PGI User Forum

[alechand]

Hello.
I am testing the simple program, modified for OPENACC, called "picalc" from NVIDIA website :

############################

program picalc
implicit none
integer, parameter :: n=1000000
integer :: i
real(kind=8) :: t, pi
pi = 0.0
!$acc parallel loop
do i=0, n-1
t = (i+0.5)/n
pi = pi + 4.0/(1.0 + t*t)
end do
!$acc end parallel loop
print *, 'pi=', pi/n
end program picalc

############################


This program is simple, but my system gives this error:


############################

alechand@pcsantos2:~/gravity$ pgfortran -fast -Minfo=all -o TEST picalc.f90 -ta=nvidia
picalc:
7, Accelerator kernel generated
7, CC 1.3 : 24 registers; 32 shared, 36 constant, 0 local memory bytes
CC 2.0 : 23 registers; 0 shared, 52 constant, 0 local memory bytes
8, !$acc loop gang, vector(256) ! blockidx%x threadidx%x
10, Sum reduction generated for pi
7, Generating compute capability 1.3 binary
Generating compute capability 2.0 binary
alechand@pcsantos2:~/gravity$ ./TEST
call to cuMemcpyDtoH returned error 700: Launch failed
CUDA driver version: 5050

############################

I have this problem also in other codes.
I installed the PGI compiler 12.10,
and i am using kubuntu 12.04.

Can you help me ?
thanks

[mkcolg]

Hi alechand,

There's something going on with your device or driver. Can you post the output to the command "pgaccelinfo"? Also, are you able to run a simple CUDA program?

Thanks,
Mat

[alechand]

thanks for the reply.
Here is the output :

#####################################

alechand@pcsantos2:~$ pgaccelinfo
CUDA Driver Version: 5050
NVRM version: NVIDIA UNIX x86 Kernel Module 319.17 Thu Apr 25 22:14:10 PDT 2013

Device Number: 0
Device Name: GeForce GTX 680
Device Revision Number: 3.0
Global Memory Size: 2147155968
Number of Multiprocessors: 8
Number of SP Cores: 1536
Number of DP Cores: 512
Concurrent Copy and Execution: Yes
Total Constant Memory: 65536
Total Shared Memory per Block: 49152
Registers per Block: 65536
Warp Size: 32
Maximum Threads per Block: 1024
Maximum Block Dimensions: 1024, 1024, 64
Maximum Grid Dimensions: 2147483647 x 65535 x 65535
Maximum Memory Pitch: 2147483647B
Texture Alignment: 512B
Clock Rate: 1058 MHz
Execution Timeout: No
Integrated Device: No
Can Map Host Memory: Yes
Compute Mode: default
Concurrent Kernels: Yes
ECC Enabled: No
Memory Clock Rate: 3004 MHz
Memory Bus Width: 256 bits
L2 Cache Size: 524288 bytes
Max Threads Per SMP: 2048
Async Engines: 1
Unified Addressing: No
Initialization time: 314151 microseconds
Current free memory: 2095439872
Upload time (4MB): 994 microseconds ( 843 ms pinned)
Download time: 1733 microseconds ( 759 ms pinned)
Upload bandwidth: 4219 MB/sec (4975 MB/sec pinned)
Download bandwidth: 2420 MB/sec (5526 MB/sec pinned)

###################################

what do you mean a cuda program ?
can you give me an example ?

thanks

[mkcolg]

The pgaccelinfo output all looks fine. My test system is a GTX690 so very similar to yours. The only difference is that your driver is newer. I'll see if I can update my driver to see if that's causing the problem.

Can you now try setting the environment variable "PGI_ACC_DEBUG=1" and run your program again?

[alechand]

i used

PGI_ACC_DEBUG=1

but the behaviour was the same as before.
I installed the cuda driver from pgi compiler,
can you help me to find this sample ?