Native crash in ntdll.dll when executing a Kernel that works fine in aparapi r997

What steps will reproduce the problem?
1. Launch class test.MaxKernelTest from MaxKernel-for-recent-aparapi.zip

or, in r997:

1alt. Launch class test.MaxKernelTest from MaxKernel-for-aparapi-r997.zip 
(import declarations change in order to reflect package changes in aparapi 
classes)

What is the expected output? 

Desired goal: finding maximum values per-row, in a float matrix stored 
row-major with r=1000 rows, each having c=90 columns. Thus, output should be a 
1000-float array.
Computation should be carried out in groups each having 2^n threads, with n 
chosen in order to having 2^n less or equal than c.

What do you see instead?

Native crash with this message:

#
# A fatal error has been detected by the Java Runtime Environment:
#
#  EXCEPTION_ACCESS_VIOLATION (0xc0000005) at pc=0x0000000077023290, pid=11280, 
tid=12964
#
# JRE version: Java(TM) SE Runtime Environment (7.0_45-b18) (build 1.7.0_45-b18)
# Java VM: Java HotSpot(TM) 64-Bit Server VM (24.45-b08 mixed mode 
windows-amd64 compressed oops)
# Problematic frame:
# C  [ntdll.dll+0x53290]
#
# Failed to write core dump. Minidumps are not enabled by default on client 
versions of Windows
#
# An error report file with more information is saved as:
# D:\Workspaces\SHAPEnew\com.exeura.shape.algo\hs_err_pid11280.log
#
# If you would like to submit a bug report, please visit:
#   http://bugreport.sun.com/bugreport/crash.jsp
# The crash happened outside the Java Virtual Machine in native code.
# See problematic frame for where to report the bug.
#

On the same code, aparapi r997 works fine.

What version of the product are you using? 

Latest aparapi distribution available (though not officially ... sorry! 
Couldn't compile for x86_64 :( ) from link 
https://groups.google.com/group/aparapi-discuss/attach/5eeea362b440b1ca/dist_win
dows_x86_64.zip?part=4&authuser=0

(link to post: 
https://groups.google.com/d/msg/aparapi-discuss/mFVA0tet2NM/yrFAtGKj7l4J )

Other aparapi version I used is r997, available from: 
https://code.google.com/p/aparapi/downloads/detail?name=Aparapi_2013_01_23_windo
ws_x86_64.zip&can=2&q=

On what operating system?

Windows 7 Professional 64-bit

Please provide any additional information below.

OpenCL devices available:

GPU: Nvidia Quadro FX 880 M, 1 GB DDR3 with 128 bit channel (drivers: ForceWare 
320.49)
CPU: i7 840QM, 8 GB DDR3 (OpenCL software: Intel OpenCL SDK for Application 
2013 64-bit)
Also, I installed AMD-APP v923.1 during a failed attempt to compile aparapi, 
thus a OpenCL CPU device is showing in the list (though not used in test)

In APARAPI words:

-----------
Platform: PlatformId 8791249725960
Name:Advanced Micro Devices, Inc.
Version:OpenCL 1.2 AMD-APP (923.1)
Device: Device 233889248
  type:CPU
  maxComputeUnits=8
  maxWorkItemDimensions=3
  maxWorkItemSizes={1024, 1024, 1024}
  maxWorkWorkGroupSize=1024
  globalMemSize=8575078400
  localMemSize=32768
Platform: PlatformId 185264896
Name:Intel(R) Corporation
Version:OpenCL 1.2 
Device: Device 202790872
  type:CPU
  maxComputeUnits=8
  maxWorkItemDimensions=3
  maxWorkItemSizes={1024, 1024, 1024}
  maxWorkWorkGroupSize=1024
  globalMemSize=8575078400
  localMemSize=32768
Platform: PlatformId 107992320
Name:NVIDIA Corporation
Version:OpenCL 1.1 CUDA 4.2.1
Device: Device 239888432
  type:GPU
  maxComputeUnits=6
  maxWorkItemDimensions=3
  maxWorkItemSizes={512, 512, 64}
  maxWorkWorkGroupSize=512
  globalMemSize=1073741824
  localMemSize=16384
---------------

obtained with code:

---- (for r997) ----
        for (final OpenCLPlatform openCLPlatform : OpenCLPlatform.getPlatforms())
        {
            System.out.println("Platform: " + openCLPlatform);
            for (final OpenCLDevice openCLDevice : openCLPlatform.getDevices())
            {
                System.out.println("Device: " + openCLDevice);
            }
        }
---- (for r997) ----
Original issue reported on code.google.com by lgallucci@gmail.com on 22 Oct 2013 at 2:46
Attachments:
rzel / aparapi

Native crash in ntdll.dll when executing a Kernel that works fine in aparapi r997 #131
