topic complex-valued scalar products in Intel® oneAPI Math Kernel Library

complex-valued scalar products

Bollhöfer__Matthias — Mon, 09 Sep 2019 06:26:25 GMT

Is there any problem calling BLAS functions zdotc, zdotu in recent version of MKL?

When calling these functions from a C-code using recent version of the MKL, the code crashes or returns fault values. It works properly when using soft-coded FORTRAN BLAS. I am using LINUX, I tried three different versions of the MKL (2018.5.274, 2019.1.144 und 2019.3.199).

Here is how I call the MKL

cc -O -fPIC -fopenmp -m64 -mcmodel=medium -I $MKLROOT/include test.c -L $MKLROOT/lib -lmkl_intel_lp64 -lmkl_intel_thread -lmkl_core -liomp5 -lm

Here is a sample C-code

#include <stdio.h>
#include <stdlib.h>

typedef struct { double r, i; } doublecomplex;

doublecomplex zdotc_(int *,doublecomplex *,int *, doublecomplex *,int *),
zdotu_(int *,doublecomplex *,int *, doublecomplex *,int *);

#define N 4

int main(int argc, char **argv)
{
doublecomplex *v, *w, val;
int i=N,j=1,k=2,l,m;
v=(doublecomplex *)malloc((size_t)N *sizeof(doublecomplex));
w=(doublecomplex *)malloc((size_t)2*N*sizeof(doublecomplex));

for (l=0; l<N; l++) {
v.r = 1.0; v.i =(double)l;

w.r =-1.0; w.i = 1.0;
w[N+l].r= 1.0; w[N+l].i= 0.0;
}

val=zdotc_(&i, v,&j, w,&k);
printf("val=(%8.1le,%8.1le)\n",val.r,val.i);

val=zdotu_(&i, v,&j, w,&k);
printf("val=(%8.1le,%8.1le)\n",val.r,val.i);

free(v);
free(w);

return 0;
}

I rewrote this case by

Gennady_F_Intel — Wed, 18 Sep 2019 04:49:36 GMT

I rewrote this case by applying recommended MKL_COMPLEX16 datatypes and I don't see the problem on my side. the example (test_zdot.cpp) is attached to this tread.

The example has been build

Gennady_F_Intel — Wed, 18 Sep 2019 04:54:09 GMT

The example has been build against mkl_threading, lp64 modes and then tested on AVX, AVX2 and AVX-512 based systems with MKL_VERBOSE mode enabled:

AVX - 512:

MKL_VERBOSE Intel(R) MKL 2019.0 Update 5 Product build 20190808 for Intel(R) 64 architecture Intel(R) Advanced Vector Extensions 512 (Intel(R) AVX-512) enabled processors, Lnx 2.20GHz lp64 intel_thread
MKL_VERBOSE ZDOTC(0x7ffef279cc00,4,0x24e80c0,1,0x24f92c0,2) 162.29ms CNR:OFF Dyn:1 FastMM:1 TID:0 NThr:28
zdotc : val=( 1.0e+00,-2.0e+00)
MKL_VERBOSE ZDOTU(0x7ffef279cc00,4,0x24e80c0,1,0x24f92c0,2) 17.16us CNR:OFF Dyn:1 FastMM:1 TID:0 NThr:28
zdotu : val=(-1.0e+00, 6.0e+00)

AVX2:

./test_zdot.out
MKL_VERBOSE Intel(R) MKL 2019.0 Update 5 Product build 20190808 for Intel(R) 64 architecture Intel(R) Advanced Vector Extensions 2 (Intel(R) AVX2) enabled processors, Lnx 2.20GHz lp64 intel_thread
MKL_VERBOSE ZDOTC(0x7ffd3aefbf80,4,0x1d560c0,1,0x1d672c0,2) 2.23ms CNR:OFF Dyn:1 FastMM:1 TID:0 NThr:44
zdotc : val=( 1.0e+00,-2.0e+00)
MKL_VERBOSE ZDOTU(0x7ffd3aefbf80,4,0x1d560c0,1,0x1d672c0,2) 7.40us CNR:OFF Dyn:1 FastMM:1 TID:0 NThr:44
zdotu : val=(-1.0e+00, 6.0e+00)

AVX:

./test_zdot.out
MKL_VERBOSE Intel(R) MKL 2019.0 Update 5 Product build 20190808 for Intel(R) 64 architecture Intel(R) Advanced Vector Extensions (Intel(R) AVX) enabled processors, Lnx 2.80GHz lp64 intel_thread
MKL_VERBOSE ZDOTC(0x7ffe7aec0e00,4,0xca40c0,1,0xcb52c0,2) 2.79ms CNR:OFF Dyn:1 FastMM:1 TID:0 NThr:20
zdotc : val=( 1.0e+00,-2.0e+00)
MKL_VERBOSE ZDOTU(0x7ffe7aec0e00,4,0xca40c0,1,0xcb52c0,2) 13.45us CNR:OFF Dyn:1 FastMM:1 TID:0 NThr:20
zdotu : val=(-1.0e+00, 6.0e+00)

forget to add: applying with

Gennady_F_Intel — Wed, 18 Sep 2019 05:08:18 GMT

forget to add: applying with compiler's options you use, didn't chanched changed nothing:

icc -O -fPIC -fopenmp -m64 -mcmodel=medium -mkl test_zdot.cpp -o test_zdot.out
./test_zdot.out
MKL_VERBOSE Intel(R) MKL 2019.0 Update 5 Product build 20190808 for Intel(R) 64 architecture Intel(R) Advanced Vector Extensions 512 (Intel(R) AVX-512) enabled processors, Lnx 2.20GHz lp64 intel_thread
MKL_VERBOSE ZDOTC(0x7ffe42d70b80,4,0x1b000c0,1,0x1b112c0,2) 2.87ms CNR:OFF Dyn:1 FastMM:1 TID:0 NThr:28
zdotc : val=( 1.0e+00,-2.0e+00)
MKL_VERBOSE ZDOTU(0x7ffe42d70b80,4,0x1b000c0,1,0x1b112c0,2) 18.40us CNR:OFF Dyn:1 FastMM:1 TID:0 NThr:28
zdotu : val=(-1.0e+00, 6.0e+00)