Streaming fabric accelerates data transport, says Synopsys
A streaming fabric to monitor silicon can accelerate data transport and “significantly reduce test time” said Synopsys. It enables real-time analytics for silicon health monitoring of increasingly large, complex designs and is claimed to shorten both silicon data access and test time by up to 80 per cent.
The fabric is generated by Synopsys TestMAX DFT (design for test) tool and part of Synopsys’ silicon lifecycle management flow. It is an on-chip network that quickly transports silicon data to and from multiple design blocks and multi-die systems which reduces the time to efficiently test and analyse the overall health of the chip for anomalies and failures.
Constant access and analysis of chip or multi-die system data based on parameters such as process, voltage and temperature, ensure silicon health. In large, advanced node designs, engineers typically incorporate data access into each design block and then connect the blocks to the chip-level pins. While traditional networks can be rigid and need considerable planning, the new streaming fabric technology requires minimal planning because it can be programmed to accommodate various block speeds and data interface sizes. To minimise the routing, the fabric supports simplified branching for blocks with small data interfaces. These capabilities ensure block-level data access methodologies are physical design-friendly, require minimal effort and provide the shortest amount of required access time, to reduce test costs. By simplifying branching connections, there is minimal physical impact on designs, enabling engineering teams to quickly deploy the streaming fabric using the Synopsys Digital Design family.
Data, including test programs, applied to chips or multi-die systems in the field must not cause the silicon to use too much power, which could either damage the part or invalidate results. The TestMax ATPG pattern generation determines power drawn at data application time and assigns results from the company’s PrimePower RTL-to-signoff power analysis technology. As a result, power drop is mitigated, avoiding incorrect silicon data results and damage, reported Synopsys.
“Efficient, cost-effective silicon data access is a fundamental requirement to achieving reliable device operation during their lifecycles, which is essential for high uptime of mission-critical applications,” said Amit Sanghani, senior vice president of the Synopsys Hardware Analytics and Test team. He added that the new streaming fabric and more accurate power-estimation capabilities support customers to meet these goals while achieving design and schedule targets.