Run C function with CUDA calls in Delphi program - c

My objective is to have a Delphi( or freepascal) code, that will call the C function func like this one:
The C/Cuda file:
/* this is the "progcuda.cu" file */
#include <stdio.h>
__global__ void foo(int *a, int *b, int *c, int n){
/*
add all the vector's element
*/
}
void func(int *a, int *b, int *c,int n){
int *da,*db,*dc;
cudaMalloc(&da, n*sizeof(int));
cudaMalloc(&db, n*sizeof(int));
cudaMalloc(&dc, n*sizeof(int));
cudaMemcpy(da,a,sizeof(int)*n,cudaMemcpyHostToDevice);
cudaMemcpy(db,b,sizeof(int)*n,cudaMemcpyHostToDevice);
cudaMemcpy(dc,c,sizeof(int)*n,cudaMemcpyHostToDevice);
foo<<<1,256>>>(da,db,dc);
cudaMemcpy(c,dc,sizeof(int),cudaMemcpyDeviceToHost);
/* do other stuff and call another Host and Device functions*/
return;
}
The pascal main file:
// this is the "progpas.pas" file
program progpas;
{$mode objfpc}{$H+}
uses unitpas;
var
...
begin
...
func(a, b, c, len);
...
end.
The pascal unit file:
// this is the "unitpas.pas" file
unit unitpas;
{$link progcuda.o}
interface
uses ctypes;
procedure func(a, b, c : cpint32 , n:cint32); cdecl; external;
procedure foo(a, b, c : cpint32 , n:cint32);cdecl; external;
implementation
end.
I've found this post Programming CUDA using Delphi or FreePascal
, but it shows more a way to program CUDA in delphi.
I don't want to program CUDA in Delphi, I want to program in CUDA in pure C/C++ code and only call that C function in delphi.
What is the problem?
How can I link the .cu code to the delphi one?
I'm using linux ubuntu 16.04 LTS, but I also have CUDA and VS in windows if necessary.
Note: if you guys could explain in detail how to do it, would help ( new to pascal and linking files )
I've already tried to generate the .o object file and link it in free pascal with
$ nvcc progcuda.cu -c -o progcuda.o then $fpc progpas.pas
but it fails at linking.
Note: I've tried once to link a normal .o generated by C code to pascal code, using gcc and freepascal compiler, and it worked, but if I use nvcc instead of gcc and rename the extension to .cu ( still same code), the linking fails.
note: new account in stack overflow, i cannot repply answers yet.

I don't know anything about Delphi and FreePascal, but I do know about CUDA, C and C++, so maybe my solution will also work for you.
I'll be demonstrating it with a simple problem:
Content of f.cu:
int f() { return 42; }
Content of main.c:
extern int f();
int main() {
return f();
}
The following works:
$ gcc -c -xc f.cu # need -xc to tell gcc it's a C file
$ gcc main.c f.o
(no errors emitted)
Now when we try replacing gcc with nvcc:
$ nvcc -c f.cu
$ gcc main.c f.o
/tmp/ccI3tBM1.o: In function `main':
main.c:(.text+0xa): undefined reference to `f'
f.o: In function `__cudaUnregisterBinaryUtil()':
tmpxft_0000704e_00000000-5_f.cudafe1.cpp:(.text+0x52): undefined reference to `__cudaUnregisterFatBinary'
f.o: In function `__nv_init_managed_rt_with_module(void**)':
tmpxft_0000704e_00000000-5_f.cudafe1.cpp:(.text+0x6d): undefined reference to `__cudaInitModule'
f.o: In function `__sti____cudaRegisterAll()':
tmpxft_0000704e_00000000-5_f.cudafe1.cpp:(.text+0xa9): undefined reference to `__cudaRegisterFatBinary'
collect2: error: ld returned 1 exit status
The problem here is that nvcc adds references to some symbols from the CUDA runtime API when compiling f.cu, and these symbols have to be linked to the final executable. My CUDA installation is in /opt/cuda, so I will use that, but you have to replace it with wherever CUDA is installed on your system. So if we link libcudart.so when compiling the library we get:
$ nvcc -c f.cu
$ gcc main.c f.o -L/opt/cuda/lib64 -lcudart
/tmp/ccUeDZcb.o: In function `main':
main.c:(.text+0xa): undefined reference to `f'
collect2: error: ld returned 1 exit status
This looks better, no strange errors, but it's still not finding the function f. That's because nvcc is treating f.cu as a C++ file, so it does name mangling when creating the object file, and we have to specify that we want f to have C, and not C++ linkage (see more here: http://en.cppreference.com/w/cpp/language/language_linkage).
To do that we have to modify f.cu like this:
extern "C" int f() { return 42; }
Now when we do:
$ nvcc -c f.cu
$ gcc main.c f.o -L/opt/cuda/lib64 -lcudart
(no errors emitted)
I hope you manage to modify this to work with your language.
EDIT: I tried a bit more complicated example:
// f.cu
#include <stdio.h>
__global__ void kernel() {
printf("Running kernel\n");
}
extern "C" void f() {
kernel<<<1, 1>>>();
// make sure the kernel completes before exiting
cudaDeviceSynchronize();
}
// main.c
extern void f();
int main() {
f();
return 0;
}
When compiling it I got:
f.o:(.data.DW.ref.__gxx_personality_v0[DW.ref.__gxx_personality_v0]+0x0): undefined reference to `__gxx_personality_v0'
collect2: error: ld returned 1 exit status
To fix it you also need to add the standard C++ libraries to the linker flags:
$ nvcc -c f.cu
$ gcc main.c f.o -L/opt/cuda/lib64 -lcudart -lstdc++
$ ./a.out
Running kernel

