➀ MediaTek's Dimensity 9500 chip is set to use TSMC's latest 3nm process N3P for production;

➁ The chip features a new all-core architecture, including 1 Travis+, 3 Altos, and 4 Gelas cores;

➂ It integrates a new microarchitecture Immortalis-Drago GPU, improving ray tracing performance and reducing power consumption, and supports full AI capabilities;

➃ It is expected to provide 100 TOPS of computing power and support 10667Mbps of LDRDR7Mbps 4.55 memory.

➀ Toshiba has introduced three four-channel dual-supply level-translator-transceivers targeting UART and SPI interfaces;

➁ These devices support level-up and level-down voltage translation from either of their dual power supplies in bidirectional communication systems;

➂ The 74AVC4T245FT and 74AVCH4T245FT models have four bidirectional level shift circuits, with the latter adding input-holds for correct output states;

➃ The 74AVC4T345FT model is designed for SPI interfaces with a single channel and a triple channel;

➄ The transceivers have direction and output-enable pins within the ‘A’ side voltage domain and are voltage tolerant up to 3.6V;

➅ They offer output strength from ±2mA to ±12mA and have a static consumption of 8μA max through the two power inputs;

➆ The packaging is TSSOP16B and operation is over -40 to 125°C.

➀ SureCore, a specialist in ultra-low power memory, has enabled KU Leuven to develop a neural processing accelerator chip for AI applications.

➁ The chip achieves dynamic power savings of over 40% compared to industry standard SRAM solutions.

➂ The technology has been demonstrated on more mature nodes, such as the 16nm process.

➀ ROHM推出了一款新的通用芯片电阻系列，以实现更小尺寸和更高的性能。

➁ 这个新系列包括高功率MCRS系列和低阻值、高功率MCRL系列，适用于汽车、工业和通信应用。

➂ 系列产品通过优化内部结构和新材料，提高了额定功率和温度系数，并支持小型化趋势。

➀ Enfabrica Corp., a promising startup, has raised $125 million in B轮融资 and attracted investment from AI giant NVIDIA, which is considered a competitor due to its AI network chip development that may challenge NVIDIA's Mellanox solutions.

➁ The company has recently completed its C轮融资, securing $115 million from heavyweights like Arm, Cisco, and Samsung.

➂ Enfabrica has developed its AI network interconnect chip, ACF-S, to address GPU network pain points and memory and storage expansion challenges.

➃ The company's ACF-S chip integrates CXL/PCIe switching and RNIC functions into a single device, eliminating the need for PCIe switches, NICs, or independent CPU connections to DRAM.

➄ Enfabrica's Millennium chip, announced at SC24, offers multi-port 800 GbE connectivity for GPU servers and provides unparalleled performance, scale, and elasticity for AI clusters.

➅ The company believes its interconnect technology will become the core of future GPU computing networks.