cancel
Showing results for 
Search instead for 
Did you mean: 
Administration & Architecture
Explore discussions on Databricks administration, deployment strategies, and architectural best practices. Connect with administrators and architects to optimize your Databricks environment for performance, scalability, and security.
cancel
Showing results for 
Search instead for 
Did you mean: 

GPU accelerator not matching with desired memory.

jaytimbadia
New Contributor II

Hello, 

We have opted for Standard_NC8as_T4_v3 which claims to have 56GB memory. 

But, when I am doing nvidia-smi in the notebook, its showing only ~16 GB, 

Why?

Please let me know what is happening here?  

Jay

3 REPLIES 3

Alberto_Umana
Databricks Employee
Databricks Employee

Hello @jaytimbadia,

When you run the nvidia-smi command, it reports the GPU memory, which is why you are seeing approximately 16GB. The 56GB refers to the overall system memory available for the instance, not the memory available on the GPU itself.

The discrepancy you are observing is due to the difference between the total system memory and the GPU memory. The Standard_NC8as_T4_v3 instance type has a total of 56GB of system memory (RAM), but the NVIDIA T4 GPU within this instance has only 16GB of GPU memory

jaytimbadia
New Contributor II

Ok, so where is this mentioned? Can you please provide some transparency?

Alberto_Umana
Databricks Employee
Databricks Employee

Connect with Databricks Users in Your Area

Join a Regional User Group to connect with local Databricks users. Events will be happening in your city, and you won’t want to miss the chance to attend and share knowledge.

If there isn’t a group near you, start one and help create a community that brings people together.

Request a New Group