I fixed the files as #Goran Flegar explained:
Add extern "C" int func(...); to the .cu file. And then tried to compile/link the .cu code, but with no device calls (yet with device code), and all worked well.
but when i add a device call ( foo<<<Nb,Nt>>>(...) ) and compile with:
$nvcc progcuda.cu -c
$fpc progpas.pas -ofinal.exe -Fl/usr/local/cuda/lib64
i get:
Free Pascal Compiler version 3.0.4 [2017/12/13] for x86_64
Copyright (c) 1993-2017 by Florian Klaempfl and others
Target OS: Linux for x86-64
Compiling prog1.pas
Linking sum.exe
/usr/bin/ld: aviso: link.res contém seções de saída; você se esqueceu -T?
/usr/bin/ld: sum.o: undefined reference to symbol '_Unwind_Resume##GCC_3.0'
//lib/x86_64-linux-gnu/libgcc_s.so.1: error adding symbols: DSO missing from command line
prog1.pas(16,1) Error: Error while linking
prog1.pas(16,1) Fatal: There were 1 errors compiling module, stopping
Fatal: Compilation aborted
Error: /usr/bin/ppcx64 returned an error exitcode
So there's still some missing libs.
Solution:
Found that linking the stdc++ and gcc_s lib to pascal solved the compilation problem.
unit unitpas;
// file "unitpas.pas"
{$LINK progcuda.o}
{$LINKLIB c}
{$LINKLIB cudart}
{$linklib stdc++}
{$linklib gcc_s}
interface
uses ctypes;
function func(x,y: cint32): cint32; cdecl; external;
implementation
end.
Run
$nvcc progcuda.cu -c
$fpc progpas.pas -ofinal.exe -Fl/usr/local/cuda/lib64
and everything works.

Related

How can I call a specific function at program start using MinGW compiler? [duplicate]

How to change the entry point of a C program compiled with gcc ?
Just like in the following code
#include<stdio.h>
int entry() //entry is the entry point instead of main
{
return 0;
}
It's a linker setting:
-Wl,-eentry
the -Wl,... thing passes arguments to the linker, and the linker takes a -e argument to set the entry function
You can modify your source code as:
#include<stdio.h>
const char my_interp[] __attribute__((section(".interp"))) = "/lib/ld-linux.so.2";
int entry() //entry is the entry point instead of main
{
exit(0);
}
The ".interp" section will let your program able to call external shared library.
The exit call will make your entry function to exit program instead of return.
Then build the program as a shared library which is executable:
$ gcc -shared -fPIC -e entry test_main.c -o test_main.so
$ ./test_main
If you are on a system that provides GNU Binutils (like Linux),
you can use the objcopy command
to make an arbitrary function the new entry point.
Suppose a file called program.c containing the entry function:
$ cat > program.c
#include <stdio.h>
int entry()
{
return 0;
}
^D
You first compile it using -c to generate a relocatable object file:
$ gcc -c program.c -o program.o
Then you redefine entry to be main:
$ objcopy --redefine-sym entry=main program.o
Now use gcc to compile the new object file:
$ gcc program.o -o program
NOTE: If your program already has a function called main, before step 2, you can perform a separate objcopy invocation:
objcopy --redefine-sym oldmain=main program.o
Minimal runnable example and notes on other answers
main.c
#include <stdio.h>
#include <stdlib.h>
int mymain(void) {
puts("hello");
exit(0);
}
compile and run:
gcc -nostartfiles -Wl,--entry=mymain -o main.out main.c
# or -Wl,-emymain
./main.out 1 2 3
The notes:
without -nostartfiles, the link fails with:
/usr/lib/gcc/x86_64-linux-gnu/7/../../../x86_64-linux-gnu/Scrt1.o: In function `_start':
(.text+0x20): undefined reference to `main'
collect2: error: ld returned 1 exit status
presumably because the glibc setup code that runs before main in _start normally calls main.
command line arguments are not setup for you, presumably because they would be setup by the glibc code that runs before main, so trying to use them prints undefined values. I haven't found a method that works for them.
Tested in Ubuntu 20.10.

