40154577-1-610x393-cloud-umbrella-businessmanI started 2013 very upbeat about EDA in the cloud. I pictured EDA very slowly moving to a SaaS business, using public clouds as a scalable infrastructure to adjust to the irregular computing resource requirements throughout the lifetime of a complex system design. I knew it would take years, but I felt that the moment was right.

I bet a year ago that a major semiconductor company would use public cloud to peak their compute power (e.g., for logic simulation, extraction, physical verification) and meet their schedule.

Well, so far I only saw a few attempts, and a lot of aborted discussions.

We first saw Cadence making some of its design solutions available in its private cloud, with only a handful of customers to date.

We all saw Synopsys putting its VCS solution available in AWS back in 2011. With exactly zero customer to show.

We saw Nimbic putting its SW in the cloud. With unclear outcomes.

What else did I see over the past 2 years (names purposely omitted)?

I saw a top-3 EDA vendor successfully conducting a real-scale experiment to reduce design validation from days to less than one hour, using 1000’s of cores dynamically allocated in a public cloud. But at the end, the viability of the solution depended on all the partners temporarily putting their IP in the cloud, and an agreement was never reached.

I saw a top-15 semiconductor company test-driving a cloud-based solution to transparently augment their compute power. This on-demand service allowed them to bring 1000’s of core fully operational in less then 15mn, ready for simulation and verification. But at the end, the old-fashioned way prevailed: buy more machines, even if that means that lots will stay idle; or simply miss the deadline.

I saw another semiconductor company opened to using cloud for the same purpose (peak compute power). But they eventually gave up, although that time for a good reason: economics. They are better off with their own high-performance data center where they can squeeze out every bit of performance of their very pricey EDA licenses. And we are talking serious high-performance computing: 2000 cores overclocked to 4GHz with liquid cooling, and a GPFS at 5GB/s for the file system. Pretty good setup, which even Amazon’s HPC cannot match.

I saw several EDA startups embracing the concept, but unable to drive more sales with it.

And I saw many potential customers turning down the prospect of EDA in the cloud, essentially for one and one reason only: security.

Of course, this is not always the real reason. Quite often the company’s IT department will work hard to obliterate the possibility of using a public cloud –because that would mean less power for them. But security is the sure topic to scare away the executives.

It does no matter if you demonstrate VPN, encrypted communications, with dynamically changing keys. It does not matter if you show that data-at-rest is always encrypted and is never in clear but in RAM. It doesn’t matter if you have a monitoring system that reports any activity that does not fall within some acceptable scenario. It doesn’t matter if the person arguing against cloud-based solutions because of security has no second thought sharing her credit card information for her on-line shopping. It doesn’t matter if every study shows that the most likely source of a security breach in a company does not come from some outside hackers, but from within: its own employees. Security is the bogyman in the world of semiconductor.

It is striking to see so many industries moving to SaaS and public cloud. One simple example: Big Data. Yes, it is a buzzword, but for many it is very real, and could not be achieved without software operating in public or hybrid clouds (don’t take my word for it, look it up). Seeing the complexity of HW/SW only increasing, I am a bit at a loss to justify the reluctance of the electronic design industry to embrace cost-efficient, flexible solutions.

So unless there is a pressing reason (read: the economics trumping the politics), it is unlikely that EDA evolves to SaaS in the cloud as quickly as I hoped. Looks like the whole industry needs some serious rejuvenation at the top.

Tags: , ,

10 Comments on EDA in the cloud still looking for takeoff

  1. Salut, Olivier! The biggest concern for cloud computing is security. The recent news involving security (Target, NSA, etc.) is not helping. But one area to consider, which is hard to gather data on, is in-house clouds. The big players who may be using them would probably be the last ones to go public with their experiences.

  2. You’re right about the big players with their own private cloud. Intel has 40,000 cores available in their private cloud. IBM has obviously its own private solution in addition with its public cloud offering.

    But I’m still puzzled with the smaller players, which could benefit a lot from EDA + public cloud.

