Synopsys Cloud
Cloud native EDA tools & pre-optimized hardware platforms
Request a Free Trial →
Synopsys Cloud
Unlimited access to EDA software licenses on-demand
Expect the Unexpected
A deeply embedded and complex bug is discovered and it¡¯s going to require significant engineering rework. Coding rework starts, then the demand for simulation balloons, way beyond your current simulation capacity. Not enough compute, and insufficient licenses¡ Panic.
Or, maybe, imagine this: Today, your engineering stack offers enough compute and simulation licenses to cover predictable verification needs. Everyone is happy! The CEO and team have an off-site, and their new growth plans define a new product to add to the roadmap. Knock-on costs have been superficially explored, but the CFO is not planning on a major expense to expand on-prem compute capacity¡ Panic.
Both dilemmas could be real and many of you reading this will have come across variations of these themes. They all add up to the same thing: predicting capacity for compute, simulation, emulation, licenses, and FPGAs needs to be more like a science than an art.
This time we are going to focus on simulation, so the cloud is the obvious answer and would certainly help the engineering teams in these examples. However, saying ¡°let¡¯s move simulation to cloud¡± it is not the same as knowing how to do it. As Einstein said,
¡°Problems cannot be solved by thinking within the framework in which the problems were created.¡±
Synopsys Cloud Verification Instance, designed for simulation, is a new framework in which some of the old challenges can be managed in a more effective way, especially for smaller companies with engineering teams operating within very constrained resources and skill assets.
The Data Challenge ¨C Analytics to Predict, Avoid, and Mitigate
Before tackling new frameworks, the primary challenge for many organizations is assembling enough information to characterize what capacity uplift is required and how to justify moving workloads to the cloud. Good analytics contribute a great deal here as they can help predict actual capacity uplift requirements in both cases, even if the scenarios didn¡¯t give much warning of an impending issue.
Companies large and small seem to struggle to develop information that would give them insights into what they are doing: how many cycles have been used (doing what) and how useful have they been? Capturing this information is a great investment and would help offer insights to avoid capacity shortfalls, increase utilization, and avoid effort in areas that are not productive.
Assuming the need to move to the cloud is recognized and the management team is ready to make the investment in as low-risk way as possible, the Synopsys Cloud Instance for Verification considers all the elements required to create a self-contained environment, suitable for engineering teams wishing to make the transition.
The Inertia Challenge ¨C Where to Start?
The barriers to entry for cloud are sometimes perceived to be quite high. Most organizations have gotten used to the idea of putting highly confidential information in the cloud. Because of the huge investments made by them, a cloud vendor¡¯s security capabilities are likely to out-class engineering platforms running on in-house compute.
That said, most engineering cycles probably still run on-prem because it is considered too complicated or would require too much effort to make the transition to the cloud.
For smaller companies, moving simulation workloads to the cloud may not be considered simple and might raise valid concerns about data upload and download, storage, and access to data files for various workflows, debug cycle and power analysis, to name a few. Larger engineering teams might have the resources to manage all of this themselves; however, many smaller companies will not.
What is required is a well-thought-out environment which is designed with all the potential gotchas resolved. The primary objective of the Synopsys Cloud Verification Instance is to provide that ¡°ready-to-go¡± environment, which addresses the concerns teams may want answers to.
The Methodology Challenge ¨C What Happens Where?
From an engineering point of view there are many questions to be answered before simulation workflows can be moved to the cloud. ¡°Make your own¡± cloud solutions at the most basic level involve running a simulator in the cloud, which is something most EDA vendors can support. But what about related workflows that have to be integrated into that flow?
Ideally, you would want an integrated environment allowing you to manage your debug workflow dynamically alongside the simulation being run. What about power analysis? Is the flow capable of managing multiple power domains in the design using the Unified Power Format (UPF)?
The Capacity Challenge
Cloud offers opportunities to improve access, especially for smaller engineering teams which are resource constrained, with very limited compute or limited access to simulation licenses. Even if a smaller organization has funding to increase on-prem compute and licenses, it takes a good deal of time to find or build a location, purchase requisite hardware, and set everything up. This should have to be a problem engineering teams need to worry about.
What I want is fast access to the cloud when project milestones demand it.
Cloud offers a short-cut to getting much more compute and access to simulation licenses and should be the preferred route for smaller organizations. For this reason, it is also being tested out by larger organizations that might choose to run a hybrid environment to handle peaks in compute requirements that cannot be serviced by existing on-premise resources.
The Engineering Productivity Challenge
How do you reach first simulation faster? Once you have access to capacity in the cloud, there is a real opportunity to improve productivity. An integrated verification tools environment has been on the wish list of many engineering teams for a long time because it would iron out some of the complications of setting up an optimized environment. An integrated tools environment may also help reduce errors created by incorrect or non-optimal tools set-up.
Especially useful for small, or less skilled, teams is an environment with pre-configured flows that is offered by the Synopsys Cloud Verification Instance. Examples would be for debug or power management¨Ctools (simulator switches) and flows can be pre-configured to present engineers with a low-risk, productive environment to work in. Many engineering teams using on-prem engineering platforms would be very happy to see that level of tools and workflow integration.
Summary
Engineering teams struggling with constrained resources for validating their IP to check for bugs are likely to be less productive and run the risk of product delays as they struggle to reach a level of quality¨Cthe penalty is time to delivery.
It is possible bugs can remain undetected in code due to insufficient resources to sign off to a satisfactory level. Perhaps there just hasn¡¯t been enough time to complete all the simulations the team might want, since they have to meet key customer delivery deadlines. That¡¯s when senior management ends up making difficult decisions resulting in customers finding errata. Nobody wants that.
The Synopsys Cloud Verification Instance offers fast access to a complete environment which enables the team to operate at a higher level of sign-off confidence and lower level of risk, without having to invest in more on-prem compute resources. To ensure there is predictable ROI for using the cloud, analytics are available so the team can track the costs of accessing the instance.
As an example of the advantages of leveraging the cloud, consider Astera Labs. ¡°By using Synopsys Cloud and BYOC (bring your own cloud), we are able to complete our chips in 30% less time and with higher quality, which is a significant time-to-market and competitive advantage,¡± said Jitendra Mohan, the company¡¯s CEO and co-founder.
Indeed, it is time to think within a different framework.
Continue Reading
Blog
4 min read
/
Nov 18, 2024
AI Startups are Using the Cloud to Accelerate Chip Design and Time-to-Market
Blog
4 min read
/
Oct 21, 2024
Accelerating AI Chip Development and Reducing Risk with Pre-silicon, Secure, Cloud-based Emulation
Blog
4 min read
/
Sep 17, 2024
AI Chip Startups Turn to the Cloud to Compete
Tags:
Cloud,
Chip Design Insights
Blog
4 min read
/
Aug 26, 2024
Accelerating the Pace and Precision of AI Chip Innovation: Synopsys Expanding ZeBu Cloud Capacity
