artificial intelligence (AI) News - Page 95 of 121 - CNX Software

December 26, 2019June 9, 2020 by Siji Sunny - No Comments

MediaPipe is an Open Source Perception Pipeline Framework Developed by Google

MediaPipe is an open-source perception pipeline framework introduced by Google, which helps to build multi-modal machine learning pipelines. A developer can build a prototype, without really getting into writing machine learning algorithms and models, by using existing components. This framework can be used for various vision & media processing applications (especially in VR) such as Object Detection, Face Detection, Hand Tacking, Multi-hand Tracking and Hair Segmentation. MediaPipe supports various hardware and operating system platforms such as Android, iOS & Linux by offering API’s in C++, Java, Objective-c, etc. And this framework also capable of utilizing GPU resources. MediaPipe Components The framework is comprised of three major components A framework for inference from the pipeline data Tools for evaluation And a collection of reusable inference and processing components It follows the approach of Graph-based frameworks in OpenCV and all processing happens with the context of the Graph. The Graph contains a […]

December 13, 2019 by Jean-Luc Aufranc (CNXSoft) - No Comments

GreenWaves GAP9 IoT Application Processor Enables AI on Coin-cell Powered Devices

GreenWaves Technologies GAP8 multi-core RISC-V microcontroller was introduced last year for artificial intelligence (AI) at the edge at ultra-low power consumption. GAP8 AIoT SoC was shown to perform inference at 3.7mA against 60mA for the same workload on STM32F7 Arm Cortex-M7 MCU. The company has now expanded its GAP IoT application processor family with GAP9 that delivers five times lower power consumption compared to GAP8 microcontroller while enabling inference on neural networks 10 times larger. Greenwaves GAP9 will bring machine learning and signal processing capabilities to (coin cell) battery operated or energy harvesting devices such as IoT sensors in consumer and industrial markets, wearables, smart building, smart farming and so on. GAP9 is said to combine architectural enhancements with Global Foundries 22nm FDX process to achieve a peak cluster memory bandwidth of 41.6 GB/sec and up to 50 GOPS compute power while consuming only 50mW. The increased memory bandwidth (20x […]

December 9, 2019December 10, 2019 by Jean-Luc Aufranc (CNXSoft) - 4 Comments

Testing NVIDIA Jetson Nano Developer Kit with and without Fan

A few weeks ago I received NVIDIA Jetson Nano for review together with 52Pi ICE Tower cooling fan which Seeed Studio included in the package, and yesterday I wrote a getting started guide showing how to setup the board, and play with inference samples leveraging the board’s AI capabilities. I’ll now test the board with the stock heatsink in both 5W and 10W modes, and see if thermal throttling does occur, and then I’ll fit the tower cooling fan to find out if we can extract more performance that way and how much lower the CPU temperature is. Jetson Nano Stress Tests with Stock Heatsink Let’s install SBC-Bench testing utility,

wget https://raw.githubusercontent.com/ThomasKaiser/sbc-bench/master/sbc-bench.sh
chmod +x sbc-bench.sh

1 2	wget https://raw.githubusercontent.com/ThomasKaiser/sbc-bench/master/sbc-bench.sh chmod +x sbc-bench.sh

check it’s properly installed,

sudo ./sbc-bench.sh -m
Time        CPU    load %cpu %sys %usr %nice %io %irq   Temp
15:05:06:  922MHz  0.05   5%   1%   2%   0%   0%   0%  35.0°C
15:05:11:  922MHz  0.13   3%   1%   1%   0%   0%   0%  35.0°C
15:05:16:  922MHz  0.12   3%   1%   1%   0%   0%   0%  34.8°C

sudo ./sbc-bench.sh -m

Time CPU load %cpu %sys %usr %nice %io %irq Temp

15:05:06: 922MHz 0.05 5% 1% 2% 0% 0% 0% 35.0°C

15:05:11: 922MHz 0.13 3% 1% 1% 0% 0% 0% 35.0°C

15:05:16: 922MHz 0.12 3% 1% 1% 0% 0% 0% 34.8°C

and run it in 5W mode:

