Portable Stimulus Gains Momentum

Next month will see a significant milestone for Portable Stimulus. On September 15^th the review period for the Early Adopter release of the Accellera Portable Stimulus Standard (PSS) will close and with it the opportunity to make your voice heard. This is an exciting time for Breker, the market leader in this space for the past decade, and signals a time when the industry can transition from a technology only available to a few aggressive adopters, to making it available to the mainstream.

The year started off with a lot of momentum, especially after the highly successful events of DVCon. There, the training session was packed with attendees who wanted to know more about the upcoming new language. They came armed with questions about what the standard would provide, what tools might be a result of it and the scope of the impact that it will have on the industry. A couple of panels and technical sessions were also dedicated to it, and they too were well attended with the audience full of questions and anticipation.

The momentum continued at DAC when Accellera devoted most of its breakfast event to the standard. Unlike the DVCon sessions mainly staged by tool vendors, the Accellera panel was filled with some early adopters ready to talk about the impact and difficulties associated with the incorporation of this technology into their development flows. The panel was moderated by Tom Fitzpatrick, vice chair of the committee, and participants included Faris Khudakjie from Intel, who chairs the committee, Karl Whiting of AMD, and Qualcomm’s Sanjay Gupta.

As should be expected from such early days of a new capability, their messages were cautiously optimistic but also realistic about the challenges they face. Each clearly attempted to identify the places in their flows where the impact would be the greatest. Sanjay identified the gap between simulation and emulation and sees Portable Stimulus supplying the necessary bridge. Being able to generate testbenches that target specific execution platforms is one of the major thrusts of the effort and is often called horizontal portability. Faris sees the opportunity to add more automation into the process, something lost when the Intel group transitioned from internally developed tools to a UVM-based flow. Portable Stimulus combines notions of stimulus, checkers, scoreboarding and coverage all into a single model and directly enables users to target coverage holes in a logical and prioritized manner. Karl was optimistic about IP integration at the SoC level and the ability to bring software and verification groups together. Another focus of portable stimulus is that models are composable, which means that verification models developed for IP blocks can be directly used at the sub-system level and at the system level with no modification. This is what is referred to as vertical portability and will have a huge impact on overall productivity.

But it is equally important to look at the challenges they face. Karl said that change is difficult and that there is a barrier to adoption because the group that writes the spec is not the group that benefits the most. Faris also pointed out that it asks users to take a step back and think about the system level and this will be a lot of effort. Panelists also noted that they have used Portable Stimulus on small or pilot projects so far. Faris admitted that scalability is a key factor and they need to prove that the tools have the ability to scale with the problems. Karl seconded that sentiment after his group saw that the adoption of UVM at the IP level did not scale to the SoC level.

Another panel asked if Portable Stimulus could be the one ring to bind them all? While setting such lofty goals may not be attainable or even desirable for any language, it provided an interesting metaphor to discuss many aspects of the standard. Mark Glasser from NVIDIA talked about how this is a new abstraction for verification and the industry has to think beyond it being just a testbench language. He believes that it will impact large parts of both the design and verification flows of the future. Monica Farkash from NXP thinks that this is the opportunity needed to bring test-driven development into the semiconductor industry and as a result, verification engineers will rise to the forefront of the entire development process.

It was my privilege to have been sitting among such panelists, even if some of them have started to identify me as the ‘Grand-daddy’ of the technology (I must thank John Cooley for this nickname!). It is important to realize that Portable Stimulus is not just one abstraction, but several seamlessly tied together. Some of those abstractions, such as transactions and registers, are familiar to people, while others, such as hardware/software interfaces and resources, are new. These new abstractions are what will enable us to extend verification to the whole system and not just the detailed hardware.

We have a long way to go before the industry can realize the full benefit of this new abstraction and to devise methodologies that will provide all of the benefits that these verification models make available.  We took the next step along the journey and it is not the effort of a single company. It took tool developers and users partnering to ensure the right standard and solutions are created. We still have ground to cover to ensure that the final release of the standard meets all of your needs. Call us today if you have question and we will be happy to look at your requirements and see if they are covered by the committee draft. Together we can do this.

Tags: Accellera, AMD, dac, dvcon, Faris Khundakjie, Grand-daddy of Portable Stimulus, Intel, John Cooley, Karl Whiting, Mark Glasser, NVIDIA, portable stimulus, portable stimulus market leader, Qualcomm, Sanjay Gupta, uvm

This entry was posted on Thursday, August 24th, 2017 at 1:11 pm. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.