Frequently Asked Questions

What is GVM Server?

+
-
GVM Server is our hypervisor that was developed for enterprise-grade GPU virtualization. It is present in all of our servers and allows us to provide our customers with powerful GPU-accelerated compute to meet the needs of all of their high performance compute (HPC) workloads, like the development and implementation of artificial intelligence (AI) and machine learning (ML).

How does your trial work?

+
-
When submitting a request for our 7-day free trial there are a few steps that you have to take. Firstly, you will be asked to fill out a quick contact form. Assuming you have provided a valid email address, you will be contact by a trial onboarding representative. You will be asked a series of questions to help us understand your use case for the trial. The more detail you offer us, the more likely you are to qualify for the trial. Please note that the trial is intended for business use and not for personal use. In order to qualify you must have a compelling business case for using Arc's GPU cloud. Assuming you do qualify, we will create your GPU instance and send you the the access credentials

I've signed up for the free trial and haven't heard back?

+
-
Free trial submissions are only reviewed during regular business hours, Monday-Friday 9AM-5PM. If you have submitted a form  you will be contacted shortly. Please be patient. We receive many applications and process them one at a time. Assuming you qualify, the clock will be reset to 7-days whenever you first gain access to your trial instance.

Can Arc Compute support GPU virtualization for all the GPUs, such as Tesla V100, Tesla K80, P100, A100, etc?

+
-
Yes. NVIDIA’s enterprise GPUs are currently supported. Supported microarchitectures include Maxwell, Pascal, Turing, and Ampere.

Can GPUs be virtualized for specific jobs, based on allocation percentage? (example: allocate 30% of a GPU for a training job?)

+
-
Yes. Isolation of workloads can be accomplished via MIG or via traditional % timesharing (example: %30 scheduling share)

Can Arc's solutions support all the frameworks such as TensorFlow, PyTorch, MxNet, Caffe2 and traditional ML frameworks?

+
-
Yes, all of the frameworks mentioned are compatible, along with many others.

How easily can it be integrated with other platforms? Through what mechanisms?

+
-
Quite seamlessly through the support of a REST API. We can provide support on integration with existing platforms as needed.

Is it compatible with Slurm Workload Manager?

+
-
Yes. the Slurm Workload Manager may be installed under GVM Server (Arc's hypervisor).

Does Arc provide dedicated/static IPs for VMs?

+
-
We sure do!

What is a hypervisor?

+
-
A hypervisor, or a virtual machine monitor, is a piece of software that can be used to create and run multiple virtual machines (VMs) on a single host server.

A hypervisor treats computing resources – like CPU, GPU, and storage – as a pool that can be reallocated between new and existing VMs within the hypervisor.

Hypervisors allow systems to use more of their available resources and provide IT with increased mobility due to the created VMs being independent from the host hardware.

Hyperborea, our bleeding-edge hypervisor, combines traditional enterprise hypervisor features along with GPU virtualization which allows the use of a GPU to accelerate graphics or GPGPU (general-purpose computing on graphics processing units) applications running on VMs, at no additional cost.