Making GPU computing more accessible, affordable and easy to use.
With Phoenix you can start training models on GPUs without the fuss of server management, SSH tunneling, or data and library management gymnastics.
Phoenix
$1.10/ hr
$2.46/ hr
$4.10/ hr
you can be up and running with just one line of command. Not only that,We provide instant access to different GPU providers at best prices to save you over 73% vs Azure, AWS, and GCP.
Use the full power of google colab, which allows you to write and execute Python in your browser, with
- Zero configuration required
- Easy sharing
- Access to GPUs free of charge
And whenever your colab resources are not enough for your research, work or your hobby you can scale your resources with Phoenix as simple as it can be.
Far faster than your laptop
Spin up a new dev environment for any sized project in seconds with prebuilt images. Phoenix’s own 35GB dev image starts in under 10 seconds. Scale your cloud resources up to 128 VCPUs and 512GB of RAM and up to 4 A100 GPUs with 40GB VRAM each. you won’t even remember it’s not your local machine.
CPUs up to
128 cores
Memory up to
512 GB
GPUs up to
4 X A100
Easy to use CLI
Launch a training job or a notebook, preloaded with whatever you need from your local computer with one line of command.
phxrun --cluster XCloud --flavor A100-Xlarge --name awesome-test python train-GPT3.py
bucket: awesome-A2E67BEED57D2DA57B3BD4BC
job: awesome-4206EFBD41ADE503C383450E In order to get job statuses, run: $ phx status
phxstatus
pooyan-8F234E2ED4C9F37259E6F1E39: DONE (exit code 0)
zeinab-B19647BB315FF26C6E224D7A6: DONE (exit code 0)
awesome-test-B19647BB315FF26C6E224D7A6: RUNNING
3 items returned
phxjupyter create --cluster arvan-cloud --flavor default-small --name awesome-jup
bucket: awesome-jup-EFD444C702CDDCFAD4C9
jupyter: awesome-jup-60FD74EB191C9CA49B26
Try to attach to your jupyter, run:
$ phx jupyter attach
In order to get jupyter statuses, run:
$ phx jupyter status
phxjupyter attach awesome-jup-60FD74EB191C9CA49B26
phx 2022/11/11 20:05:02 Copy http://localhost:8888/ into your Google Colab Local Kernel dialog.
Phoenix
Phoenix is a serverless computing service on top of a gpu marketplace which everyone can add their machines to the Phoenix network and:
- Rent out their GPUs to other
- Manage their compute resources more efficient
- Switch between their own resources or other provider with zero cost
- Access to wide selection of GPUs
EGG
Eggs allow you to effortlessly run parallel model training experiments across dozens of GPUs. Just provide the code repository and the location of the data, we handle the rest. No instance provisioning, environment setup, or worrying about turning it off when you're done.
Install Phoenix into your own server and get more out of your GPUs.
Hosting
Rent out your GPU whenever you don't need it, to make your server pay for itself. Or Transform your mining farm into a GPU training center and earn ~2x to ~4x more per gpu-hour than mining cryptocurrency. We connect you with customers and provide simple tools to streamline hosting. You set your own prices and schedules.
GPU resource management
Easily share on-premise or cloud GPUs with your team.
- Scheduling and automatic queuing resources
- Share resources between users on a team
- Create group-owned resources
- Use admin tools to monitor and control usage
Democratizing GPUs for AI community
Our unique approach combine serverless computing and GPU marketplaceWhich enable teams, company and organizations to benefit from.
Launch a training job or a notebook, preloaded with whatever you need from your local computer with one line of command.
- hybrid model for computation(on-prem machine & cloud machines)
- make cloud adoption easy and very logical
- Very easy to use and time saving for data scientists
- Dramatically, lowering the compute cost
- Raising income by switching from mining to renting out GPUs for deep learning
We think that it is a golden time for our solution because As The Ethereum Merge has obviated the need to secure the blockchain network, GPUs are no longer needed on a mass scale and will go for a significant discount.In this situation renting out the GPUs previously active on the ethereum network is a more profitable and eco-friendly choice.