Cloud Compute for Deep Learning, AI & HPC
GPU-accelerated virtual machines for data scientists to develop & train CNN models and perform high performance computing
Deep Learning, AI & HPC in the Cloud
3XS Systems AI & HPC workstations and servers are the gold standard when it comes to professional video applications. Powered by professional-grade NVIDIA RTX GPU accelerators, all the capabilities of our award-winning 3XS Systems are now available in the cloud from any device, anywhere. Cloud services enable any organisation to take advantage of highly responsive GPU-accelerated compute without the need to invest in on-premise hardware and its associated maintenance. Available in rental increments as small as one month, cloud services can be used as little or as often as needed.
The Scan Cloud service is designed from the ground up for two purposes. Firstly - to deliver GPU intensive application and workflows over a virtual infrastructure, commonly called vGPU - to any device, anywhere. Secondly - to be as simple and straightforward as possible. Many hyperscale cloud providers leave you largely to your own devices to understand which of their myriad offerings will work best for your workloads and projects and although short billing periods may sound attractive there are often extra costs you may not expect - such as uploading or moving data. Also, if you run into issues there is little or no support available.
Scan Cloud offers a refreshing new take on cloud - you pay a monthly fee only for the service you provision - with no hidden extras and we’re always on-hand to advice and support you - by a human cloud expert - not a chatbot.
| SCAN CLOUD VS HYPERSCALERS | ||
|---|---|---|
| SCAN CLOUD | Hyperscalers | |
| Usage | Tailored for GPU intensive workloads | Any workloads |
| Proof of Concept trial | Yes - free of charge | No |
| Data upload / download costs | No | Yes |
| Support from cloud experts | Yes | No |
| Billing timeframes | Per month | Per second |
| Suitability | Smaller organisations | Larger organisations |
Scan Cloud has been extensively tested during its development by digital artists, designers, engineers and more to ensure it delivers on every level and is almost indistinguishable from the powerful physical systems we offer. Once you’ve begun using a cloud service dedicated to GPU-accelerated applications, you’ll wonder how you ever coped being tied to a desk in a single location. Being able to work anywhere on a lightweight device normally not capable of dealing with such demanding workflows is a truly ground breaking experience.
Furthermore, we can also provide real-time collaboration via NVIDIA Omniverse Enterprise - it allows multiple users to work on the same project at the same time in real-time, so everybody can see the results and input. It also allows files from previously incompatible applications to be imported into a single collaborative environment.
| VISUAL WORKLOADS | COMPUTE WORKLOADS |
|---|---|
|
3D design |
Scientific Computing |
|
Animation |
High Performance Computing (HPC) |
|
Rendering |
Machine learning |
|
Video transcoding |
Deep learning |
CAE & CFD simulation |
Artificial Intelligence (AI) |
The Scan Cloud service has been created so you have an alternative to deploying physical workstation or servers. We won’t lie though - cloud can be daunting so whether you’re upgrading desktop devices or wanting an entire new network infrastructure, our in-house cloud experts are able to guide you through the entire process and minimise the stress.
Perhaps the first and most important thing to understand about cloud is that it is effectively a monthly subscription service. Traditionally for hardware upgrades, an organisation would use capital expenditure, or CAPEX - normally associated with one time major, long term expenses, but as cloud services occur as a monthly payment, these can be taken from operating expenditure, or OPEX - concerned with general day-to-day expenses - making planning and budgeting much easier. Learn more about how cloud can simplify your day-to-day business by watching the video opposite, or read our detailed cloud buyers guide.
Optimised for AI & HPC
We understand how important it is to tune virtual machines for each specific Linux-based application, so our cloud workstation options include pre-configured profile options and the ability to customise your own bespoke profile too.
Data Preparation
Data preparation is the process of readying data for the development, training, testing and implementation of an algorithm. It’s a multi-step process that involves data collection, structuring and cleaning, feature engineering, and labelling. These steps play an important role in the overall quality of your deep learning model, as they build on each other to ensure a model performs to expectations. We can provide profiles specifically tailored for data preparation with expanded storage.
Development
Once the data is ready it can be used withs the libraries and frameworks to begin building your AI models. Model development usually begins with smaller and lower complexity models with smaller data sets. However this can be a very iterative process very early on in the development model cycle. This development stage can be carried out on relatively low powered hardware – typically one or two GPUs is sufficient, although more GPUs will reduce the time to your ideal version.
Training
The training of deep learning and machine learning models happens with large datasets and is very iterative. As model developers train, calibrate and re-calibrate parameters and hyperparameters of the models, large data volumes are needed to be repeatedly passed through the models. This process can be very resource hungry in terms of compute, storage, memory, bandwidth. That’s why our training profiles have multiple GPUs with optimised hardware to help speed up training tasks.
HPC
NVIDIA GPUs have revolutionised high performance computing and big data, enabling far quicker time to results and analysis than when run on traditional CPUs. Our HPC virtual machines, powered by multiple NVIDIA GPUs, CUDA-X and NVIDIA HPC SDK, are available in a wide variety of specifications, each optimised for different levels of precision, allowing you to select the perfect configuration for your project.
Popular Languages, Libraries and Frameworks
Accelerated by NVIDIA GPUs
The most powerful professional data science and compute GPUs are the NVIDIA series. These GPUs, delivered virtually by Scan Cloud, bring a host of features to data scientists, enabling far greater control and capability.
NVIDIA CUDA
NVIDIA CUDA is a parallel computing platform and programming model for compute applications. With CUDA, developers are able to dramatically speed up computing applications by harnessing the power of NVIDIA GPUs. In GPU-accelerated applications, the sequential parts of the workload runs on the CPU – which is optimized for single-threaded performance – while the compute intensive portion of the application runs on thousands of GPU cores in parallel. When using CUDA, developers program in popular languages such as C++, Python and MATLAB.
NVIDIA GPU Cloud
The NVIDIA GPU Cloud (NGC) provides researchers and data scientists with simple access to a comprehensive catalogue of GPU-optimised software tools for deep learning and high performance computing (HPC) that take full advantage of NVIDIA GPUs. The NGC container registry features NVIDIA tuned, tested, certified, and curated containers for the top deep learning frameworks and applications. It also offers third-party managed HPC application containers, NVIDIA HPC visualisation containers, and partner applications.
Added Benefits of Scan Cloud
Each 3XS AI and HPC virtual machine is provided with at least 1TB of high performance SSD storage for the operating system and applications. Storage of project data is provided via AI and HPC-optimised storage nodes that can be accessed by the virtual machines in your account. Our application specialists will work with you to identify the optimum capacity for this shared storage, which can be easily scaled as your needs develop.
We offer cloud workstations as an alternative to our physical workstations, giving you the freedom to work on any device anywhere. All processing is done using server-grade hardware in secure datacentres and shared remotely with your device.
Like buying a physical workstation, we offer cloud workstations tailored to your needs, find out more and contact the cloud team using the link below.