4 Replies Latest reply on Dec 18, 2014 11:11 PM by oasanov

    Do you plan to keep AMD Core Math Library (ACML) for Linux built with Intel® FORTRAN?

    oasanov

      Hello all,

       

      I'm very interesting to use the latest ACML version (6.1.x) for Linux with Intel FORTRAN support.

      Can I expect such release or Intel FORTRAN support is dropped on Linux?

       

      Other question is about crashes on an Intel Haswell CPU. Does it fixed in ACML 6.1.0?

       

      Best Regards,

      Oleg

        • Re: Do you plan to keep AMD Core Math Library (ACML) for Linux built with Intel® FORTRAN?
          bragadeesh

          Hi Oleg,

           

          I assume you are using the Fortran interfaces on Linux? Unfortunately, we decided to drop support for Intel fortran builds on Linux to reduce the number of configurations we needed to support. But we are evaluating this and we may turn it back on in the future. If you are using C interfaces, you could use the gfortran builds. Let us know if you have any other feedback.

           

          Thanks,

          • Re: Do you plan to keep AMD Core Math Library (ACML) for Linux built with Intel® FORTRAN?
            bragadeesh

            Regarding the crashes on Haswell, we know that at least one issue of crash on intel CPU was fixed. Could you check on your side? thanks.

              • Re: Do you plan to keep AMD Core Math Library (ACML) for Linux built with Intel® FORTRAN?
                oasanov

                Hi,

                 

                Thanks a lot! Your answer is helpful for me.

                I'll try to use gfortran build and will notify you if any issues will be detected.

                 

                Best Regards,

                Oleg

                • Re: Do you plan to keep AMD Core Math Library (ACML) for Linux built with Intel® FORTRAN?
                  oasanov

                  Hi,


                  I try to use ACML 6.1.0 on Intel CPU E5-2670. It's 2 CPU systems with 16 cores per CPU.

                  flags  : fpu vme de pse tsc msr pae mce cx8 apic mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx pdpe1gb rdtscp lm constant_tsc arch_perfmon pebs bts rep_good xtopology nonstop_tsc aperfmperf pni pclmulqdq dtes64 monitor ds_cpl vmx smx est tm2 ssse3 cx16 xtpr pdcm dca sse4_1 sse4_2 x2apic popcnt aes xsave avx lahf_lm ida arat epb xsaveopt pln pts tpr_shadow vnmi flexpriority ept vpid

                   

                  I have used gfortran 4.8.3 and 4.9.1.

                  ACML examples (from release ACML 6.1.0\gfortran64_mp\examples) are crashed.

                   

                  sgetrf_example:

                  Program received signal SIGSEGV: Segmentation fault - invalid memory reference.

                  Backtrace for this error:
                  #0  0x7F65FC751347
                  #1  0x7F65FC75195E
                  #2  0x3C84832ABF
                  #3  0x7F65FD7AE5D6
                  #4  0x7F65FD7B21F9
                  #5  0x7F65FD7B2C3B
                  #6  0x401097 in MAIN__ at sgetrf_example.f:0
                  #0  0x7F65FC751347
                  #1  0x7F65FC75195E
                  #2  0x3C84832ABF
                  #3  0x7F65FD7AE5D6
                  #4  0x7F65FD7B21F9
                  #5  0x7F65FC52FB0D
                  #6  0x3C854077E0
                  #7  0x3C848E68EC
                  #8  0xFFFFFFFFFFFFFFFF

                   

                  dgetrf_example:

                  Program received signal SIGFPE: Floating-point exception - erroneous arithmetic operation.

                  Backtrace for this error:
                  #0  0x7F569465F347
                  #1  0x7F569465F95E
                  #2  0x3C84832ABF
                  #3  0x7F5695234D3C
                  #4  0x7F5695235C90
                  #5  0x7F56952362A3
                  #6  0x40106C in MAIN__ at dgetrf_example.f:0

                   

                  I have similar issues when I try to use ACML in my fortran tests:

                  dgesv call produce: 38460 Floating point exception

                   

                  Best Regards,

                  Oleg