I wrote a few posts on EDA and cloud computing –the latest was as recent as last week, where I posted about security in the cloud after a thread of comments on a previous post–. Bottom line was, EDA in the cloud will happen, because it’s a necessity.

Today, Synopsys announced that it will use AWS (Amazon Web Services) to provide a cloud computing solution for VCS simulation. Synopsys is not the first big EDA player venturing into cloud computing –Cadence already did–. But as the market leader, Synopsys sends a strong signal to the EDA community and its customers: EDA SaaS (Software as a Service) in the cloud is about to get very real.

Just need to figure out a good business model…

Related posts:

1. Cloud computing: an opportunity for EDA
2. EDA in the cloud: shall we be scared?
3. Why Synopsys buying Magma is good

I wrote, like many others, that cloud-based EDA solutions are inevitable: there is no magic algorithm that will reduce the ever-increasing complexity of designing and verifying a digital device (FPGA, ASIC, or SW/HW co-design). The only way to keep pace with the complexity is massive parallelism.

Some claim that EDA is not ready for cloud computing because it requires a lot of CPU power with very fast access to a massive amount of data. That is because EDA tools have not been designed to take advantage of very large clusters of machines with a relatively low bandwidth network, each machine having a fraction of the data. It will not be long before tools are re-architected for that purpose. E.g., physical and logical verification are the most obvious candidates to benefit from partitioning techniques and to become SaaS in the cloud.

The other obstacle to EDA in the cloud is not specific to EDA: security is the most cited reason to explain the resistance of potential customers. Semi conductor companies and design houses are reluctant to let their sensitive data go into a cloud they feel they have no control of.

These are the typical questions when security in the cloud is raised:

1. Who has privileged access to the data?
2. Which data encryption is used, and how are managed the keys?
3. Where is the data located?
4. Is the data segregated from other customer’s data?
5. Can the data be recovered in case of disaster?

The relevance of (1) and (2) is no different than when the data is managed internally. Topics (3) and (4) come up when customers feel safer with a precise hosting location, or by excluding some location (e.g., some foreign country). However the principle of data fragmentation hosted in different, non-predictable locations, makes the whole data safer, because breaching one or more data center is not sufficient to rebuild the complete file. This also answers question (5): assuming the complete loss of a few data centers, it is possible to reconstruct the whole data thanks to fragmentation and embedded redundancies hosted in the other data centers.

Customers will feel more confident if these questions are clearly answered by providers, and if independent security audits assess the quality of the services. Also the definition of widely accepted security certification would help the adoption of cloud services.

The reality is that cloud services, as other IT services, are the target of thieves and spies. There have been and there will be well-publicized security breaches in clouds (Gmail, Twitter), like there are many told and untold intrusions in private networks. I think it is misleading to believe that hosting one’s data in one’s own facility is any safer than relying on a well-vetted cloud: most of the cloud providers will be better at security than customers will ever be.

Related posts:

1. Cloud computing: an opportunity for EDA
2. Synopsys is getting into the cloud
3. Plunify, a glimpse at EDA in the cloud

ASIC and digital system design is a computing resource-hungry task that would certainly benefit from cloud computing. Still, EDA is mostly staying on the sideline while so many other industries are rapidly shifting towards cloud-based platforms.

Granted, there have been a few incursions of EDA into the cloud.

Synopsys and other EDA vendors have been using Amazon Web Services (AWS) to provide web-based training services. Both Synopsys and Cadence now prefer Xuropa as a web-based training platform, because Xuropa offers a service more tailored to EDA needs (e.g., input/output languages and format).  These services are Infrastructure as a Service (IaaS), i.e., it gives the customer the ability to use processing, network, and storage resources in a flexible manner.

Physware and Plunify are two startups sitting firmly on the cloud-based Software as a Service (SaaS) side. Physware simulates an IC with a true 3D field solver, providing signal integrity, power integrity, and electro-migration analysis. Plunify runs multiple synthesis scenarios to offer a wide area/performance tradeoff to the user. Both use cloud computing to offer a very short, scalable, turn-around-time to their customers. Cadence also has its own cloud-based SaaS offering, with mitigated success so far.

Oh yes, and IBM has been using cloud computing for its own EDA tools for years –with more than 20,000 cores, 150 Tb of memory, running 40,000 jobs per day.

But where does that leave the big three, Synopsys, Cadence, and Mentor, as provider of EDA solutions in the cloud? For instance, physical verification is known to be very tolerant to design partitioning, so a number of physical verification tools can easily take advantage of cloud computing. Logic simulation can also benefit from massive parallelism, even though it is more challenging –partitioning a test bench is trivial, but taking advantage of a design partition for logic simulation is tricky.

