of each widget.
All of the ...
- 4,266
0
votes
0
answers
102
views
Why does my Radix Sort with OpenMP+AVX not sort correctly, while the OpenMP-only version works?
I'm implementing a bitwise Radix Sort to sort an array of 64-bit unsigned integers (key_t is uint_fast64_t) that represent encoded points.
I have two versions of the algorithm:
One using OpenMP only ✅...
- 1
3
votes
1
answer
98
views
OpenMP atomic: Difference between write and update
I am reading the OMP documentation (6.0) about atomic operations.
As I came across clauses I read about write and update.
I understand the difference between the two:
write is used to atomically ...
0
votes
2
answers
133
views
OpenMP target function with gcc compiler
I am working on a project which performs extensive matrix operations and takes advantage of GPU offloading via OpenMP. Things work well for most of the code, except when I try to run some target-...
0
votes
0
answers
78
views
Adding OpenMP reduction clause to loop inside a function
I have a function that contains an OpenMP-parallelized for loop, which calls a callback at each iteration, similar to this:
template<class Callback>
void iterate(const Callback& callback, ...
- 9,230
2
votes
1
answer
88
views
OpenMP behavior of for outside of parallel
I have a function that uses an OpenMP-parallelized for loop in its implementation. It should be possible to turn on/off the parallelization at runtime. Currently it is like this:
void iterate(bool ...
- 9,230
1
vote
0
answers
120
views
Parallelization of a for loop in a C application: is the same iteration held by both threads?
I'm working on a C code application, running on Windows 10; my CPU is an i7-7500u with 2 cores. My C compiler is
anaconda3/Library/bin/gcc.exe
and I use the following flags:
-O2 -march=native -...
- 11
1
vote
2
answers
72
views
How to debug OpenMP scalabilty
I'm trying to implement Conway's Game of Life in C, showing that it can scale with OpenMP. The results in the following picture are from running it on a socket of a machine with 64 cores equally ...
- 13
0
votes
2
answers
44
views
OpenMP system variables and environments
I'm currently dealing with some benchmarks that are highly dependent on OpenMP settings. I noticed that the results can vary greatly from setting values for variables such as
OMP_PLACES
OMP_PROC_BIND
...
8
votes
0
answers
131
views
What is the difference between the OpenMP loop and for constructs?
The OpenMP specification names the worksharing-loop construct with syntax
#pragma omp for
and the loop construct with syntax
#pragma omp loop
The description in the linked pages sounds a bit like ...
- 94.8k
0
votes
1
answer
89
views
LLVM OpenMP GPU Offloading Docker Image
I've been trying to get clang OpenMP GPU offloading working in the Docker Image and it builds fine at first (cmdline:
~/llvm_project/llvm/utils/docker:$ bash build_docker_image.sh
--source nvidia-...
- 123
0
votes
0
answers
109
views
How is a multithreaded code (ex using OpenMP) converted to assembly?
For example, When I multithread a "for loop" using OpenMP, making the iterations as different threads, How does it get translated to Assembly?
Also can a multithreaded code run on hardware ...
3
votes
1
answer
108
views
Does omp parallel for scheduling affect memory access?
I have a found myself most confused by the result of my tests regarding effects of different scheduling for #pragma omp parallel for. Basically my confusion sprouted from the following problem:
I have ...
- 41
0
votes
0
answers
43
views
OpenMP crashes on Android when APK contains large assets files (100MB+), works fine after removing them
I'm encountering a strange issue where OpenMP parallel code crashes on certain Android devices when the APK contains large files (100MB+) in assets or res/raw. The crash disappears after removing ...
- 1
1
vote
2
answers
159
views
Is there a way to “split” an OpenMP synchronization barrier?
I have some code that previously looked like:
std::vector<MyObject> data;
do { // May run 100 times or more
prepare();
remaining = compute1(my_data);
remaining -= compute2(my_data);
...
- 225
0
votes
0
answers
112
views
Why `mul_mat` in ggml slower than llama.cpp?
I use the following command to compile an executable file for Android:
cmake \
-DCMAKE_TOOLCHAIN_FILE=$ANDROID_NDK/build/cmake/android.toolchain.cmake \
-DANDROID_ABI=arm64-v8a \
-...
- 11
0
votes
1
answer
109
views
Strange behavior with pointers in derived types when using OpenMP
I am having trouble identifying a problem in an OpenMP parallelized Fortran code. In doing so, I created a small reproducible that mimics the derived types I am using in the code. The small ...
- 43
2
votes
0
answers
78
views
OpenMP vs OpenMP API
For a school assignment I need a precise definition of OpenMP. I find the Wikipedia and other sources defective.
Specificaly I need to understand what is the difference between OpenMP and OpenMP API.
...
1
vote
0
answers
63
views
Correct usage of multi-dependencies and mutexinoutset in OpenMP
Recent OpenMP standards introduced task multi-dependencies. This feature was originally proposed in OmpSs if I'm not mistaken. I found an example of applying multi-dependencies to the assembly step of ...