  3. Patrick Groeneveld says:

    Salut Olivier.
    Yes, cloud/SAAS and EDA dont mix. Not in 2000 (when it was called internet-based CAD), not in 2013, and not in 2023 either. It was a misguided idea in any era. Let me play the big devil’s advocate here, as there are several good reasons for why EDA has it’s feet firmly on the ground rather than its head in the cloud:
    First, the EDA business is dominated by only a few hands full of big customers that all have huge private compute farms. I have not seen any movement to a public cloud, and until that happens EDA will not invest in it. It is telling that in all these years no EDA startup came up with a compelling cloud based business, so the major EDA companies have good reasons to comfortably ignore it.
    Second, SAAS makes poor business sense for high-touch tools like EDA. EDA makes good and predictable money selling multi-year license deals per seat, especially on seats that are not used. A large part of the license fee actually subsidizes support effort. Splitting it up into license and support cost would make the latter uncomfortably large. Moving to a pay-per-use SAAS model also makes income unpredictable and it would entice the customer to reduce tool usage.
    Third, cloud does not really solve a technical problem. Only few tools benefit from massive parallelism: mainly verification/analisys. For most other tools will not run any faster nor better in the cloud.
    Fourth, EDA tools require a significant engineering setup and support effort, something that does not go away by calling it cloud. In many cases the problem is not really run time, it’s the many iterations it takes to find and debug problems.
    Fifth: public clouds hardware are often not set up for the problem sizes that EDA needs. We need machines with 500G RAM and more, but hardly care about silly over clocking since the speed is IO limited anyway. Public clouds are mainly populated by bread and butter machines instead which is a different sweet spot.

  4. Hallo Patrick,

    You have some good points. Let me comment on them.

    (1) Yes, the top customers are big and have their own private compute farms, so they are not interested in using public cloud. However smaller companies or startups would benefit from CAD tools available in a public cloud. Why is there pretty much no new startups in HW and system design, while you have so many in mobile SW and web-based applications? True, the latter can claim a larger customer base. But isn’t that true that the cost of entering HW design is so high that it has become a real obstacle for design innovation? When was the last time you got a cool HW gadget that didn’t come from Apple, Samsung, Sony, or the like?
    (2) I can’t disagree with this. CAD companies are comfortable with multi-year licenses that provide them with a predictable cushion of revenue 3 to 5 years in advance. And seeing the lack of success of SNPS and CDNS with cloud-based solutions will not help.
    (3) I have to disagree here, and you kind of make my point. Indeed only a few tools benefit from massive parallelism: logic verification, extraction, physical verification. I would also throw timing sign-off in the batch. But verification is a huge contributor to the cost and time spent in design (70% is the commonly accepted figure). And by design, I really mean HW + SW: you want to be able to simulate the behavior of complex applications and get them right within a short amount of time (think mobile or gaming). If you’re not one of the top 10 guys, you will come short on compute power. This is where a cloud-based solution comes handy. However there are two problems: (a) security, and (b) I/O. I believe (a) is a perception issue, I won’t go through it. As for (b), that needs to be considered. If you were to use a public cloud to run your logic simulation, you’d have to move lots of data in and out of the cloud (10-100’s of GB). Physical verification is a one-time thing (more like 2-3 times), but requires 500GB to 1TB for large chips. As a base, AWS gives you about 38Mbps upload to and 19Mbps download from their cloud, so uploading 100GB takes you about 6 hours. This may not be practical if you need to do this every day, so you can get a dedicated 1Gbps line which allows you to upload your 100GB of data in less than 15mn. Then comes the I/O from within: verification can produce a lot of data, and you need to write it (and read it) fast. We’re talking GB/s, so your average 140MB/s local drive or 36MB/s NFS definitely won’t cut it. You need a high performance file system, like GPFS or some flavor of Gnutella, which can deliver 2 to 7 GB/s depending on the setup and application. That is a real bottleneck, because you need to build your own GPFS in the cloud, and that’s not easy with today’s public cloud. However we may have a healthy competition for the high-end compute cloud between AWS and Google. Last but not least, high-performance I/O is evolving fast thanks to SSD drives and Big Data requirements, so I expect substantial progresses here.
    (4) Yes, many tools requires AEs and manual tweaking. How good would that be for EDA and semi companies to be able to share live, interactive access, to some design data in a matter of minutes? How about giving a secured access to a 3rd party for debugging? With high-definition GUI and no noticeable latency (<300 ms), with nobody physically traveling? That’s exactly what you can deliver with a cloud-based solution. And yes, I claim it is more secured than 99% of in-house networking solutions, because that’s the bottom line of cloud providers. You can setup a temporary, secured environment, continuously monitored, and dismantle it once no longer needed. During that time, only a handful of people had access to the network, via keys that are useless once the setup is gone, to data that was never stored in clear. Also you can setup the machines anywhere in the world to reduce latency. Or to overcome some legal issues (that’s a topic that by itself deserves a dedicated answer…)
    (4-2) Also cost of deployment of SW solutions in the cloud is much cheaper than SW installation in some customer-specific environment.
    (5) You’ll find 500GB RAM machines in AWS’ HPC. I/O is another matter, as discussed above.

