EDACafe Editorial Roberto Frazzoli
Roberto Frazzoli is a contributing editor to EDACafe. His interests as a technology journalist focus on the semiconductor ecosystem in all its aspects. Roberto started covering electronics in 1987. His weekly contribution to EDACafe started in early 2019. AI calling for new fabs; EDA updates; workload-aware AVS; China’s Risc-V activity; Renesas adding GaN capabilityJanuary 25th, 2024 by Roberto Frazzoli
Artificial Intelligence keeps creating “circles” – that is, producing consequences that feedback on AI itself, further amplifying its impact. Besides the well-known chip performance “virtuous circle” (powerful processors enabling AI, and AI enabling even more powerful processors), another circle is looming – virtuous or vicious, depending on the point of view. It can be described as follows: chip fabs capacity enables the construction of large datacenters, and the success of AI applications fueled by large datacenters generates the demand for even larger fab capacity. Clearly, this insatiable hunger for GPUs is “virtuous” from the point of view of semiconductor sales, but can be considered “vicious” from an energy consumption standpoint. According to a CB Insights estimate, in 2024 three million operational H100 Nvidia chips – in datacenters around the world – will collectively consume almost 14 GWh, more than the whole Guatemala state (13 GWh). Sam Altman reportedly looking to spur the construction of new fabs According to two press reports (here and here), OpenAI’s CEO Sam Altman is trying to launch a network of new fabs to ensure his company an abundant supply of GPUs and reduce its reliance on Nvidia. Reportedly, Altman has been discussing the project with investors from United Arab Emirates and Japan’s SoftBank Group – as well as with potential foundry partners like TSMC, Intel and Samsung. OpenAI would be the new fabs’ primary customer. “Big tech” companies are the biggest buyers of Nvidia’s H100 GPUs. In a Thread post (quoted by Reuters), Mark Zuckerberg recently wrote “We’re building a massive amount of infrastructure. At the end of this year, we’ll have ~350k Nvidia H100s — and overall ~600k H100 equivalents of compute if you include other GPUs”.
EDA and other software updates: Cadence, Altium, Perforce, Segger Cadence has announced a new portfolio of domain-specific applications that enhance the capabilities of its Palladium Z2 Enterprise Emulation System. The “4-State Emulation App” enables acceleration of simulations requiring X-propagation such as for low-power verification of complex SoCs with multiple switched power domains; the “Real Number Modeling App” enables acceleration of simulations on mixed-signal designs; the “Dynamic Power Analysis App” is a massively parallel architecture for multi-billion-gate, million-clock-cycle power analysis of complex SoCs that – according to the company – is up to 5X faster than its previous versions. Altium has launched the BOM Portal within the Altium 365 platform. The BOM Portal aims to enhance collaboration between engineering and procurement teams, offering a unified approach to managing Bills of Materials in electronics design. With visibility into upcoming designs before they are released, procurement professionals can catch supply issues early in the development process. The BOM Portal on Altium 365 integrates with data sources like Octopart, SiliconExpert, and S&P Global (formerly IHS Markit). According to a study quoted by Altium, up to 80% of PCB designs require part replacements due to sourcing issues. Altium has also recently integrated its platform with Z2Data, an electronics supply chain risk management solution. This integration will provide Altium 365 users with direct access to Z2Data’s database of over a billion electronic components. Perforce has announced the availability of Helix Core Cloud, a Perforce-managed and hosted offering of their version control platform, Helix Core. Helix Core Cloud is now available as a turnkey, expertly pre-configured solution in the Microsoft Azure Marketplace for teams of fifty people and under. Segger’s new Embedded Studio IDE (V8.10) now also supports multiple architectures with a single setup. The same software can be used to build and debug programs for both Risc-V and Arm targets. Reducing power consumption with “functional-workload aware” adaptive voltage scaling ProteanTecs has launched its power reduction solution which uses on-chip telemetry, machine learning and predictive analytics to enable workload-aware system-on-chip power reduction during production testing and in-field operations. As for in-field power savings, the solution is based on proteanTecs’ AVS Pro, a closed-loop hardware-firmware application for reliability and functional-workload aware adaptive voltage scaling (AVS). Based on timing margin agents, AVS Pro leverages excessive guard-bands to reduce power while guaranteeing failure prevention. According to the company, this technology has enabled customers to reduce their power consumption by an average of 8-14%. For power reduction during production, proteanTecs offers applications for prediction-based VDDmin optimization per individual chip and system, with deep data analytics for process grading. These applications are deployed on the tester with advanced analysis on the proteanTecs cloud platform. Partnering with ATE vendors, the company enables parametric power and performance visibility for inline decision making. China’s Risc-V activity raising concerns An article published by the Jamestown Foundation – a Washington, D.C.-based conservative defense policy think tank – focuses on the role played by Risc-V in China’s technological ambitions. According to this analysis, China believes that it has advantages that allow it to gain a lead in the development and commercialization of Risc-V technology. Reportedly, there are now over 300 companies in China developing products using Risc-V technology, and the capabilities of the latest XiangShan processor designed by the Chinese Academy of Sciences surpass those of Arm’s Cortex-A76, making it the most powerful open-source core in the world. The article also highlights that Chinese entities in the Risc-V International Foundation outnumber those based in the United States, in terms of both high-level members (12 vs 7) and strategic members (49 vs 41). As for the potential use of Risc-V in military operations, the analysis quotes a paper from the Chinese “PLA Information Engineering University” (where PLA stands for People’s Liberation Army) describing the use of Risc-V to achieve a 700 percent acceleration in the AES algorithm. Acquisitions Renesas will acquire Transphorm (Goleta, California), a company designing and manufacturing high performance and high reliability GaN semiconductors for high voltage power conversion applications. With manufacturing operations in Goleta and Aizu, Japan, Transphorm produces the industry’s first JEDEC and AEC-Q101 qualified high voltage GaN semiconductor devices. Renesas will implement Transphorm’s automotive-qualified GaN technology to develop new power solution offerings, such as X-in-1 powertrain solutions for EVs, along with computing, energy, industrial and consumer applications. The acquisition of this GaN technology will add to Renesas’ previously announced establishment of an in-house SiC production line. Switzerland-headquartered Inficon – a provider of instrumentation, sensor technologies, and Smart Manufacturing /Industry 4.0 software solutions – has acquired all assets of FabTime (San Luis Obispo, CA), a niche provider of cycle time management software and consulting services to semiconductor manufacturers. Indian giant Infosys, a digital services and consulting firm, has announced a definitive agreement to acquire InSemi, an Indian semiconductor design and embedded services provider. Siemens has completed the acquisition of Heliox, a Netherlands-based specialist in DC fast charging solutions, serving eBus and eTruck fleets and passenger vehicles. People Synopsys’ executive Deirdre Hanford has been appointed as Chief Executive Officer of Natcast, a new, purpose-built independent non-profit entity created to become the operator of the National Semiconductor Technology Center (NSTC) consortium – once established, in the context of “CHIPS for America” initiatives. |