Snapdragon 780G Doubles AI Speed
Compared with last year’s Snapdragon 765G, Qualcomm’s latest mid-premium smartphone processor doubles AI performance and integrates a new CPU cluster that boosts multicore processing by 78%.
Compared with last year’s Snapdragon 765G, Qualcomm’s latest mid-premium smartphone processor doubles AI performance and integrates a new CPU cluster that boosts multicore processing by 78%.
Syntiant NDP120 Sharpens Its Hearing
Although mainly for speech recognition, Syntiant’s second-generation chip adds audio processing and can even handle simple object detection. In always-on mode, it consumes less than 500 microwatts.
Although mainly for speech recognition, Syntiant’s second-generation chip adds audio processing and can even handle simple object detection. In always-on mode, it consumes less than 500 microwatts.
Ceva Bluebud Streams Wireless Stereo
Bluebud is a turnkey solution for implementing true-wireless-stereo earbuds. It includes several of the company’s licensable cores along with software that supports Bluetooth Classic and the new Low Energy Audio.
Bluebud is a turnkey solution for implementing true-wireless-stereo earbuds. It includes several of the company’s licensable cores along with software that supports Bluetooth Classic and the new Low Energy Audio.
IBM Demonstrates New AI Data Types
An IBM test chip can train and inference neural networks with impressive efficiency. It uses smaller data types (8-bit FP and 4-bit integer) than most other AI accelerators yet maintains the same output accuracy.
An IBM test chip can train and inference neural networks with impressive efficiency. It uses smaller data types (8-bit FP and 4-bit integer) than most other AI accelerators yet maintains the same output accuracy.
MediaTek Sets the Smart-TV Standard
A powerhouse in the Android-TV segment, MediaTek has again raised the bar by releasing the MT9638, which features an AI accelerator. The chip can also be used in streaming dongles.
A powerhouse in the Android-TV segment, MediaTek has again raised the bar by releasing the MT9638, which features an AI accelerator. The chip can also be used in streaming dongles.
Lightmatter Shines in Data-Center AI
Lightmatter aims to deliver the industry’s first photonic AI processor. After successfully testing the prototype it introduced last year, the startup is scaling the architecture by 16x for its production system.
Lightmatter aims to deliver the industry’s first photonic AI processor. After successfully testing the prototype it introduced last year, the startup is scaling the architecture by 16x for its production system.
Rocket Lake Lifts Off From Sunny Cove
Intel’s new 11th generation desktop-PC processor finally replaces the Skylake CPU with the Sunny Cove microarchitecture, delivering a greater gain in single-core performance than its three predecessors combined.
Intel’s new 11th generation desktop-PC processor finally replaces the Skylake CPU with the Sunny Cove microarchitecture, delivering a greater gain in single-core performance than its three predecessors combined.
ST Raises IoT Power Efficiency
STMicroelectronics’ new STM32U5 microcontroller is a fast, power-efficient chip for consumer and industrial-IoT systems. Its 160MHz Cortex-M33 CPU, large memories, and integrated peripherals set it apart.
STMicroelectronics’ new STM32U5 microcontroller is a fast, power-efficient chip for consumer and industrial-IoT systems. Its 160MHz Cortex-M33 CPU, large memories, and integrated peripherals set it apart.
i.MX8ULP Uses FD-SOI to Cut Power
NXP has extended its i.MX8 series to serve battery-powered devices by replacing the i.MX7ULP. The new model upgrades to more modern CPUs including Cortex-A35 and Cortex-M33.
NXP has extended its i.MX8 series to serve battery-powered devices by replacing the i.MX7ULP. The new model upgrades to more modern CPUs including Cortex-A35 and Cortex-M33.
R-Car V3U Runs Lockstep Neural Nets
The new Renesas R-Car V3U delivers 66 TOPS for Level 2 and Level 3 ADASs. To support ASIL D functional safety, it employs hardware redundancy in all compute subsystems, including its DLAs.
The new Renesas R-Car V3U delivers 66 TOPS for Level 2 and Level 3 ADASs. To support ASIL D functional safety, it employs hardware redundancy in all compute subsystems, including its DLAs.
Sunway Gives China HPC Independence
Jiangnan Computing Lab recently disclosed details of Sunway, a made-in-China architecture that powers some of the world’s fastest supercomputers, including a next-generation exascale design.
Jiangnan Computing Lab recently disclosed details of Sunway, a made-in-China architecture that powers some of the world’s fastest supercomputers, including a next-generation exascale design.
AMD Milan Extends Server Lead
The third-generation Epyc processors get a big boost from the new Zen 3 CPU, enabling AMD to, for the first time, surpass Intel Xeon processors in both single- and multithread performance.
The third-generation Epyc processors get a big boost from the new Zen 3 CPU, enabling AMD to, for the first time, surpass Intel Xeon processors in both single- and multithread performance.
Driver Monitoring Makes ADAS Safer
A driver-monitoring system is essential for Level 2+ autonomous vehicles and can save lives in any car. With shipments set to soar in the next few years, chipmakers are focusing on this new application.
A driver-monitoring system is essential for Level 2+ autonomous vehicles and can save lives in any car. With shipments set to soar in the next few years, chipmakers are focusing on this new application.
MediaTek’s M80 Delivers mmWave
The new 5G modem provides the first real millimeter-wave competition for Qualcomm. It brings an extensive upgrade by offering faster downlinks and uplinks with support for 3GPP Release 16.
The new 5G modem provides the first real millimeter-wave competition for Qualcomm. It brings an extensive upgrade by offering faster downlinks and uplinks with support for 3GPP Release 16.
Deep Vision Optimizes AI Data Flow
The startup has developed the ARA-1 inference coprocessor for edge applications along with a broad software stack. The chip delivers strong performance on ResNet-50 and MobileNet at just 2W.
The startup has developed the ARA-1 inference coprocessor for edge applications along with a broad software stack. The chip delivers strong performance on ResNet-50 and MobileNet at just 2W.
