search

googlecolab / colabtools

Python libraries for Google Colaboratory
Apache License 2.0
2.21k stars 725 forks source link

Feature Request: Add NVIDIA A100 or H100 with 80GB Memory to Google Colab Pro + #3784

Open timechild666 opened 1 year ago

timechild666 commented 1 year ago

Current compute limitations in Google Colab Pro + can sometimes restrict the handling of large-scale datasets or intricate models. An ideal resolution would be to introduce higher-performance GPUs, specifically the NVIDIA A100 or H100 with 80GB memory, into the Pro + tier. Such an upgrade could offer a significant boost in computing capabilities for substantial projects. Naturally, these powerful GPUs would necessitate an adjustment in the current compute credit structure, reflecting their superior power and value.

Switching to alternative platforms like AWS, Azure, or third-party providers such as Runpod or Paperspace is an option, but they bring their own set of challenges. These include steep learning curves, additional costs, and in the case of third-party providers, a lack of regular software updates which results in an added maintenance hassle.

The user-friendly nature of Google Colab, coupled with its collaborative capabilities, make it a preferred choice for many. By offering high-performance GPUs, Google Colab could potentially attract a broader range of users with more demanding computational needs. This could lead to an expanded user base, increased revenue, and strengthen its competitive standing in the cloud-based machine learning platform market.

cperry-goog commented 1 year ago

Thanks for the feedback - keeping this open as we think through hardware refreshes.

timechild666 commented 3 days ago

Hi there, just checking in on this feature request. It’s been a while, and I noticed there have been some updates to Colab recently. So I was wondering if a hardware refresh might be coming too?

The A100 is starting to show its limits even for medium-scale projects, so I was curious if the H100 might be on the roadmap.