undefined reference error for linking CUDA static or shared library with gcc

gcc and CUDA question
Hi,
I have compiled a CUDA shared library but can't link it with the main program that uses it. I am compiling the main program with gcc.
The code:
simplemain.c
#include <stdio.h>
#include <stdlib.h>
void fcudadriver();
int main()
{
printf("Main \n");
fcudadriver();
return 0;
}
test.cu
__global__ void fcuda()
{
}
void fcudadriver()
{
fcuda<<<1,1>>>();
}
I compile test.cu as --> It works
nvcc --compiler-options '-fPIC' -o libtest.so --shared test.cu
I compile simplemain.c as ---> It gives error :(
gcc simplemain.c -L. -ltest
/tmp/ccHnB4Vh.o:simplemain.c:function main: error: undefined reference to 'fcudadriver'
collect2: ld returned 1 exit status
try using g++ instead of gcc. nvcc uses c++ style linking conventions. (You don't need to rename any files.)
alternatively, if you must use gcc, preface your void fcudadriver() function definition like this:
extern "C" void fcudadriver()
C and C++ name the functions in different way.
Since nvcc treat the CPU code in .cu file as C++, you could rename your simplemain.c to simplemain.cpp, and compile it with g++
Another solution could be adding extern "C" before the function definition in the .cu file.

Cuda C - Linker error - undefined reference

I am having a hard time compiling a simple cuda program consiting of only two files.
The main.c looks like this:
#include "my_cuda.h"
int main(int argc, char** argv){
dummy_gpu();
}
The cuda.h looks like this:
#ifndef MY_DUMMY
#define MY_DUMMY
void dummy_gpu();
#endif
And the my_cuda.cu file loos like this:
#include <cuda_runtime.h>
#include "my_cuda.h"
__global__ void dummy_gpu_kernel(){
//do something
}
void dummy_gpu(){
dummy_gpu_kernel<<<128,128>>>();
}
However if I compile I allways receive the following error:
gcc -I/usr/local/cuda/5.0.35/include/ -c main.c
nvcc -c my_cuda.cu
gcc -L/usr/local_rwth/sw/cuda/5.0.35/lib64 -lcuda -lcudart -o md.exe main.o my_cuda.o
main.o: In function `main':
main.c:(.text+0x15): undefined reference to `dummy_gpu'
collect2: ld returned 1 exit status
Thank you for your help.
You have a problem with symbol name mangling. nvcc uses the host C++ compiler to compile host code, and this implies that symbol name mangling is applied to code emitted by the CUDA toolchain.
There are two solutions to this problem. The first is to define dummy_gpu using C linkage, so change your my_cuda.cu to something like this:
extern "C" {
#include "my_cuda.h"
}
.....
extern "C"
void dummy_gpu(){
dummy_gpu_kernel<<<128,128>>>();
}
Note that you will need to change your linkage command to this:
gcc -L/usr/local_rwth/sw/cuda/5.0.35/lib64 -o md.exe main.o my_cuda.o -lcuda -lcudart
because the CUDA shared libraries need to be specified after the object files that use them.
Your second alternative would be to use either g++ or nvcc to do the linking, in which case the whole problem should disappear.
You have a C/C++ linkage problem. nvcc is decorating things in a C++ fashion but your gcc compiler is handling things using C style linkage. A simple way to fix it is to rename your main.c to main.cpp and then repeat your commands using g++ instead of gcc

