Any chance MKL could be released under the Intel Open Source License? Would be nice to distribute MKL in projects using GPL libraries.
↧
Intel Open Source License
↧
mkl 64 bit dll gives strange error
I am trying to run anaconda python 64 bit:
(base) D:\Anaconda2>python.exe
Python 2.7.16 |Anaconda, Inc.| (default, Mar 14 2019, 15:42:17) [MSC v.1500 64 bit (AMD64)] on win32
Type "help", "copyright", "credits" or "license" for more information.
>>> import numpy as np
INTEL MKL ERROR: The operating system cannot run %1. mkl_intel_thread.dll.
Intel MKL FATAL ERROR: Cannot load mkl_intel_thread.dll.
in a popup windows it says:
The ordinal 242 could not be located in the dynamic link library
.....\mkl_intel_thread.dll
I can note also that the Intel Python itself does not run, producing the same popup and the same announcement as above.
Furthermore linking 64 bit Fortran programs with 64 bit mkl multithreaded does not work (for me), it only works with mkl sequential.
↧
↧
How to get Cholesky diagonal?
Hello
I need to extract diagonal from cholesky LDLt sparse factorization. Can I do this with Intel MKL?
I can use pardiso to factor and solve sparse linear system. But cholesky factor matrix itself is unreachable for me.
Thanks
↧
mkl 64 bit dll gives strange error
I am trying to run anaconda python 64 bit:
(base) D:\Anaconda2>python.exe
Python 2.7.16 |Anaconda, Inc.| (default, Mar 14 2019, 15:42:17) [MSC v.1500 64 bit (AMD64)] on win32
Type "help", "copyright", "credits" or "license" for more information.
>>> import numpy as np
INTEL MKL ERROR: The operating system cannot run %1. mkl_intel_thread.dll.
Intel MKL FATAL ERROR: Cannot load mkl_intel_thread.dll.
in a popup windows it says:
The ordinal 242 could not be located in the dynamic link library
.....\mkl_intel_thread.dll
I can note also that the Intel Python itself does not run, producing the same popup and the same announcement as above.
Furthermore linking 64 bit Fortran programs with 64 bit mkl multithreaded does not work (for me), it only works with mkl sequential.
↧
Packed GEMM APIs and dynamic batch size
Hi,
I'm interested in further optimizing my application using the packed GEMM API. However, I'm unclear how it behaves in the case of dynamic batch sizes. For example,
- X, the input of shape [M, K] where M is the batch size
- W, the weight of shape [N, K]
The GEMM function should compute X*WT where W can be packed as it remains constant.
How does a change in M affect the packed representation of W? Do cblas_gemm_*_compute functions silently repack W if any of M, N, K is different? Or should it be done manually?
Thanks,
Guillaume
↧
↧
serial vs parellel: different behaviour
Hi
we wrote a header-only library, where we use IntelMKL (wrapped by Armadillo) and open MP in a nested way.
In broad strokes, in the header-only library we do something like this
for (int step =1;step<N_steps; i++){
serial code: Some linear algebra (SVD/ Pseudoinverse).
#parallel openmp for
matrix multiplication
}
My projects usually have the following include structure:
exe using IntelMKL parallel in the VS->property->Intel Performance Libraries-> Use Intel MKL
^
static_lib I compile the header-only library in some function, it includes just IntelMKL headers
^
header-only including IntelMKL headers
we repeat this structure for different project, where the header-only library is in common.
For SOME of the projects, the code in the header-only library crashes in some random way, sometimes in the serial part (the SVD fails with message:
Intel MKL ERROR: Parameter 4 was incorrect on entry to DLASCL.)
sometimes in the loop, where some out of bound location in vector is accessed. If remove the #openMP pragma, it just fails in the SVD at some point.
If I switch in the exe options IntelMKL to serial, it works just fine. Behaviour is the same if I include or exclude the OpenMP support from visual studio.
Any clue on what is causing this? The code spends most of the time in the parallel for, where the intelMKL should be serial anyway, but we would like to use any speedup we can have.
Our setup:
C++ 17
VS 15.9.14
Intel MKL 2019.4.245
CPU intel Xeon Gold 6126 CPU @ 2.59 GHz
SO: Windows 10
but we had this issue on different machines, and with previous version of VS, intel mkl and on different machines.
Happy to provide any information you might require.
