Building A Micrometer-Level Displacement Sensor With 3D Printed Parts

A compact NVIDIA Jetson Orin NX carrier board designed for robots and UAVs

WeAct N006 is a compact 90x60mm carrier board for NVIDIA Jetson Orin NX 8GB/16GB modules, targeting robotics and UAV applications. It features HDMI output, Gigabit Ethernet, dual M.2 sockets for NVMe SSD and WiFi/Bluetooth, two MIPI CSI camera interfaces, three USB 3.2 ports, CAN Bus, GPIO, I2C, and UART connectors. Power input is 16V–28V via XT30 connector with protection against undervoltage, overvoltage, overcurrent, and reverse polarity. WeAct provides a GitHub repository with documentation and a getting started guide for JetPack 7.x on Ubuntu 24.04. The board is available on AliExpress for $110.50 and competes with similar offerings like the Aerium Lumen board.

Nvidia accidentally made the RTX 3080 too good, and it spent the next five years fixing that "mistake"

A data-driven analysis arguing that the RTX 3080 was an anomaly in Nvidia's GPU lineup — offering 83% of the flagship's CUDA cores and 81% of its memory bandwidth at $699, with a 65% performance uplift over its predecessor. Since then, Nvidia has systematically reduced the relative core counts and memory bandwidth on its 80-class and 70-class GPUs: the RTX 4080 dropped to 59% of flagship cores, and the RTX 5080 to just 49%. Memory bandwidth followed the same trend. Meanwhile, prices climbed ($699 → $1,199 → $999) while generational performance gains shrank. The piece concludes that Nvidia recognized the RTX 3080 disrupted its tiered pricing model and has been course-correcting ever since, making a repeat unlikely given Nvidia's dominance and focus on AI revenue.

Solution, Name that Ware May 2026 « bunnie's blog

The solution to the May 2026 'Name that Ware' challenge is revealed: a large PCB from a Rodgers Instrument Co. church organ. The board contains core memory used to store and recall organ stop settings. The photographed board is the largest of five, featuring 19×10 bit words (190 total, with 180 actually used across 200 cores). The unredacted version includes a patent number for further technical details, and the board is praised for its design-for-repair, self-documenting layout. No prize was awarded this month as nobody correctly identified the ware.

Your PC isn't getting slower, it's getting dustier, and there's a fix that takes 20 minutes

PC chậm đi thường do bụi bẩn tích tụ chứ không phải do phần cứng cũ đi. Bụi ngăn luồng khí làm mát, khiến CPU/GPU giảm xung nhịp do quá nhiệt, gây tụt khung hình. Vệ sinh 20 phút (tắt nguồn, xả tụ, thổi khí nén ngược chiều vào quạt tản nhiệt) có thể khôi phục hiệu suất. Theo dõi nhiệt độ bằng MSI Afterburner để kiểm chứng, đồng thời sử dụng quạt hút khí lọc bụi để hạn chế bụi bám sau này.

Lập trình viên nên đọc bài này vì việc làm sạch bụi trên máy tính có thể giải quyết những rắc rối về hiệu suất và nhiệt độ làm việc lâu dài, giúp máy hoạt động ổn định hơn trong các dự án cần tốc độ cao.

Clicks shows off its BlackBerry-inspired phone in a new hands-on video

Clicks Technology has released a new hands-on video showcasing the Clicks Communicator, a $499 BlackBerry-inspired smartphone set to ship in Q4. The device features a tactile touch-sensitive keyboard, a customizable 'Signal Light' for notification management, swappable back covers, a 3.5mm headphone jack, physical SIM tray, eSIM support, and expandable microSD storage up to 2TB. Running Android with the Niagara Launcher, it targets productivity-focused users and those seeking a less addictive smartphone experience. The video previews pre-production hardware and software ahead of the full launch.

GCC 17 Compiler Lands SpacemiT X100 Core Targeting

GCC 17 has gained upstream support for -mcpu=spacemit-x100 and -mtune=spacemit-x100 targeting the SpacemiT X100 RISC-V core. The X100 makes up the primary cores of the SpacemiT K3 SoC, the first widely-available RVA23 profile-compliant RISC-V design, featuring eight X100 cores at up to 2.4GHz alongside eight AI-focused A100 cores. A follow-up patch for the spacemit-a100 target is also on the GCC mailing list and expected to be upstreamed soon. The A100 is an in-order, dual-issue core with VLEN=1024 vector length and the xsmtvdotii extension.

Intel Kills Off AMX-TF32 Support Before It Even Shipped In Diamond Rapids

Intel has removed AMX-TF32 from its ISA programming reference manual before the feature ever shipped in hardware. AMX-TF32 was a planned extension to Advanced Matrix Extensions (AMX) that would have natively supported NVIDIA's TensorFloat-32 format, offering FP32 range with FP16-like performance for AI/matrix workloads. The feature was previously confirmed for Diamond Rapids Xeon processors expected in 2027. Intel is now patching GCC to remove AMX-TF32 support, citing no actual hardware implementing the feature. This follows a similar fate to AMX-TRANSPOSE, which was also cancelled last year. Back-ported patches are needed to remove AMX-TF32 from GCC 15 and GCC 16.