Author Topic: Dear All.. Is NVIDIA's road is the best for a Robotics Start-Up?  (Read 2793 times)

0 Members and 1 Guest are viewing this topic.

Offline salihkanberTopic starter

  • Regular Contributor
  • *
  • Posts: 140
  • Country: us
  • Hello all.. It's nice to be here..
    • Dynamic, Versatile, Innovative Start-Up
Dear All.. This is Salih, Founder of Sunday Robotics.. I am trying to decide the route to take for our AI and DeepLearning goals. NVIDIA is doing a lot today, would their ware be the best route for a Robotics & AI Start-Up? Can I have your suggestions? Thanks.
 

Offline rstofer

  • Super Contributor
  • ***
  • Posts: 9886
  • Country: us
Re: Dear All.. Is NVIDIA's road is the best for a Robotics Start-Up?
« Reply #1 on: October 28, 2021, 05:00:26 pm »
For small projects, the NVIDIA Jetson Nano is interesting as is the Coral Dev Board (Google product).  These both support full Linux as well as specialized cores for all the math required.  But the number of cores is fairly small - 128 in the case of the Nano.

For larger projects, I'm not aware of any company even coming close to the NVIDIA graphics cards with over 10,000 CUDA cores.  That's a LOT of parallel computing.

The new $1,500 RTX 3090 has 10,496 cores, for 36 teraflops.  We went to the Moon and back using machines capable of 2-3 megaflops!

You can do a lot of machine learning with 36 teraflops.
 
https://www.engadget.com/nvidia-rtx-3090-3080-3070-cuda-core-int32-fp32-210059544.html

Tensor cores are the new big deal - worth searching Google...
 
The following users thanked this post: salihkanber

Offline salihkanberTopic starter

  • Regular Contributor
  • *
  • Posts: 140
  • Country: us
  • Hello all.. It's nice to be here..
    • Dynamic, Versatile, Innovative Start-Up
Re: Dear All.. Is NVIDIA's road is the best for a Robotics Start-Up?
« Reply #2 on: October 29, 2021, 09:13:31 pm »
For small projects, the NVIDIA Jetson Nano is interesting as is the Coral Dev Board (Google product).  These both support full Linux as well as specialized cores for all the math required.  But the number of cores is fairly small - 128 in the case of the Nano.

For larger projects, I'm not aware of any company even coming close to the NVIDIA graphics cards with over 10,000 CUDA cores.  That's a LOT of parallel computing.

The new $1,500 RTX 3090 has 10,496 cores, for 36 teraflops.  We went to the Moon and back using machines capable of 2-3 megaflops!

You can do a lot of machine learning with 36 teraflops.
 
https://www.engadget.com/nvidia-rtx-3090-3080-3070-cuda-core-int32-fp32-210059544.html

Tensor cores are the new big deal - worth searching Google...

That is fantastic, certainly more than we need right now but I see the potential. For a Robotic start-up that deals with Rasp Pi's right now, its a bit confusing to create the path, since we can't have it all, the best way would be to focus on a specific problem and try to develop robotic and AI solutions upon it. Don't you think?
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf