top of page


Designing Chips in the Cloud

Find out more about how Cloudwerx is assisting in cloud transformation for the EDA industry.

cloud logo white.png




Optimized. Reliable. Secure.

The Current Challenge

EDA tooling, particularly simulation, is a compute intensive task with traditionally on-prem HPC hardware. With ever increasing IC design complexity and reduced time to market expectations, cloud computing offers fast scaling of compute nodes in an elastic manner. That is, compute nodes are scaled-up when needed and scaled nearly down to zero when usage is low. Importantly, it can help EDA companies reduce time to market, by tapping into massive compute at scales that are often not replicated with on-prem resources.

IP security has been paramount in EDA space, and with advanced security management that cloud offers, switching to cloud with peace of mind has never been easier. Google Cloud provides state of the art security measures to lock down IP data as well as access control.

By decoupling compute-intensive resources from development resources, each remote engineer can be given a medium sized ‘cloud PC’ experience. When running simulations, the jobs can be offloaded to a scalable compute cluster that scales just-in-time. EDA customers can focus on IP development and optimizations for faster chip to market rather than managing IT infrastructure

CloudWerx: Your Cloud Transformation Partner

2 copy.png

Step 1 - Assess

  • CloudWerx’s expert teams start with understanding the customer’s EDA tooling, workflow and current pipeline. Any current challenges and suboptimality in the development process (such as access to engineering resources, remote engineers, file shares, etc.) are also identified.

  • Since each customer’s EDA tooling can differ, a customized GCP Cloud Migration Plan is developed in a tight feedback loop with the customer.

  • For development machines and engineering needs, if deemed optimal, we begin migrating on-prem development machines to cloud for:

CLOUDWERX icon case studies.png

Easier IT administration

CLOUDWERX icon case studies.png

Scaling VM type depending on engineer’s role

CLOUDWERX icon case studies.png

Engineers running EDA simulations, as the first step in moving to a central elastic compute cluster

CLOUDWERX icon case studies.png

Bringing the development tools, test vectors, designs and source code to the cloud, ready for next steps -- elastic compute cluster


Step 2 - Develop Proof of Concept + Production Architecture

  • Set up Google Filestore based architecture to share large EDA data amongst company employees and compute loads.

  • Design custom tooling that uses (gcloud CLI and API) that allows client to, in an automated, hands-free manner:

    1. Create a golden image based off of CentOS 8 with custom software, permissions, share points and security settings.

    2. Spin up a new VM, fully provisioned with new credentials, auto-mouted NFS-shares, EDA tooling installed and connected to appropriate license server, remote access enabled and appropriate security measures to avoid intellectual property loss.

    3. Ability to delete a VM simply by specifying the username for whom the VM was created in the first place.

    4. Ability to run arbitrary scripts and commands against all VMs in an automatic manner using Ansible scripting as backbone.

    5. Setup credentials

    6. Setup remote desktop for each user (Chrome Remote Desktop, or other) – disable clipboard sync if possible

    7. Ability to add users to spin up new instances with the ability for admin to login to any instances

    8. Disconnect web-access, other than connection to our Source Control Server

    9. Setup shared NFS drive between all instances, which is automatically mounted on startup.

    10. Setup logging of user access

  • Set up a SLURM to allow elastic scaling of compute nodes. Client’s tooling (such as Synopsys Execman) is configured to use said cluster.


Step 3 - Deploy + Deliver

  • Set up a project with required permissions, settings, logging, monitoring and billing is set up within Google Cloud Platform to host the said move of EDA workflow to cloud.

  • Move assets to the cloud in a staged manner to have minimal to zero impact on business and productivity.

  • Once users start using cloud based resources, legacy on-prem setup can be torn down.

  • Additional tooling for billing insights are set up based on the client's requirements and scope of usage.


Step 4 - Optimize + Maintain

  • Seamless billing and cost optimization will run through CloudWerx’ GCP Partner Managed Billing Platform providing a single point of contact and resource for all billing cost optimization needs.

  • Ternary, a build on GCP for GCP and Best-in-Class cost optimization tool will be set up and configured complimentary as part of the CloudWerx GCP Partner Managed Billing.

  • CloudWerx expert GCP engineers and architects will be available for consulting and review of all ongoing changes, further optimizations or maintenance of the deliverables.

  • After deployment, CloudWerx architects remain available to the customer to ensure the cloud experience scales optimally and fulfills all business requirements now and in the future.

  • As new technologies and architectures become available in the ever evolving cloud platforms, the customer is kept up-to-date on advancements and solutions that would further improve cost savings and increase customer’s productivity.

  • CloudWerx will perform Quarterly Reviews to identify optimization opportunities, review architecture best practices and train the customer on new features of the platform.


The Cloud is (finally) ready for chip design.