sudo nvpmodel -m 1

sudo ./sbc-bench

sbc-bench v0.6.9

Memory performance:
memcpy: 3685.3 MB/s 
memset: 8555.4 MB/s 

7-zip total scores (3 consecutive runs): 2877,2885,2854

OpenSSL results:
type             16 bytes     64 bytes    256 bytes   1024 bytes   8192 bytes  16384 bytes
aes-128-cbc     284837.64k   525113.11k   639412.05k   706251.09k   728449.02k   729841.66k
aes-128-cbc     284316.13k   525028.93k   634287.70k   704675.84k   728088.58k   728973.31k
aes-192-cbc     262002.90k   458230.17k   544725.93k   588999.68k   604075.35k   604607.83k
aes-192-cbc     261583.66k   458583.96k   538986.92k   588138.84k   602303.15k   604067.16k
aes-256-cbc     247370.60k   405101.35k   466444.29k   501432.32k   512816.47k   513370.79k
aes-256-cbc     247650.51k   405270.40k   469783.72k   502266.54k   513187.84k   512977.58k

Full results uploaded to http://ix.io/23rg. Please check the log for anomalies (e.g. swapping
or throttling happenend) and otherwise share this URL.

sudo nvpmodel -m 1

sudo ./sbc-bench

sbc-bench v0.6.9

Memory performance:

memcpy: 3685.3 MB/s

memset: 8555.4 MB/s

7-zip total scores (3 consecutive runs): 2877,2885,2854

OpenSSL results:

type 16 bytes 64 bytes 256 bytes 1024 bytes 8192 bytes 16384 bytes

aes-128-cbc 284837.64k 525113.11k 639412.05k 706251.09k 728449.02k 729841.66k

aes-128-cbc 284316.13k 525028.93k 634287.70k 704675.84k 728088.58k 728973.31k

aes-192-cbc 262002.90k 458230.17k 544725.93k 588999.68k 604075.35k 604607.83k

aes-192-cbc 261583.66k 458583.96k 538986.92k 588138.84k 602303.15k 604067.16k

aes-256-cbc 247370.60k 405101.35k 466444.29k 501432.32k 512816.47k 513370.79k

aes-256-cbc 247650.51k 405270.40k 469783.72k 502266.54k 513187.84k 512977.58k

Full results uploaded to http://ix.io/23rg. Please check the log for anomalies (e.g. swapping

or throttling happenend) and otherwise share this URL.

The temperature never went over 44.5°C, and no throttling occurred. tegrastats during 7-zip multi-core test:

CPU [100%@1428,100%@1428,off,off] EMC_FREQ 3%@1600 GR3D_FREQ 0%@76 APE 25 PLL@41.5C CPU@43.5C PMIC@100C GPU@43.5C AO@52.5C thermal@43.5C POM_5V_IN 3348/2567 POM_5V_GPU 0/0 POM_5V_CPU 1549/912
RAM 1211/3956MB (lfb 515x4MB) SWAP 0/1978MB (cached 0MB) IRAM 0/252kB(lfb 252kB) CPU [100%@1428,100%@1428,off,off] EMC_FREQ 3%@1600 GR3D_FREQ 0%@76 APE 25 PLL@41.5C CPU@44C PMIC@100C GPU@43.5C AO@52C thermal@43.5C POM_5V_IN 3348/2568 POM_5V_GPU 0/0 POM_5V_CPU 1549/912
RAM 1239/3956MB (lfb 508x4MB) SWAP 0/1978MB (cached 0MB) IRAM 0/252kB(lfb 252kB) CPU [100%@1428,100%@1428,off,off] EMC_FREQ 3%@1600 GR3D_FREQ 0%@76 APE 25 PLL@41.5C CPU@44C PMIC@100C GPU@43.5C AO@52.5C thermal@43.5C POM_5V_IN 3348/2569 POM_5V_GPU 0/0 POM_5V_CPU 1549/913
RAM 1265/3956MB (lfb 502x4MB) SWAP 0/1978MB (cached 0MB) IRAM 0/252kB(lfb 252kB) CPU [100%@1428,100%@1428,off,off] EMC_FREQ 3%@1600 GR3D_FREQ 0%@76 APE 25 PLL@41.5C CPU@43.5C PMIC@100C GPU@43.5C AO@52.5C thermal@43.5C POM_5V_IN 3348/2571 POM_5V_GPU 0/0 POM_5V_CPU 1510/914

