How to use LightOn Cloud (General guide)

LightOn Cloud availability

  • During the early release stage of LightOn Cloud, we are trying to make our service available to as many users as possible, without overwhelming our current OPU capacity.
  • As more OPUs are added to the LightOn Cloud we will be able to satisfy more access requests.
  • You can express your interest to use LightOn Cloud through this form. LightOn Cloud operates on the basis of a credit system. During your initial request for access, you also select your prefered credit package. Pricing information can be found here.
  • You can request access to the LightOn Cloud for Research program (free credits to use LightOn Cloud, under certain terms & conditions) through this form.

What happens after my request?

  • If the sufficient capacity exists we will send you all the necessary information to proceed with the payment.
    • Upon receiving the payment, you’ll get your credentials within 1 working day.
    • If sufficient capacity does not exist, we will try to inform you about the availability, but we cannot guarantee any timeline to do so.
    • Alongside your credentials, you will receive all other necessary information to use LightOn Cloud (e.g. links for documentation, available support channels, etc.)
  • And if you want to buy more credits just let us know at support[at]lighton.io.

LightOn Cloud characteristics

  • In LightOn Cloud you have access to LightOn Aurora 1.5 OPU alongside NVIDIA® Tesla® V100 32GB and Intel® Xeon® Gold 16 cores (see details here).
  • LOC supports Jupyter Lab: before the start of each booking, you will receive a link at your email. You can install any extension to it and you are able to set extension installation at each boot to automate the process.
  • You have in your disposal a private data directory, persistent across bookings (mounted on ~/data). You can access these data anytime (no booking is required) through SFTP.

Please keep in mind that only the data folder (~/data) is persistent across bookings. Everything else is erased at the end of your booking. For example, simulation results should go in this data folder.

  • You have access to a shared storage space (mounted on /data/ml_data) where you can find datasets relevant to the examples that exist in your private data directory.

  • For each booking, you start with a Python environment that has a complete environment for Machine Learning task with OPU, GPU and CPU

  • If you prefer SSH access to the compute you have to provide an SSH username and your public SSH key here. (The same SSH key is used for SFTP access to your home directory (at sftp.cloud.lighton.ai), please refer to the instructions here).

  • During each booking, the compute (OPU+CPU+GPU) is fully yours and you have “sudo” rights, but be aware that system modifications are lost at the end of each booking!

  • For “ root ” rights you have to use sudo -i or sudo [cmd] ( su - doesn’t work) .

Before booking a time slot, please take into account that in order to offer a satisfactory level of service to all our users and due to inherent limitations of the booking system, the following restrictions apply:

  • Booking duration: The minimum booking duration is 10 minutes and the maximum booking duration is 24 hours.
  • Booking starting time: You may not plan a booking for less than 10 minutes in advance (minimum booking start time = Now+10’). Additionally, a time buffer of 10 minutes exists between bookings for the same OPU. (e.g. if a booking ends at 15:00 a new booking may start after 15:10).
  • Booking modification: After a booking has been made, it cannot be modified. Alternatively, you may delete your booking and make a new one. (The credits from the deleted booking will be immediately added back to your account).
  • Booking deletion: You may delete a booking until 1 minute before the starting time. (e.g. a booking that starts at 15:00 cannot be deleted after 14:58).

Please, make sure that you have read the Terms of Service as well.

If you have been selected to use LightOn Cloud please refer to the following article First-time user? Start here!