It is not like the Big Three do not know what is at stake: unless there is some revolutionary technology in the making, the largest SoC will simply exceed the capacity of today’s synthesis and verification tools. Which means that the semiconductor industry should be eager to access cloud-based SaaS. However there are a few obstacles on the way:

• Tools needs to be revamped to fit the cloud infrastructure. For verification and simulation, this is not a major bottleneck though.
• It is unclear what will be the business model of a SaaS EDA in the cloud. But this is a chance to propose new models that would lift the EDA market. Shall we charge by CPU/hour plus bandwidth? Or by the TAT reduction (the more servers, the smaller the TAT, the higher the fee)?
• Last but not least, security is a major obstacle for design houses to let their IP go into the cloud. But those that express their concerns about security are the same that have a private email in the cloud (Yahoo email, Gmail, Hotmail) and go happily shop on-line. Also data are arguably more reliable in the cloud because of the inherent redundancy required by fault-tolerant platforms.  With time people will come to accept that the cloud is secured enough.

Given that the cloud infrastructure becomes more mainstream thanks to the many open-source resources and accumulated experience in many industries, it is just a matter of time before EDA and its customers are serious about cloud-based SaaS. Let’s just hope that the EDA companies will size the opportunity to reiterate itself as a major enabler of the semiconductor industry, and to propose a new business model that would benefit the industry.

[UPDATE: follow-up post on security and the cloud]

Related posts:

1. Cloud computing is not grid computing
2. EDA in the cloud: shall we be scared?
3. Synopsys is getting into the cloud

It is good to learn that the customers of the EDA industry are doing better –if they do badly, EDA will do too. But will that seemingly economic improvement of the semiconductor industry translate into better days for EDA? Nothing is less certain. The recent quarterly reports of Synopsys, Mentor, Cadence, and Magma, although slightly above guidance, show a bleak outlook. Most of the book-to-bill ratios decreased, and they all carefully announcing a lean year ahead.

I recently ran into some acquaintance working for a leading semiconductor company (in the top 15), who told me that they are reaching out to services companies to get more values out of them. The numbers speak for themselves: they will put $12 millions down for an evaluation project that will encompass the full backend part of the design cycle –about 8 months project. That is only for an evaluation! When was the last time any EDA company was given that amount of cash for a real-life trial?

More numbers? Let us only look at the VLSI service companies in India, i.e., in no specific order: HCL Technologies, KPIT Cummins Infosystems Ltd, MindTree Ltd, Sasken Communication Technologies, Tata Consultancy Services, Wipro Technologies. According to the India Semiconductor Association, VLSI design service revenues in India could hit $1.13 billion in 2009, while hardware and board design could reach $560 million and embedded design and services about $7.29 billion. Yes, that’s nearly $9 billion overall, nearly twice the EDA market, and China is not even in the picture yet. Despite the dramatic downturn in 2009, some of these services companies did quite well, and most expect an uptick with a recovery in the semi industry next year.

The truth is that EDA companies have been providing software solutions that are more and more seen as commodities. The license renewal rate is dropping and its volume is decreasing.  In a flat, if not slowly shrinking market, the EDA firms have to eat their competitors’ share if they want to grow or just survive. They drop their prices and fork free AE support to sweet the deal for the customer. The vast majority of the designs can be done with last year’s generation suite, thus there is no urgency to buy new design tools. Then semiconductor companies might indeed be better off with a dedicated service company, which provides hands-on design expertise, and will be judged on results, i.e., the final tapeout. This is a win-win situation: the customer can fully rely on the service company, and since this business model commands a much higher fee than for a software license, the service company can expand and further invest to be an intimate part of their customers’ flows.

EDA has better look around and see what is happening. Semiconductors companies will more and more rely on service companies, tailored to their needs. Chip design and verification looks more and more like an IP assembly that requires an expertise that EDA tools do no longer deliver. The value-added is in that expertise, not in the tools that are becoming more and more push-process.

It is true that VLSI service companies buy tools to EDA companies, but the service companies factorize the license usage between several customers, which means that overall, less licenses are needed. Today, TCS can easily rent any EDA tool from the big 3 by the week or by the month. Imagine tomorrow Synopsys, Cadence, and Mentor dealing only with the top 6 hardware design service companies, themselves servicing the top 20-30 in the semi industry: EDA will loose a lot of leverage in the sale negotiation process. If Magma ends up as a cheap provider of IC implementation solutions, all developed in India, it will lower the bar even more.

EDA has to evolve quickly if it does not want to be sidelined as just an enabler. The EDA industry must be part of the design expertise, and work closely with its customers, even if it means its solution is no longer generic. And yes, as I said in the past, the value-added is in the system-level software, and this is where resides the growth of hardware designs. So the EDA industry must go into chip software design and verification if it wants to be relevant in five years from now.

No related posts.