↧
Cubic spline interpolation on 3 samples
Hello,
I'm using the data fitting functions of MKL for cubic spline interpolation (DF_PP_CUBIC, DF_PP_DEFAULT) using a not-a-knot condition (DF_BC_NOT_A_KNOT). In theory at least 4 samples are needed to perform this kind of interpolation, which is the unique cubic interpolation polynomial in this case.
My question is: What is the result of the MKL routine in case of 3 samples, which can also be executed without error code? Is this undefined behavior?
The MATLAB routine interp1 performs a simple quadratic polynomial interpolation for example. But this result differs from MKL's.
Thanks for clarification
↧
How can download an old version of MKL like 8.0
Hi, all. I want to use an old version of MKL like 8.0 for a old version software. The old version MKL has libs such as -lmkl_lapack -lmkl_em64t -lguide, which can be applied.
However, I can not find out where to download MKL 8.0.
Thanks.
↧
Intel MKL FATAL ERROR: Cannot load symbol MKLMPI_Get_wrappers. Intel MKL FATAL ERROR: Cannot load symbol MKLMPI_Get_wrappers.
Hi all,
I am trying to compile a tensor library from source and cythonize the code. The code is taken from a github repository: https://github.com/cyclops-community/ctf .Unfortunately, it seems that the cythonized code is throwing me errors when I ran the test cases. I pass the linkers by running the following command in my configure files:
./configure --no-static LD_LIB_PATH="-L$MKLROOT/lib/intel64/" LD_LIBS="-lmkl_scalapack_lp64 -lmkl_intel_lp64 -lmkl_intel_thread -lmkl_core -lmkl_blacs_intelmpi_lp64 -lpthread -lm" CXXFLAGS="-O3 -no-ipo"
I have attached the configure file here. Does anyone know what error is implied by " Intel MKL FATAL ERROR: Cannot load symbol MKLMPI_Get_wrappers. Intel MKL FATAL ERROR: Cannot load symbol MKLMPI_Get_wrappers. ". Googling this shows me it might be because I have a non-dynamic version of mkl_blacs_intelmpi_lp64. I do have it in my directory though.
Best,
Adrian
| Attachment | Size |
|---|---|
Download configure.txt | 41.27 KB |
↧
↧
Using original PARDISO library with Intel MKL
Hello!
I am trying to integrate the original PARDISO library (v 6.0) with Intel MKL 2019 for performance comparison but I am receiving the following error message upon compilation:
libpardiso600-GNU800-X86-64.so: undefined reference to `log2f@GLIBC_2.27'
libpardiso600-GNU800-X86-64.so: undefined reference to `logf@GLIBC_2.27'
Here is the command line:
icc pardiso_unsym.c -o p.out -LPARDISO libpardiso600-GNU800-X86-64.so -I$MKLROOT/include -L${MKLROOT}/lib/intel64 -lmkl_gf_lp64 -lmkl_sequential -lmkl_core -lm -ldl -lgfortran -fopenmp
I am using Linux and the sample file can be downloaded from here: https://www.pardiso-project.org/manual/pardiso_unsym.c
Any idea what could be the issue? I really appreciate any help.
Best,
Afshin
↧
What is the largest matrix that can be handled through LAPACK in mkl_lp64?
Hello,
I am having troubles with diagonalizing a symmetric matrix of size 40,000 x 40,000. Just storing the unique elements already exceeds the largest integer we can store with 32-bit. However, it does not exceed the limit of 64-bit integer.
I thought with LAPACK through mkl_lp64 I should be able to diagonalize this but it crashed with an error:
Intel MKL ERROR: Parameter 8 was incorrect on entry to DSYEVD.
The parameter 8 is lwork and it is above the maximum 32-bit integer.
Is it the case that even with mkl_lp64 I cannot use LAPACK to diagonalize matrices of size 40,000 x40,000?
Should I be using mkl_ilp64?
Thank you!
↧
Eigen::PardisoLLT is slower than Eigen::SimplicialLLT
Hi,
I followed this page(https://eigen.tuxfamily.org/dox/TopicUsingIntelMKL.html) to use Eigen::PardisoLLT to solve a sparse linear system(88869x88869). I have MKL(student version) installed. I’ve successfully compiled, linked and run it. However, it’s 5 times slower than Eigen::SimplicialLLT. Am I doing anything wrong there?
Best!
