Voltus^TM IC Power Integrity Solution is a power integrity and analysis signoff solution that is integrated with the full suite of design implementation and signoff tools of Cadence to deliver the industry’s fastest design closure flow. The aim of this blog series is to broadcast the voices of different experts on how design engineers can effectively use the diverse Voltus technologies to achieve high-performance, accuracy, and capacity for next-gen chip designs.

Have you ever spent long days or even weeks compiling and re-compiling ultra-large designs before a big tapeout? Have you experienced the frustration of random crashes or disk capacity issues killing a long run, leaving you no option but to start over from scratch? 

In the digital IC realm, System-on-chip (SoC) designs grow larger and more complex each year, and signoff tools like Voltus need to keep up with that aggressive pace. A power integrity solution must accurately debug and verify power and IR drop results, but that alone is not enough to support the size and complexity of present-day chips. Half-hearted attempts at job distribution fall short in the face of the hundred-billion node designs and storage requirements approaching hundreds of terabytes. While commercial and open-source systems are available for job distribution and resource management, the EDA domain presents unique challenges with the sheer size of its inputs and intermediate data. To meet these challenges, design engineers need a power signoff tool with scalable performance and high fault-tolerance, one that leverages all available resources for fast, reliable closure.

The new cloud-ready Extensively Parallel (XP) solution from Voltus has brought significant improvements in capacity and scalability, letting our customers innovate with highly complex designs while speeding up time-to-market. With adequate resources, Voltus-XP analyzes massive designs in record time, distributing tasks among thousands of CPUs while seamlessly processing terabytes of data.

The following Voltus-XP features allow design engineers to easily leverage their existing infrastructure and environment with minimal fuss:

• Compatibility with a range of existing grid management systems and simple integration with new ones
• Minimal machine and OS requirements
• No persistent services or costly pre-run initialization
• Dynamic scalability to leverage all available resources

The Voltus Extensively Parallel Technology

Voltus-XP has three core capabilities that uniquely address the challenges facing modern design engineering teams: 

Resource Management: Voltus-XP delivers a dynamic resource management system that acquires resources on-demand and efficiently distributes active jobs, with the ability to integrate tens to thousands of CPUs automatically as they become available. This means that work can begin swiftly, as soon as the initial set of resources are made available, rather than idly waiting for the full set.  Sometimes resource allocation is very slow, especially at times when many engineers are competing for shared grid resources.  Voltus-XP’s flexible resource scaling allows work to proceed even when limited resources are immediately available. When a larger fixed-resource footprint is required (for example during Voltus-XP simulation), it will pause the run until enough resources are acquired.

Job Management: Voltus-XP provides a highly resilient system for distributed jobs.  In any large distributed system, transient failure is nearly inevitable and automatic recovery is a necessity.  Voltus-XP will silently recover from certain types of failure with no user intervention, including resource (machine) loss and transient NFS failure. For other critical failures, such as running out of disk space part way through a run, Voltus-XP can resume the stopped job automatically once the issue is addressed and pick up where it left off.

Data Management: Voltus-XP optimizes the data flow through the distributed system as hundreds or thousands of concurrent jobs produce and consume data across the network.  Large data takes a toll on network throughput and storage capacity as both intermediate data and final results move through a distributed system. While massive runs are already possible, Voltus-XP is taking significant strides in continuing to improve both network and disk utilization.

A Customer-Centric Approach for Allocating Resources 

Voltus-XP does a great job of automating the configuration process for a specific run.  With minimal user input, it can be configured to leverage a particular set of resources.  But how should resources be chosen for a given run?  The machines being tasked are limited resources, and effective utilization is key.

Big machines are usually in high demand and are thus harder to reserve.  But when speed is the priority, concentrating work on fewer, more powerful machines may yield faster results with less network overhead. Small machines are often easier to acquire but less powerful. Spreading work across a large number of smaller machines might be less efficient due to communication overhead, but might still result in a shorter overall runtime simply due to their availability; since smaller machines are usually easier to acquire, there’s less time spent waiting for resources.

Voltus-XP’s configuration options make it easy to select the appropriate resources for a run.  Simple configurations generally include a single resource type from a single grid management queue with a parameter controlling how many instances are allowed.  More complex configurations might span multiple queues or even multiple resource types.  For example, a single configuration can specify that some resources be acquired from the grid, some from the local (launch) machine, and some through direct ssh or rsh access.  Since there is no wait for direct-access machines or local resources, such mixed configurations allow work to start immediately even as slower-to-acquire grid resources are being requested.

Fitting right in: Two Customer Cases

Let's take a look at two dynamic power examples of faster performance, featuring very large designs with several hundred million instances at the 7nm and 5nm process nodes from an early beta customer. The traditional production flow would take approximately ~10 days to complete. Voltus-XP finished the tasks in less than 26 hours—a ~10X speed up, resulting in a successful tapeout ahead of schedule.

Voltus XP is designed to leverage the existing infrastructure for the largest of jobs.  With flexible resource allocation, high-performance, and uncompromising reliability, XP is definitely the lean yet mean solution to tackle the next generation of customer designs.

- Tim Edwards

Related Resources

Voltus-XP: Flow and Features (Video)

For more information on Cadence digital design and signoff products and services, visit www.cadence.com.

About Voltus Voice

“Voltus Voice” showcases our product capabilities and features, and how they empower engineers to meet time-to-market goals for the complex, advanced-node SoCs that are designed for next-gen smart devices. In this monthly blog series, we also share important news, developments, and updates from the Voltus space.

Click Subscribe to visit the Subscription box at the top of the page in which you can submit your email address to receive notifications about our latest Voltus Voice posts.