"We came to realize early on that as the deep learning landscape continues to grow and expand, there was a serious need for a cloud-based multi-GPU service for deep learning that wasn't limited to one or two GPUs", stated Mike LaPan, Director of Marketing and Cloud Services, Cirrascale Corporation. "Our new service, which makes use of our GX8 Series rackmount servers, meets this need by providing the ability to peer up to 16 GPUs on a single root complex for increased performance and scalability, while being cost effective in its overall rental model."
The company's Cloud service offers the ability for customers to load their very own instances of popular deep learning frameworks, such as Caffe, Torch, Theano and TensorFlow, or to choose an approved partner image. Because Cirrascale's offering gives users access to the raw horsepower of a modern multi-GPU system, it is also proving attractive to customers outside of deep learning. Blazegraph, creator of the industry's first GPU-accelerated high-performance database for large graphs, is the first ISV to provide a ready-to-run image on the service. Blazegraph Database, an ultra-scalable, high-performance GPU-enabled graph database with support for Blueprints and RDF/SPARQL APIs, takes advantage of the GX8 series power to accelerate various graph applications.
"Partnering with Cirrascale as its launches its GPU cloud services makes powerful graph database and analysis technologies accessible to those data scientists and organizations whose work is limited by traditional solutions", stated Brad Bebee, CEO, Blazegraph. "Together we offer great flexibility for exploiting the superior bandwidth to main memory and effective parallelism of GPUs to achieve graph application acceleration of between 10x-1,000x, with a graph traversal rate of 32 billion traversed edges per second (GTEPs)."
The Cloud service is available immediately with configurations supporting the latest GPU accelerators from NVIDIA, such as the NVIDIA Tesla M40 GPU Accelerators, Tesla K80 Dual-GPU Accelerators, and GeForce TITAN X GPUs.