Jing
↧
MKL distribution problem
I have a mixed language program that uses a fortran DLL that uses the MKL, everything works fine on the development machine but when I try to distribute it to another machine I get a module not found error when the program calls the DLL. Through a process of elimination I have tracked the problem down to the MKL. If I do not include the MKL in the link and remove all calls to MKL routines everything is fine. As soon as I add the MKL to the link, I get the module not found error even with no calls to the MKL routines. I have tried installing MKL onto the distribution machine, and adding all the redist MKL dlls to the installation folder but neither makes any difference.
I am using MKL 2019 update 4 on both machines.
Any ideas gratefully accepted.
↧
↧
Deprecating pardiso_getenv/pardiso_setenv [asking for feedback]
Hello all!
In Intel MKL we're considering deprecating pardiso_getenv/pardiso_setenv functonality (https://software.intel.com/en-us/mkl-developer-reference-c-pardiso-geten...) as we suspect that it is unused and we strongly believe it can confuse the customers in its current form.
To this end, we'd like to ask whether someone is using this feature now and know about the use cases. We hope to get some form of feedback within 30 days, otherwise we will assume that this feature doesn't affect anyone.
Thanks,
Kirill
↧
SPARSE_STATUS_NOT_SUPPORTED exit status in mkl_sparse_sp2m with symmetric sparse matrices
↧
Inconsistent C definition of ?gelsd function prototypes?
Recently started using MKL alongside other wrapper libraries like Armadillo and ran into some LAPACK function definition conflicts. Upon closer inspection, it appears that the MKL definitions of ?gelsd functions are inconsistent in mkl_lapack.h, and are incorrect for sgelsd, dgelsd, zgelsd.
For sgelsd, dgelsd, and zgelsd, the parameter for matrix "a" is defined as "const type*". This should not be the case as matrix "a" is overwritten as an output so it should not be defined with a "const" qualifier in the function parameters.
cgelsd correctly defines "a" without the "const" qualifier. This should be an easy fix but I'm unsure as how to contribute to the MKL code base since it's not open source.
MKL Version: 2019.4
↧
Different results in 32 and 64 bit applications (on same cpu)
Hello,
We are currently evaluating a replacement of our Netlib Blas/Lapack libs by the Intel MKL. As we need identical computation results on different hardware, we have done some tests and realised that the the results from the MKL routines differ not only between different cpus but also depend on the architecture of the application (32/64bit) from which they are called. Switching to MKL_CBWR_COMPATIBLE mode as described in other posts seem to make the results independent of the cpu type. However, the divergences between 32 and 64 bit application on same cpu remains.
Up to now we have not found any setting or configuration that help. So we would be interested in any tips. Anything we can try?
Our application runs on Windows 10 and is compiled in native C++ as 32 and 64 bit edition by Visual Studio 2019. MKL version is 2019 update 4.
We are focusing on eigenvector calculations. An example for calling dggev and zgeev could be made available.
Thanks and regards,
Christoph
↧
↧
Package Signature Verification Failed
Hello,
I'm developing an application that requires the use of Intel MKL. MKL was successfully installed on our Linux (RHEL) machines, and I'm trying to migrate the codebase over to Windows.
All of the computers are in private networks with no internet access. I downloaded the Intel MKL 2019 Version 4 "Full Package" (543 MB) and moved it over to the following Windows machines:
- Windows Server 2008
- Windows 10 Enterprise
When I run the w_mkl_2019.4.245.exe installer, I can select an installation directory, but it crashes barely 5% of the way in. It provides the following message on both versions of Windows:
Error: Package signature verification failed. Click Help for details.
When I click on 'help', it directs me to a webpage, which obviously won't open. I've run it as an administrator, and looked at posts on here for reference. Many of the links (like https://software.intel.com/en-us/articles/retired-articles-intel-c-and-f...) are now dead, or offer solutions exclusive to other Intel products.
What should I do to resolve this problem? (hopefully without downloading anything else, but I can do so if necessary)
Brad
↧
lapack dsyevx
<p>I used to write codes in fortran 77, of course it was in the library, now I'm using the 2018 student version and it needs to be fine-tuned. I've been trying to run a sample snippet using lapack for a long time, but it just doesn't work. I would appreciate if you help ..<br>Best regards</p>
| Attachment | Size |
|---|---|
| DownloadSource1.for | 5.07 KB |
↧
LNK2019 unresolved external symbol _DGEMM referenced in function [...]
Hi,
I want to use dgemm function from mkl lib in my (C++) program but i keep getting linking error... could you please help me with this?
I included the lib (i.e. #include "include/mkl.h"), I think it should run...
Thanks in advance...
↧