  5. From my conversations with EDA tool vendors, there are two areas that are most likely to see a breakthrough to the cloud.

    1. Newer and less popular (in terms of market share) EDA tools. If your product is not the customer’s first choice, it is much easier to have them try it on the cloud without installing. In addition, the cloud offers global reach to potential customers. So, it is mostly upside for a product with a small marketing budget. And, for smaller products that are not the bread-and-butter of the company, there is much less concerns about licenses, benchmarking, etc.

    2. Education and training. Current hands-on web-based training solutions already exist, and more are being developed. Here, security is not an issue.

    Victor Lyuboslavsky, creator of EDA Playground

  6. Evgeni says:

    I’m a system engineer in a small FPGA design group within a large corporation. A couple of years ago I implemented a tool running FPGA builds on AWS cloud. Since then, our group has been using the cloud once in a while when it runs out of capacity of the local server farm (6 high-end servers). Than happens when everybody is working on timing closure before product releases.

    As far as security goes, this tool had been reviewed by our corporate IT, and was OK-ed. They gave few recommendations, such as volume-level encryption requirement.

    So, as far as I can tell, the reason why cloud EDA doesn’t take off is not security, but the lack of “killer application”.

  7. Olivier,

    Thank you for the interesting post. You mentioned that top-3 EDA players conducted successful real-scale experiments and reduced validation time significantly. Could you please provide more details on this? What kind of design validation did you mean?

    Thank you, Ian

  8. Hi Ian,

    This was the final verification step before going to the foundry –full physical conformity. I cannot tell explicitly which of the top-3 EDA companies it is. Suffice so say it is the #1 in that field.

  9. Larry Drenan says:

    Hi Olivier:
    I appreciate your article here and, from what I observe working with customers, I think it fairly accurately describes what EDA customers are doing right now. Based on your last paragraph, I think you should actually be encouraged, because in this business economics will eventually top politics. One way or another. But there are a lot of “non-computer related costs” such as recreating one’s design environment in a cloud, and synchronizing information, that also factor in to these decisions.
    I do want to clarify one point about Cadence’s “private cloud” service, HDS. To date we have serviced over 80 different customers. Many have had contracts for 3 years or longer. This may or be not be considered “many”, but it is definitely more than “a handful”. However, our target thus far has not been a model to offer cheap computing, as much as to offer a reliable design environment with experienced CAD and IT support, for those who don’t want to spend the money on those things themselves. The economics of this are quite different from those of offloading peak computer needs.

  10. TonyA says:

    I bet you will lost, we should wait for more than 5 years to get a good.
    Up to now http://easyeda.com is the best cloud base EDA tool, but it is too young.

Leave a Reply