VQV Signal
According to a post on X by @dnystedt, Tohoku University and Aisin have built a 16 nm AI chip with 32 MB of embedded STT-MRAM. The memory holds both a compact Linux boot image and Tiny YOLOv3 weights, so the chip needs no external DRAM or SSD. The post says it boots in 65.7 milliseconds, 39 times faster, and runs object detection at about 50.7 mJ per frame using weight sharing and mixed 8- and 16-bit math.
According to a post on X by @dnystedt, Tohoku University and Aisin have built a 16 nm AI chip with 32 MB of embedded STT-MRAM. The memory holds both a compact Linux boot image and Tiny YOLOv3 weights, so the chip needs no external DRAM or SSD. The post says it boots in 65.7 milliseconds, 39 times faster, and runs object detection at about 50.7 mJ per frame using weight sharing and mixed 8- and 16-bit math.
AI-assisted summary based on the listed source.
The chip packs 32 MB of embedded STT-MRAM holding both a compact Linux boot image and Tiny YOLOv3 weights, so it needs no external DRAM or SSD. It boots in 65.7 milliseconds, 39 times faster, and runs object detection at about 50.7 mJ per frame using weight sharing and mixed 8- and 16-bit math. Why it matters:...
The post presents instant-on, low-power AI as relevant for cars and sensors, where boot time and power draw are critical. Chip designers and makers of automotive or sensor hardware may find that keeping model weights on-chip removes the need for external memory. The source does not give the baseline for the 39x figure or independent test results, so those details are worth checking before drawing conclusions.
Hardware and robotics watchers may want to track whether this becomes a product, benchmark, or deployment signal.
VQV organizes public signals from inspectable sources. It does not independently verify the underlying report.
Signal Strength reflects source quality, relevance, freshness and evidence. Public Interest helps organize discovery; it is not proof of truth.
VQV surfaced this signal because it is recent, relevant to AI Chips, connected to Scobleizer News Pulse (X).
No login, cookies, social SDKs, or automatic posting.
