LAPACK GELQF Openblas MXNet

intel_mkl_sgeqrf.md

Intel MKL ERROR: Parameter 4 was incorrect on entry to SGEQRF.
[03:23:19] ../src/operator/tensor/./../linalg_impl.h:1058: ------------- linalg---------------- gelqf
Traceback (most recent call last):
  File "debug.py", line 3, in <module>
    print(q)
  File "/home/ubuntu/chai-mxnet/python/mxnet/ndarray/ndarray.py", line 261, in __repr__
    return '\n%s\n<%s %s @%s>' % (str(self.asnumpy()),
  File "/home/ubuntu/chai-mxnet/python/mxnet/ndarray/ndarray.py", line 2570, in asnumpy
    ctypes.c_size_t(data.size)))
  File "/home/ubuntu/chai-mxnet/python/mxnet/base.py", line 246, in check_call
    raise get_last_ffi_error()
mxnet.base.MXNetError: Traceback (most recent call last):
  File "../src/operator/tensor/./../linalg_impl.h", line 1058
sgelqf: Check failed: ret == 0 (-4 vs. 0) : Workspace query failed on CPU.

lapack_gelqf_openblas64_error_log.md

In file included from ../src/operator/numpy/linalg/./../../tensor/../linalg.h:31:0,
                 from ../src/operator/numpy/linalg/./../../tensor/la_op-inl.h:29,
                 from ../src/operator/numpy/linalg/./np_norm-inl.h:35,
                 from ../src/operator/numpy/linalg/np_norm_forward.cc:26:
../src/operator/numpy/linalg/./../../tensor/.././c_lapack_api.h: In function ‘int MXNET_LAPACK_sgelqf(int, mshadow::index_t, msh
adow::index_t, float*, mshadow::index_t, float*, float*, mshadow::index_t)’:
../src/operator/numpy/linalg/./../../tensor/.././c_lapack_api.h:483:63: error: cannot convert ‘mshadow::index_t* {aka long int*}
’ to ‘int*’ for argument ‘1’ to ‘void sgeqrf_(int*, int*, float*, int*, float*, float*, int*, int*)’
       prefix##geqrf_(&n, &m, a, &lda, tau, work, &lwork, &info); \
                                                               ^
../src/operator/numpy/linalg/./../../tensor/.././c_lapack_api.h:490:3: note: in expansion of macro ‘MXNET_LAPACK_CWRAP_GELQF’
   MXNET_LAPACK_CWRAP_GELQF(s, float)
   ^
../src/operator/numpy/linalg/./../../tensor/.././c_lapack_api.h: In function ‘int MXNET_LAPACK_dgelqf(int, mshadow::index_t, msh
adow::index_t, double*, mshadow::index_t, double*, double*, mshadow::index_t)’:
../src/operator/numpy/linalg/./../../tensor/.././c_lapack_api.h:483:63: error: cannot convert ‘mshadow::index_t* {aka long int*}
’ to ‘int*’ for argument ‘1’ to ‘void dgeqrf_(int*, int*, double*, int*, double*, double*, int*, int*)’
       prefix##geqrf_(&n, &m, a, &lda, tau, work, &lwork, &info); \
                                                               ^
../src/operator/numpy/linalg/./../../tensor/.././c_lapack_api.h:491:3: note: in expansion of macro ‘MXNET_LAPACK_CWRAP_GELQF’
   MXNET_LAPACK_CWRAP_GELQF(d, double)
   ^