CPU [100%@1428,100%@1428,off,off] EMC_FREQ 3%@1600 GR3D_FREQ 0%@76 APE 25 PLL@41.5C CPU@43.5C PMIC@100C GPU@43.5C AO@52.5C thermal@43.5C POM_5V_IN 3348/2567 POM_5V_GPU 0/0 POM_5V_CPU 1549/912

RAM 1211/3956MB (lfb 515x4MB) SWAP 0/1978MB (cached 0MB) IRAM 0/252kB(lfb 252kB) CPU [100%@1428,100%@1428,off,off] EMC_FREQ 3%@1600 GR3D_FREQ 0%@76 APE 25 PLL@41.5C CPU@44C PMIC@100C GPU@43.5C AO@52C thermal@43.5C POM_5V_IN 3348/2568 POM_5V_GPU 0/0 POM_5V_CPU 1549/912

RAM 1239/3956MB (lfb 508x4MB) SWAP 0/1978MB (cached 0MB) IRAM 0/252kB(lfb 252kB) CPU [100%@1428,100%@1428,off,off] EMC_FREQ 3%@1600 GR3D_FREQ 0%@76 APE 25 PLL@41.5C CPU@44C PMIC@100C GPU@43.5C AO@52.5C thermal@43.5C POM_5V_IN 3348/2569 POM_5V_GPU 0/0 POM_5V_CPU 1549/913

RAM 1265/3956MB (lfb 502x4MB) SWAP 0/1978MB (cached 0MB) IRAM 0/252kB(lfb 252kB) CPU [100%@1428,100%@1428,off,off] EMC_FREQ 3%@1600 GR3D_FREQ 0%@76 APE 25 PLL@41.5C CPU@43.5C PMIC@100C GPU@43.5C AO@52.5C thermal@43.5C POM_5V_IN 3348/2571 POM_5V_GPU 0/0 POM_5V_CPU 1510/914

Only two Cortex-A57 cores are used even under load, and power […]

December 7, 2019June 25, 2021 by Jean-Luc Aufranc (CNXSoft) - 3 Comments

AI inference using Images, RTSP Video Stream on NVIDIA Jetson Nano Devkit

Last month I received NVIDIA Jetson Nano developer kit together with 52Pi ICE Tower Cooling Fan, and the main goal was to compare the performance of the board with the stock heatsink or 52Pi heatsink + fan combo. But the stock heatsink does a very good job of cooling the board, and typical CPU stress tests do not make the processor throttle at all. So I had to stress the GPU as well, as it takes some efforts to set it up all, so I’ll report my experience configuring the board, and running AI test programs including running objects detection on an RTSP video stream. Setting up NVIDIA Jetson Nano Board Preparing the board is very much like you’d do with other SBC’s such as the Raspberry Pi, and NVIDIA has a nicely put getting started guide, so I won’t go into too many details here. To summarize: Download the […]

December 4, 2019December 4, 2019 by Jean-Luc Aufranc (CNXSoft) - 11 Comments

ROCK PI N10 RK3399Pro SBC Sells for $99 and up

Rockchip RK3399Pro processor is based on the popular Rockchip RK3399Pro hexa-core Arm Cortex-A72/A53 processor plus an embedded neural processing unit (NPU) delivering up to 3 TOPS for AI acceleration. So far you had to spend over $200 to get either Toybrick RK3399Pro board or 96Boards RK3399Pro development kit to get started with the processor. Some other companies announced their own RK3399Pro SBC such as Pine64 RockPro64-AI, or Khadas Edge board, but those are not available yet. But there’s now a more affordable Rockchip RK3399Pro SBC courtesy of Radxa’s Rock Pi N10 available on Seeed Studio in three variants: $99 model A with 4GB RAM (3GB for CPU/GPU + 1GB for NPU), 16GB eMMC flash $129 model B with 6GB RAM (4GB for CPU/GPU + 2GB for NPU), 32GB eMMC flash $169 model C with 8GB RAM (4GB for CPU/GPU + 4GB for NPU), 64GB eMMC flash Rock Pi N10 specifications: […]

