Tesla doesn't work in multidevice system

Accelerated Computing CUDA CUDA Programming and Performance
1

Hi all,

I have GeForce 8600GT in first PCIex and Tesla C870 in second. While device query shows that each device is present, I cannot run any CUDA app on tesla. MultiGPU hangs on execution. I tried to set device to tesla in DWTHaar1D - test fails.

Computer spec :

Quad Core 2.4 GHz

Abit IP35 PRO motherboard

4 GB RAM

560 W Chieftec CFT-560-A12C

Palit Sonic+ GeForce 8600 GT (PCIEx 16x)

TESLA C870 (PCIEx 4x)

Windows 2003 Server x64 SP2

ForceWare 175.16

Device Query listing :

There are 2 devices supporting CUDA

Device 0: "GeForce 8600 GT"

  Major revision number:                         1

  Minor revision number:                         1

  Total amount of global memory:                 268107776 bytes

  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:                                    1350000 kilohertz

Device 1: "Tesla C870"

  Major revision number:                         1

  Minor revision number:                         0

  Total amount of global memory:                 1610350592 bytes

  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:                                    1350000 kilohertz

Test PASSED

Anybody knows where is the problem ?

2

Have you verified that you’re using the latest motherboard BIOS?

Tesla cards have never been tested or qualified to run in a PCI-E x4 slot. If possible, you should try swapping the 8600GT and C870.

3

I updated BIOS. Problem with multiGPU has been solved, test pass, but MonteCarloMultiGPU still fails :

main(): allocating host memory...

main(): generating input data...

main(): starting 2 host threads...

main(): waiting for GPU results...

main(): GPU statistics

GPU #0

Options         : 128

Simulation paths: 8388608

Time (ms.)      : 340.824921

Options per sec.: 375.559392

GPU #1

Options         : 128

Simulation paths: 8388608

Time (ms.)      : 36.285835

Options per sec.: 3527.547294

main(): comparing Monte-Carlo and Black-Scholes results...

L1 norm        : 4.931723E-001

Average reserve: -1.#IND00

TEST FAILED.

Shutting down...

Tesla is in PCIEx x16 slot, but it works as x4, chipsets before x38 cannot work with 2 pciex16 with full speed.

When I set the device to Tesla in DWTHaar1D - it fails.

4

If you use CUDA 2.0, it’s only supported by the 174.55 beta drivers… the devices will be listed but the program will crash ( or almost that’s what happened to me ).

If you’re using CUDA 1.1… sorry, I don’t know.

5

I swapped Tesla and GF8600, now it works :)
Before swap, I also upgraded to CUDA 2.0b2