White Papers + Analyst Reports

A new white paper from Cambrian AI Research examines the growing momentum of the Poplar software stack and ecosystem, detailing how our customer-centric focus is driving software enhancements, supporting developers and benefitting from open-source contributions from the wider AI ecosystem.

“Developers require fast, scalable accelerators to handle the massive computational loads of larger models. But AI developers also need a robust development environment that meets them in their AI development journey," writes Analyst Karl Freund.

This research examines the Graphcore data center architecture that enables highly scalable parallel processing for Artificial Intelligence (AI) and High-Performance Computing (HPC).

The architecture encompasses efficient low-latency communications between Intelligence Processing Units (IPUs) within a node, within a rack, and across a data center with hundreds, and in the future thousands, of accelerators to handle exponentially increasing AI model complexity.

IPU-Fabric dynamically connects IPU accelerators with disaggregated servers and storage. Critically, this agile platform for parallel applications supports a comprehensive software stack to develop and optimize these workloads using open-source frameworks and Graphcore-developed libraries and development tools.

Software for new processor designs is critical to enabling application deployment and optimizing performance. Graphcore’s Intelligence Processing Unit (IPU) utilizes the expression of an algorithm as a directed graph, and the Poplar software stack translates models and algorithms into those graphs for execution.

This paper explores the benefits provided by Poplar software and discusses how these capabilities could speed development and deployment of applications that run on Graphcore IPUs.

This research paper from Moor Insights & Strategy, explores Graphcore's new 7nm Colossus MK2 GC200 IPU, the IPU-M2000 and IPU-POD systems which scale to supercomputing scale for enterprise AI infrastructure.  

Analyst Karl Freund describes our second-generation platform, the IPU-M2000, as having "greater processing power, more memory, and built-in scalability for handling extremely large parallel processing workloads."

Today’s CPUs and GPUs are holding leading innovators back from making breakthroughs in machine intelligence. What is needed is a new type of processor that can efficiently support much more complex knowledge models for both training and inference. This is why Graphcore have created the Intelligence Processing Unit – the IPU. The IPU is built from the ground up to meet these innovators’ needs and to help them make new breakthroughs in machine intelligence.

来自Moor Insights＆Strategy的这份研究报告探讨了Graphcore全新7纳米 Colossus MK2 GC200 IPU、IPU-M2000和IPU-POD系统，它们可以扩展到公司AI基础设施的超级计算规模。

分析师Karl Freund将Graphcore基于第二代IPU的IPU-M2000描述为：“具有更大的处理能力，更多的内存以及内置的可扩展性，可以处理非常大的并行处理工作负载。该公司现在将把未来押注在第二代IPU平台上 。IPU-M2000即插即用的构建块，可实现大规模的可扩展性，目前在业界是独一无二的。”

科技分析机构Moor Insights & Strategy在其5月26日发表的一篇白皮书《Graphcore的软件栈：Built to Scale》中探讨了Graphcore软件Poplar? SDK所带来的种种益处，并讨论了这些能力是如何加快在Graphcore IPU上运行的应用程序的开发和部署的。

Moor Insight & Strategy高级分析师表示：“Graphcore是我们目前已知的唯一一家将其产物扩展到囊括如此庞大的部署软件和基础架构套件的初创公司。”

现有的 CPU 和 GPU 正在阻碍领先的创新者在机器智能方面的突破，他们需要一种新型处理器，可以有效地支持更复杂的知识模型，并可以同时用于训练和推理。这就是Graphcore从零开始设计IPU（智能处理器）的原因，IPU可以满足创新者的需求，突破现有硬件对于算法和模型创新的限制瓶颈，帮助他们在机器智能方面实现新的突破。

戴尔EMC DSS8440是一款采用Graphcore技术的机器学习服务器。 该白皮书详细介绍了用于戴尔EMC DSS8440服务器的IPU硬件和软件技术，架构，工作原理等。

Moor Insights & Strategy? ? ?? ???? ?????? AI ???? ?? ????? ??? ???? ?????? ??? 7nm Colossus MK2 GC200 IPU IPU-M2000? IPU-POD ???? ?? ??? ????.  

?????? Karl Freund? ?????? 2?? ???? IPU-M2000? "???? ?? ?? ????? ??? ? ?? ?? ??? ?? ??, ? ? ??? ? ?? ??? ???? ??? ????. ?????? ?? ?? ??? ??? ???? ?? ??? ? ??? ?? ??? ? 2?? ???? ??? ??? ?????."?? ??????.

??? ???? ??? ?? ?????? ?????? ??? ?? ???? ?? ?????. ??? ??? ? ?????? ?????? ???? ?? ??? ?????? ?????? ??? ??? ?? ????. ?????? IPU(??????)? ????? ???? ??? ???(directed graph)?? ????, ??? ????? ??? ??? ????? ?? ???? ??? ?????. ? ?????? AI ? ?? ???? ?? ? ??? ????? ?? ?????? ??? ???? ??? ?????. ?? ??? ?????? ?????? ???? ???? ?? ????, ?? ???? ?? IPU?? ???? ??????? ?? ? ?? ??? ???? ??? ?? ?????.

??? CPU ? GPU? ?? ?? ?? ???? ?? ????? ???? ??? ??? ? ???? ???? ????.

??? ??? ??? ?? ?? ???? ?? ??? ????? ??? ? ?? ??? ??? ????? ?????. ??? ?? ?????? IPU(?? ?? ??: Intelligence Processing Unit)? ??? ?????. IPU? ?? ?? ??? ?? ??? ???? ??? ?? ????? ???? ??? ???? ??? ? ??? ?? ?? ???????.