December 4, 2019 by Jean-Luc Aufranc (CNXSoft) - 3 Comments

Think Silicon NEOX|V is the First RISC-V ISA based GPU

We are seeing more and more RISC-V microcontrollers and processors hitting the market, but so far they all lacked a GPU for 3D graphics acceleration. Think Silicon, the make of NEMA GPU for IoT and wearables, has now announced it will demonstrate NEOX|V GPU, the first RISC-V ISA based 3D, at the RISC-V Summit at the San Jose Convention Center, on December 10-12, in San Jose, California. NEOX|V key features: Parallel multi-core and multi-threaded architecture based on the RISC-V64GC ISA instruction set with adaptive NoC (Networks-on-Chip) Configurable from 4 to 64 cores Variety of cache sizes and thread counts organized in 1 to 16 cluster elements Variety of cluster/core configurations with compute power ranging from 12.8 to 409.6 GFLOPS at 800 MHz Support for FP16, FP32, and FP64 plus SIMD instructions Beside 3D graphics, the RISC-V GPU can also be used for machine learning, vision/video processing, and open GPGPU compute […]

December 3, 2019December 3, 2019 by Jean-Luc Aufranc (CNXSoft) - 6 Comments

Imagination Unveils IMG A-Series GPU Designed For Everything from IoT to Mobile and Server

Imagination Technologies has just launched IMG A-Series GPU which they claim is “The GPU of Everything” and “The fastest GPU IP ever”. IMG A-Series can be customized and scaled for various applications & markets from automotive, AIoT, set-top box, mobile, and server. Compared to the company’s earlier PowerVR 9Series GPU, IMG A-Series GPU delivers 2.5 times more performance, eight times faster AI processor, and 60% less power while running complex content with the same process node, area, and under similar conditions. IMG A-Series GPU supports the latest API standards including OpenGL ES, Vulkan, OpenCL, and Imagination provides a Unified AI API for use in combination with PowerVR neural network accelerators. It also offers 5x performance density compared to the best current shipping PowerVR devices and supports PVRIC4 lossless or virtually-lossless compression guaranteeing a 50% bandwidth and footprint reduction. The new GPU also leverages HyperLane Technology with up to eight individual, […]

November 29, 2019November 30, 2019 by Jean-Luc Aufranc (CNXSoft) - 24 Comments

Orange Pi 4/4B Board with RK3399, 4GB RAM Launched for $50 and Up

Orange Pi 4 is yet another Rockchip RK3399 SBC with 4GB RAM, while Orange Pi 4B adds Gyrfalcon Lightspeeur 2801A AI Accelerator. Both models were announced about 10 days but were not for sale just yet. The company has now launched the board on both Aliexpress and their newly opened Amazon store with three options: Orange Pi 4 with no eMMC flash – $49.90 Orange Pi 4 with 16GB eMMC flash – $59.90 Orange Pi 4B with 16GB eMMC flash, Gyrfalcon NPU – $69.90 Quick reminder of Orange Pi 4/4B specifications: SoC – Rockchip K3399 hexa-core big.LITTLE processor with two Arm Cortex A72 cores, four Cortex A53 cores, and an ARM Mali-T860 MP4 GPU with support for OpenGL ES 1.1 to 3.1 support, OpenVG1.1, OpenCL and DX 11 System Memory – 4 GB LPDDR4 Storage – Optional 16 GB eMMC flash, micro SD card NPU (Orange Pi 4B only) – Gyrfalcon […]