Graphcore created a completely new processor, the IPU, specifically designed for AI compute. The IPU's unique architecture lets AI researchers undertake entirely new types of work to drive the next advances in machine intelligence.
VENDOR WEBSITE VIEW PRODUCTWATCH WEBINAR
Graphcore's Intelligence Processing Unit (IPU) hardware and Poplar® software helps innovators create next generation machine intelligence solutions. The IPU is the first processor to be designed specifically for Machine Intelligence and offers significant performance advantages compared to other computing hardware, typically used inartificial intelligence. In addition to outperforming other technologies at today’s most common workloads, the Graphcore IPU has been architected to excel at next-generation AI applications – including highly sparse models.
The IPU-M2000 is a revolutionary next-generation system solution built with the Colossus MK2 IPU. It packs 1 PetaFlop of AI compute and up to 450GB Exchange-Memory™ in a slim 1U blade for the most demanding machine intelligence workloads. Designed from the ground up for high performance training and inference workloads, the IPU-M2000 unifies your AI infrastructure for maximum datacentre utilisation.
To date accelerations of algorithms have revolved around SIMD architecture GPGPU, causing many classes of problems to be ‘rephrased’ in a manner to suit it. They work well, but when state needs to be considered, such as MCMC (Markov Chain Monte Carlo), natural language processing, neural net modelling and many other increasingly distributed, federated, forms of calculations, other novel approaches such as MIMD need to be considered.
Our experts will discuss how the Graphcore Intelligence Processing Unit (IPU) enables finance firms to use models of greater complexity and iterate faster than is currently possible with legacy processors.
4 x Colossus™ Mk2 GC200 IPU
Direct connect or via Ethernet switches
Virtual-IPU comprehensive virtualisation & workload manager support
The IPU-M2000 has a flexible, modular design, so you can start with one and scale to thousands. Directly connect a single system to an existing CPU server, add up to eight connected IPU-M2000s or with racks of 16 tightly interconnected IPU-M2000s in IPU-POD64 systems, grow to supercomputing scale thanks to the high-bandwidth, near-zero latency IPU-Fabric™ interconnect architecture built into the box.
To help our clients make informed decisions about new technologies, we have opened up our research & development facilities and actively encourage customers to try the latest platforms using their own tools and if necessary together with their existing hardware. Remote access is also available
Boston are proud to be Diamond sponsors of NVIDIA's GTC21 (Digital). Read on to find out more about our partner sessions and speaking slots...