- 246
1
vote
1
answer
101
views
OpenMP thread affinity not working in Rcpp package: "Affinity not supported on this configuration"
I'm developing an R package using Rcpp with OpenMP for parallelization. I'm facing an issue where threads are randomly spread across CPU cores, leading often to threads beeing scheduled primarily on ...
- 11
1
vote
1
answer
105
views
OpenMP directive in constexpr functions
According to OpenMP specification (2.1. Directive Format)
Directives may not appear in constexpr functions or in constant expressions. Variadic parameter packs cannot be expanded into a directive or ...
- 2,239
0
votes
0
answers
47
views
Does "pragma omp parallel for" only allow specific syntax in for loops? [duplicate]
In gcc (GCC) 14.2.0 the following program
int main(int argc, char *argv[]) {
#pragma omp parallel for
for (int i{0}; i < 1; ++i) {
}
return 0;
}
does compile with g++ main.cpp (without ...
- 1,647
1
vote
0
answers
55
views
Setting the stack size for nested parallel regions with Open MP (Fortran)
I'm using Open MP with Fortran 08 (compiled with GFortran) to create nested parallel regions (in my case, a 3-level nest). Before running my executable, I set some Linux environment variables to ...
- 11
1
vote
0
answers
94
views
Variable declaration in fortran : Pros and Cons between declaration in a block construct or copy in a private OMP Parallel clause?
I wrote these two programs to test OMP PARALLEL:
1 : my_idx is declared at the beginning of the program and after a private copy for eah thread is done.
program my_pgm
use omp_lib
implicit ...
- 2,525
0
votes
0
answers
46
views
Problem with construction BLOCK and openmp in gfortran
I wrote the following program :
program my_pgm
use omp_lib
implicit none
CALL OMP_set_dynamic(.FALSE.)
CALL OMP_set_num_threads(2)
!$OMP PARALLEL DEFAULT(SHARED)
...
- 2,525
1
vote
1
answer
116
views
Why do OpenMP programs run faster on a single process than on multiple processes?
The task was to implement various matrix multiplication algorithms using OpenMP. It turned out that with num_threads(1), the program runs faster than with any other number of threads. Is this due to ...
2
votes
2
answers
97
views
Standard way of calling math functions in C when using OpenMP & its offloading feature(s)?
I am writing some code in C in which I want to add the optional ability to have certain sections of the code accelerated using OpenMP, and with an additional optional ability to have them accelerated ...
- 124
4
votes
4
answers
428
views
Optimize a separable convolution for SIMD friendly and efficiency
I implemented a simple 2d separable convolution code:
void SeparableConvolution2D(const float * restrict mI, float * restrict mO, float * restrict mB, int numRows, int numCols, const float * restrict ...
- 5,133
0
votes
1
answer
106
views
Getting multiple outputs on Open MPI C++ Code
The following code gives multiple outputs
Sum of first 100000 natural numbers: 5000050000
Sum of first 100000 natural numbers: 5000050000
Sum of first 100000 natural numbers: 5000050000
Sum of ...
2
votes
1
answer
134
views
How to declare an OpenMP reduction for a std::vector inside a struct?
I'm trying to perform an std::vector sum reduction with an OpenMP reduction declaration for it:
// g++ -fopenmp MRE.cpp -o MRE
#include <vector>
#include <algorithm>
#include <omp.h>
...
0
votes
1
answer
53
views
Openmp lastprivate confusing result
I am getting following results roughly 20% of the execution with the given code and lastprivate clause is not working as expected. Pls note that end value of b should be 50 however at times it is not ...
1
vote
0
answers
81
views
loop parallelization with omp
Good morning, I was trying to parallelize a loop with OMP in a program I created for a flood simulator with the goal of achieving a performance boost. The loop I want to parallelize is as follows:
#...
- 11
1
vote
3
answers
173
views
Parallel loop in Fortran using OpenMP
I'm exploring OpenMP in Fortran, using the Mandelbrot algorithm as an example:
!$omp parallel do reduction(+:iters)
do iy=0,30
do ix=0,70
c = cmplx(xmin+stepx*ix, ymin+stepy*iy, qp)
...
- 1,789
0
votes
1
answer
102
views
File writing while compiling with flag -fopenmp
I'm trying to write to file a certain value, calculated with some parallelized operations using Eigen, in particular with the flag -fopenmp.
I've written a code that works perfectly with -g -o, but I'...
6
votes
3
answers
254
views
Using inclusive scan syntax in OpenMP in the C language
I want to use the inclusive scan operation in OpenMP to implement an algorithm. What follows is a description of my attempt at doing so, and failing to get more than a tepid speedup.
The inclusive ...
- 14.9k
1
vote
1
answer
66
views
OpenMP declare reduction: initializer refers to variable which is not 'omp_priv' nor 'omp_orig'
Leaving out non-essential code:
double* max_of_two(double *r,double *n);
int npoints = 1000000;
double* xy = (double*)malloc(2*npoints*sizeof(double));
double zero_coord[2] = {0.0,0.0};
double ...
- 5,990
0
votes
1
answer
42
views
In OpenMP, does explicit barrier synchronize implicit tasks in any ways?
In OpenMP specification 6.0, the barrier construct "specifies an explicit barrier at the point at which the construct appears", and "all threads of the team that executes that binding ...
- 1