MKL memory limitation?

[everest]

Hi,

I am trying to evaluate MKL 8.0 for my large matrix solving project and encountered the following strange behavior. I am using the function cgetrf() and cgetrs() to perform LU decomposition. For small matrix, the code runs fine and pretty fast. However, it seems to me it can not go beyond solving matrix size at about 10000x10000 (single precision complex matrix), even the computer has 2GB memory. A straight copy of code from numerical recipe can solve up to 15kX15k matrix problem on the same machine (although much slower than MKL).

So my question is

1. is this a known problem?

2. Is there anyway (such as turn on some flags, etc.) to avoid this "premature" memory limit, even it means some sacrifice of speed.

Thanks

[TimP]

If you didn't try the usual stack limit adjustments:
/link /stack:nnnnnnnnn (and its relatives, for Windows, along with /3GB boot parameter for XP-32)
info ulimit (linux)
those should apply, but they would not fit your "strange" description.
It looks like you could run into the shortage of address space under 32-bit XP-SP2 with everything set to default, but I hesitate to guess your OS.

[everest]

Hi,

I am using Linux (SLES 2.6.5-7.151-smp). The machine I am using has following CPU

processor : 0
vendor_id : GenuineIntel
cpu family : 15
model : 3
model name : Intel Pentium 4 CPU 3.40GHz
stepping : 4
cpu MHz : 3393.886
cache size : 1024 KB
physical id : 0
siblings : 2
fdiv_bug : no
hlt_bug : no
f00f_bug : no
coma_bug : no
fpu : yes
fpu_exception : yes
cpuid level : 5
wp : yes
flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe pni
monitor ds_cpl cid
bogomips : 6701.05

and my "limit" scenario is as follows

cputime unlimited
filesize unlimited
datasize unlimited
stacksize 8192 kbytes
coredumpsize 0 kbytes
memoryuse unlimited
vmemoryuse unlimited
descriptors 1024
memorylocked unlimited
maxproc 32255