Designing semiconductor devices is hard; really hard. We get it. It’s a constant battle between innovation, schedule, risks and costs. At least now you don’t need to worry about creating specialized computing capabilities for hosting your EDA tools, IP and design flow – with some help from CloudWerx and Google. According to Cisco: “By 2021, 94-percent of workloads and compute instances will be processed by cloud data centers.” It’s time to leverage the power of this technology when designing semiconductors.


Faster time-to-market, lower costs

Key Differentiator 1: On-demand Compute Resources

Harnessing the flexibility of cloud-compute can give semiconductor development a dynamic access to more compute resources for peak load times

As a result, projects can achieve faster design cycles and reduce schedule risk -- speeding time-to-market and lowering total development costs. Cloud-based Electronic Design Automation (EDA) software hosting makes this a viable option. CloudWerx makes it secure. Buying your own server hardware limits your compute resources to those you have on hand from an earlier purchase decision. 


Upgrading hardware adds time, complexity, and expense. The flexibility of cloud computing provides state of the art resources on demand. Google and CloudWerx are ready to support successful chip design and implementation. Leading EDA companies like Synopsys now provide cloud-friendly licensing models. This means chip designs can enjoy cloud benefits so many other industries have already taken advantage of.

Key Differentiator 2: Time-to-Market

When design iterations happen faster, we get to market faster 

Chip design projects are notorious for challenging schedules. With computing resources capped to the compute resources available, locally or in the data center, engineering teams compete for server resources during peak load times. This can delay projects, costing millions of dollars a month in expense and lost revenue and sacrificing the competitive advantage of time. Cloud compute offers the ability to scale compute resources in minutes — and as the project ramps down, there are no lingering CapEx headaches from having too much compute.

Licensing, Security, and Scaling Accommodate Cloud and Hybrid EDA

Now, design teams can:

6 copy.png

Take advantage of peak compute scaling

any time for any team

8 copy.png

Securely enable use of the skills and knowledge of partners around the world


Operate with the same secure-by-default practices as


Utilize state of the art hardware and compute acceleration at all times

Key Differentiator 3: Enhanced Security and Collaboration 

Combine the best design infrastructure with the best security protection

It seems like we are constantly wrestling between multi-user and multi-location access versus security risks. Early on, a concern with the cloud were security vulnerabilities that came with such accessibility. Today, Google Cloud is a secure-by-design infrastructure architected by the teams accountable for the highest profile and most accessed websites in the world. Cloud resources architected and operated by leading experts of CloudWerx enable smooth collaboration while lowering the risk profile. This can improve security for semiconductor development projects, by utilizing the latest and best security that the industry has to offer. 


Partners at Your Side 

For the demanding, high value and complex infrastructure of chip design, you need a partner close at hand. CloudWerx brings legendary expertise and at-the-ready support for designing semiconductor devices in the cloud.

The Customer

NovuMind helps companies leverage the power of AI in their products and services.  Their NovuTensor chip is an industry breakthrough – designed to enhance business functions by “making things think.” They offer a wide range of deep learning tools to help apply NovuTensor in a multitude of operations, allowing businesses to optimize for AI computation.

The Challenge

Novumind had no prior GCP experience and was working under extreme time constraints (July- August)  to complete the PoC before their EDA tools license expiration. They are ramping up, requiring them to expand their environment very quickly but working with a globally dispersed team and only a single Systems Engineer for their entire infrastructure.

CloudWerx Solution

CloudWerx delivered a custom solution to create, deploy, and [eventually] destroy a 'CloudPC' experience for the globally-dispersed engineers. This solution increased security, lowered infrastructure costs, and improved the user onboarding experience. Within the CloudPC, Cloudwerx created a mechanism to allow users to install software packages without compromising security or system stability.

cloud logo white.png
 CloudWerx was engaged to design and implement a GCP PoC with performance and configuration in mind to balance performance, efficiency, security and cost.

Hybrid EDA Transformation Case Study: Novumind


Cloudwerx created a Custom ‘Cloud PC’ Experience for NovuMind Engineers and Administrators. 


Features of Engineer Experience:

  •  Each engineer is assigned a Cloud PC with persistent storage

  • Through this, they can access NovuMind assets, IP and Synopsys tooling

  • Access is available via Nomachine client or web browser


Features of Admin Experience:

  • Easily able to create new Cloud PCs 

  • Ability to use commands to update all Cloud PCs at once

  • Simple scripts to destroy old Cloud PCs to free up resources and archive data. 

CLOUDWERX icon case studies.png

System deployment time decreased from months to ~30 minutes while significantly improving user experience and enhancing information security for the client.

CLOUDWERX icon case studies.png

CloudWerx provided a detailed Optimization Report, giving NovuMind a snapshot of their baseline, and then a “Project Recap,” highlighting how CloudWerx improved performance, reduced staff workload, optimized the environment, secured the environment, and other notable impacts. 


"The entire project and experience from a business and technical perspective has been very smooth, professional and successful... [CloudWerx] took a really tight timeline, put together a great plan, and went to work efficiently executing to deliver in a narrow window."

Jeff Lien, Director of Engineering, Novumind

bottom of page