Install CUDA, Nvidia driver and cudnn for GeForce GT 730.

I want to install the correct version of CUDA, Nvidia driver and cudnn for GeForce GT 730 in Ubuntu 16.04 X86_64 OS, could anyone tell me the right version for this graphic card?

The GeForce GT 730 comes in 2 different flavors, one of which is compute capability 3.5, the other is compute capability 2.1

If you have the cc 2.1 version, cuDNN will not work with that GPU (it requires 3.0 or higher).

Other than that, choose the latest driver for your gpu using the wizard at

If you have the cc 3.5 GPU, CUDA 8, 9, or 9.1 will work with that GPU. If you have the 2.1 version, only CUDA 8 or prior will work.

People are often interested in cuDNN because they want to run something that depends on it, such as Tensorflow. If that is the case, then choose your CUDA and cuDNN versions based on the requirements of the specific TF version you want to run (or the requirements of whatever software you want to run that depends on cuDNN/CUDA).

1 Like

How to check my GeForce GT 730 is in which flavor? Thank you!

do a proper install of CUDA 8 and run deviceQuery on it

what the result after you command “nvidia-smi”? is it no supported like below

| NVIDIA-SMI 390.77 Driver Version: 390.77 |
| GPU Name Persistence-M| Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. |
| 0 GeForce GT 730 Off | 00000000:01:00.0 N/A | N/A |
| 65% 30C P8 N/A / N/A | 182MiB / 1982MiB | N/A Default |

| Processes: GPU Memory |
| GPU PID Type Process name Usage |
| 0 Not Supported |

Kindly try what Robert Crovella recommended in #4, and copy the output of the deviceQuery app here.

I’m hoping for an expert opinion on the following case:

I tried cuda-11-2 with a GeForce gt730, and some of the /usr/local/cuda/samples programs seemed to run okay. At least they did not report any errors.

However, what I really wanted to do was to test the memory of my gt730 card using memtestG80, which I compiled using cuda-11-2. When I ran the memtestG80, it reported 4e9 errors in a very short time, as seen here:

Test iteration 1 (GPU 0, 4 MiB): 0 errors so far
Moving Inversions (ones and zeros): 4294967295 errors (0 ms)
Memtest86 Walking 8-bit: 4294967288 errors (0 ms)

I’m guessing that the memory test did not really run at all? Can the experts confirm my suspicion? Thanks. I am looking into using cuda-9-1 as suggested above, also, but I’m expecting a bit of a struggle installing it on ubuntu 20.04 since it was built for /on17.04.

Note: From ./deviceQuery

Device 0: “GeForce GT 730”
CUDA Driver Version / Runtime Version 11.2 / 11.2
CUDA Capability Major/Minor version number: 3.5
Total amount of global memory: 981 MBytes (1028980736 bytes)
( 2) Multiprocessors, (192) CUDA Cores/MP: 384 CUDA Cores

UPDATE: I managed to install cuda-9.1 as well as g++-6 and gcc-6, at which point I could compile memtestG80 and run it against my gt730 card. Now I get 0 errors, and it takes some 300ms per iteration to test 960MB of memory.

Looks like my previous question can now be ignored.