Cuda mixed C project linking

I have a large project in C and i'm trying to integrate some Cuda kernels in it. I'm compiling my c-files with "gcc -c main.c" and my .cu files with "nvcc -c cuda_GMRES.cu" and then I try to link the 2 object files with nvcc: "nvcc -o main.o cuda_GMRES.o" and receive the following error:
/usr/lib/gcc/x86_64-redhat-linux/4.1.2/../../../../lib64/crt1.o: In function
_start':
(.text+0x20): undefined reference tomain'
collect2: ld returned 1 exit status
It's the first time I'm trying to combine cuda with C files and I might have done something wrong.Can someone help me please. I'm on a GPU Cluster with Rocks OS.
My main.c file:
#include <stdio.h>
#include <math.h>
#include "cuda_wrapper.h" //header containing wrapper function
//cuda_GMRES that calls the kernel cuda_dot
int main (int argc,char* argv[])
{
//content
//bla bla bla
//cuda Function call
cuda_GMRES(50);
return 0;
}
My cuda_wrapper.h file:
#ifndef Cuda_GMRES_cuda_wrapper_h
#define Cuda_GMRES_cuda_wrapper_h
//wrapper function declaration
void cuda_GMRES(double a);
#endif
My cuda_GMRES.cu file that contains the kernel calling function:
#include <stdio.h>
#include "cuda_wrapper.h"
#include "cuda_dot.cu"
//kernel declaration
__global__ void cuda_dot();
//kernel calling function
extern "C"
void cuda_GMRES(double a)
{
double b;
double *dev_a;
double *res;
cudaMemcpy(dev_a, &a, sizeof(double), cudaMemcpyHostToDevice );
cuda_dot<<< 1, 1 >>>(*dev_a, res );
cudaMemcpy(&b, res, sizeof(double), cudaMemcpyDeviceToHost );
}
My cuda_dot.cu file that contains the kernel:
__global__ void cuda_dot(double a, double *help)
{
*help=2*a;
}
Your linking command appears to contain a fatal error. Supposing you first compile two objects like this:
gcc -c main.c
nvcc -c cuda_GMRES.cu
you should have two object files main.o and cuda_GMRES.o. You then do this:
nvcc -o main.o cuda_GMRES.o
This command says "link a program file called main.o using cuda_GMRES.o", ie. overwrite main.o. It is for this reason that the linker is complaining about a missing main subroutine, you are not supplying one (and you are destroying the object file which contains one at the same time).
You want something like this:
nvcc -o executable main.o cuda_GMRES.o
where executable is the name of the final linked program, or
nvcc main.o cuda_GMRES.o
which will emit a default linked program called a.out

Error when compiling with GCC

Every time I compile I get the following error message:
Undefined reference to ( function name )
Let's say I have three files: Main.c, printhello.h, printhello.c. Main.c calls function print_hello(), which returns "Hello World". The function is defined in printhello.c.
Now, here's the following code of printhello.h:
#ifndef PRINTHELLO_H
#define PRINTHELLO_H
void print_hello();
#endif
I am sure this code is fine. I still don't know why is it giving me the error, though. Can you help me?
Undefined references are the linker errors. Are you compiling and linking all the source files ? Since the main.c calls print_hello(), linker should see the definition of it.
gcc Main.c printhello.c -o a.out
The error is, I think, a linker error rather than a compiler error; it is trying to tell you that you've not provided all the functions that are needed to make a complete program.
You need to compile the program like this:
gcc -o printhello Main.c printhello.c
This assumes that your file Main.c is something like:
#include "printhello.h"
int main(void)
{
print_hello();
return 0;
}
and that your file printhello.c is something like:
#include "printhello.h"
#include <stdio.h>
void print_hello(void)
{
puts("Hello World");
}
Your declaration in printhello.h should be:
void print_hello(void);
This explicitly says that the function takes no parameters. The declaration with the empty brackets means "there is a function print_hello() which returns no value and takes an indeterminate (but not variadic) list of arguments", which is quite different. In particular, you could call print_hello() with any number of arguments and the compiler could not reject the program.
Note that C++ treats the empty argument list the same as void print_hello(void); (so it would ensure that calls to print_hello() include no arguments), but C++ is not the same as C.
Another way to do it is to explicitly build object files for the printhello:
gcc -c printhello.c -o printhello.o
gcc -o Main main.c printhello.o
This has the added benefit of allowing other programs to use the print_hello method
It seems that the error is from the linker and not the compiler. You need to compile and link both the source files. I think what you are doing is simply including the header file in Main.c and you are not compiling the printhello.c
You need to :
gcc Main.c printhello.c -o myprog
or
construct the object files first
gcc -c printhello.c
gcc -c Main.c
then link them
gcc Main.o printhello.o

Resources