A Collection of Chat logs between AI and Human

Can I Increase CUDA Memory by Adding Another Graphics Card?

Share:

Facebook
Twitter
LinkedIn

I have 4gb CUDA graphic card. I want to work on AI image processing that needs more than 4gb CUDA. Can I add another 4gb? or should I need to replace the 4gb with new 8gb?

Unfortunately, it is not possible to add more memory to a graphics card once it has been manufactured. The amount of memory on a graphics card is determined by the design and the physical components used to build the card.

If your AI image processing tasks require more than 4GB of CUDA memory, you may need to upgrade your graphics card to one with a higher amount of memory. You could consider replacing your current 4GB graphics card with a new one that has 8GB or more of CUDA memory, which would provide the additional memory you need to process your images.

It’s important to note that there are other factors besides memory that affect the performance of a graphics card for AI image processing, such as the number of CUDA cores and the clock speed of the GPU. So when choosing a new graphics card, it’s important to consider all these factors to ensure that it is suitable for your specific use case.

You didn’t answer that whether I can add another 4gb. for example, I have graphic card Asus XYZ 4gb, I am adding another Asus PQR 4GB. Will it work as 8gb?

I apologize for the confusion in my previous response. To answer your question, it is generally not possible to add another graphics card to increase the amount of CUDA memory available for a single task.

Graphics cards with multiple GPUs (such as the NVIDIA Titan RTX or the NVIDIA GeForce RTX 3090) can use both GPUs to perform AI image processing tasks, effectively doubling the amount of available memory. However, the two GPUs must be connected using a special bridge and have specific driver support.

Simply adding a second identical graphics card (Asus XYZ 4GB and Asus PQR 4GB) would not increase the amount of CUDA memory available for a single task, as each card would only have access to its own 4GB of memory. While it is possible to use multiple graphics cards for parallel processing tasks, such as running multiple tasks at the same time, this would not increase the available memory for a single task.

In short, if your AI image processing tasks require more than 4GB of CUDA memory, the best solution would be to upgrade to a single graphics card with a higher amount of memory, such as an 8GB card or higher.

here they are explaining it is possible: https://tradamaker.com/can-you-use-two-different-graphics-cards-in-one-computer/

I apologize for the confusion in my previous responses. It is possible to use two graphics cards to increase the amount of CUDA memory available for AI image processing, but it requires certain hardware and software configurations.

The configuration you are referring to is called SLI (Scalable Link Interface) on Nvidia graphics cards or Crossfire on AMD graphics cards. To use SLI or Crossfire, both graphics cards must be compatible and connected with a special bridge. The SLI or Crossfire configuration allows the two graphics cards to work together to improve performance in graphics-intensive applications, but it does not necessarily double the amount of available memory.

When using SLI or Crossfire, the memory on each graphics card is mirrored, meaning that each card has a copy of the same data. This can help improve performance in some cases, but it does not necessarily double the amount of available memory for a single task.

In the case of AI image processing, it is possible to use SLI or Crossfire to improve performance, but it is not guaranteed to provide a significant increase in available memory. In most cases, it is still recommended to use a single graphics card with a higher amount of memory if your AI image processing tasks require more than 4GB of CUDA memory.

Post Author

Author

  • Sakthi Sri

    senthil@trilliardbytes.com Sri Sakthi