Failed Tests with CUDA 2.0

Hi everyone,

I’ve run into failed test problems with the provided SDK samples.

My installation system:

Intel MacBook Pro, NVIDIA 8600M GT 128MB, 2GB RAM, MAC OS X 10.5.2

GCC 4.0.1

This was my first installation of CUDA, so no residual CUDA 1.1 files were lying around.

The steps I followed:

  • Installed the toolkit (made sure to install the CUDA.kext driver)

  • Added the environment variables

  • Installed the SDK

  • Typed ‘make’ (made sure to not compile to emulation mode)

Result of ./deviceQuery:

There is 1 device supporting CUDA

Device 0: "GeForce 8600M GT"

  Major revision number:                         1

  Minor revision number:                         1

  Total amount of global memory:                 134021120 bytes

  Number of multiprocessors:                     4

  Number of cores:                               32

  Total amount of constant memory:               65536 bytes

  Total amount of shared memory per block:       16384 bytes

  Total number of registers available per block: 8192

  Warp size:                                     32

  Maximum number of threads per block:           512

  Maximum sizes of each dimension of a block:    512 x 512 x 64

  Maximum sizes of each dimension of a grid:     65535 x 65535 x 1

  Maximum memory pitch:                          262144 bytes

  Texture alignment:                             256 bytes

  Clock rate:                                    0.75 GHz

  Concurrent copy and execution:                 Yes

Test PASSED

From the samples that I have tried, the following run fine:

  • matrix

  • matrixMulDrv

  • Mandelbrot

  • bitonic

  • boxFilter

  • nbody

  • dxtc

The following fail:

  • BlackScholes (test fails with overall screen corruption)

  • MersenneTwister (machine locks up)

  • alignedTypes (all tests fail)

  • bandwidthTest (crashes with a bus error)

  • asyncAPI (nothing happens)

Installing the latest update of Mac OS X from 10.5.2 to 10.5.5 did not improve my results.

Any advice? Could it be related to only having 128MB RAM on the graphics card?

Any help will be greatly appreciated.

I wouldn’t be surprised. Because Quartz takes up a non-trivial amount of VRAM, your free memory is very low. I’ve had no problems running those demos on a MacBook Pro with a 256MB 8600M GT, so the difference in video memory gets my vote.

Thanks for the quick reply. I managed to track down a similar post from February on this forum (available here). Using the provided code, it seems like the available memory fluctuates between 35MB and 61MB (out of 128MB total). Pretty bad news.

I will switch over to the emulator when my code starts to require more memory. Unfortunately there are no easy upgrade options for graphics cards on the MacBook Pro, and the other machine I have access to (Mac Pro) has an ATI card :/

Thanks again!

Ooops. wrong forum.