Intel MKL backend issue

[chihaoy]

Gcc version:

gcc (GCC) 11.3.1 20220421 (Red Hat 11.3.1-2)
Copyright (C) 2021 Free Software Foundation, Inc.

Icpx --version:

Intel(R) oneAPI DPC++/C++ Compiler 2025.0.4 (2025.0.4.20241205)
Target: x86_64-unknown-linux-gnu
Thread model: posix
InstalledDir: /opt/intel/oneapi/compiler/2025.0/bin/compiler
Configuration file: /opt/intel/oneapi/compiler/2025.0/bin/compiler/../icpx.cfg

Steps to reproduce the bug:

~/spblas-reference$ source /opt/intel/oneapi/setvars.sh
~/spblas-reference$ CXX=icpx cmake -B build -DENABLE_ONEMKL_SYCL=ON
~/spblas-reference$ cd build/examples/
~/spblas-reference/build/examples$ make
~/spblas-reference/build/examples$ ./simple_spgemm

        #################################################################################
        ### Running SpGEMM Example:
        ###
        ###   C = A * B
        ###
        ### with 
        ### A, in CSR format, of size (100, 100) with nnz = 100
        ### B, in CSR format, of size (100, 10) with nnz = 100
        ### C, in CSR format, of size (100, 10) with nnz to be determined
        ### using float and spblas::index_t (size = 4 bytes)
        #################################################################################
XRT build version: 2.15.0
Build hash: adf27adb3cfadc6e4c41d6db814159f1329b24f3
Build date: 2025-01-10 08:59:56
Git branch: HEAD
PID: 342916
UID: 1785379
[Wed Aug  6 17:04:33 2025 GMT]
HOST: Segmentation fault (core dumped)

I suspect it is the problem with `multiply_compute(scaled(1.f, a), b, c);
After experimenting it, I find everything works when compiled with normal gcc