fill_with_randn error

[GoogleCodeExporter]

What steps will reproduce the problem?
1. the code below has no problem if no gpu is used
2. when nsize >= 512 and num_batch>=100, the script will complain.
######
import os, sys
import gnumpy as gpu

nsize = 1024 #broken with 512
num_epoch = 1
num_batch = 100
for i in range(num_epoch):
    a = []
    for j in range(num_batch):
        if j == 0:
            a = gpu.randn(nsize, 432 )
        else:
            a = gpu.concatenate((a,gpu.randn(nsize,432) ),0)
#####

What is the expected output? What do you see instead?
The error message is below:
###########
  File "test2.py", line 17, in <module>
    a = gpu.concatenate((a,gpu.randn(nsize,432) ),0)
  File "/home/yun_tang/tools/NN/gnumpy/gnumpy.py", line 333, in randn
    return _rand__base(shapeInfo, 'normal', True)
  File "/home/yun_tang/tools/NN/gnumpy/gnumpy.py", line 291, in _rand__base
    {'uniform': _cmType.fill_with_rand, 'normal': _cmType.fill_with_randn}[distribution](ret._base)
  File "/home/yun_tang/tools/NN/gnumpy/cudamat/cudamat.py", line 369, in fill_with_randn
    raise generate_exception(err_code)
cudamat.cudamat.CUDAMatException: CUDA error: no error
#########

What version of the product are you using? On what operating system?
The version of cudamat is released at 01-15-2010. I tested it on a linux gpu 
server (NVIDIA Tesla K10 GPU).

Please provide any additional information below.
The script has no problem with numpy (using cpu)

Original issue reported on code.google.com by hott...@gmail.com on 8 Feb 2013 at 8:30