Avnet Ultra96 was unveiled last year as one of the four 96Boards AI platforms designed to develop applications leveraging hardware to accelerated artificial intelligence workloads. The 96Boards CE compliant board comes with a Xilinx Zynq UltraScale+ MPSoC, 2GB RAM, a 16GB microSD card, WiFi and Bluetooth connectivity and more. The company has now launched an upgraded version with Ultra96-V2 featuring most of the same specifications but with a WiFi & Bluetooth module certified in 75 countries, industrial temperature range, and Infineon’s PMIC’s for additional power control and monitoring. Ultra96-V2 specifications: SoC – Xilinx Zynq UltraScale+ MPSoC ZU3EG A484 with four Cortex A53 cores, two Cortex-R5 core, Arm Mali-400MP2 GPU, and FPGA Fabric System Memory – 2 GB (512M x32) LPDDR4 Memory Storage – 16 GB MicroSD card + adapter pre-loaded with PetaLinux environment Video Output – Mini DisplayPort (MiniDP) Connectivity – Microchip ATWILC3000 certified 802.11 b/g/n WiFi 4 / Bluetooth […]
Aller Artix-7 FPGA Board with M.2 Interface Fits into a Laptop
A few days ago, we wrote about Nitefury M.2 card equipped with a Xilinx Artix-7 FPGA, and which you can connect to any laptop, board, a computer with a spare M.2 socket. It turns out Numato Lab has done something similar with the Aller board, specifically designed for development and integration of FPGA based accelerated features into other larger designs, and provided in a standard 2280 M.2 form factor M-key slot. Aller FPGA M.2 card key features & specifications: FPGA- Xilinx Artix-7 FPGA (XC7A100T-1FGG484C) with 101,440 Logic cells, ~126K Flip-flops, ~600KiB Block RAM, and 240 DSP slices System Memory – 2Gbit DDR3 (MT41J128M16JT-125:KTR) Storage – On-board 1Gb QSPI flash memory for FPGA configuration Host Interface – 4 lane PCIe Gen1 (2.5GT/s) via M.2 Connector Interface, M-Key Debugging – JTAG header for programming and debugging Security – 1x Trusted Platform Module (AT97SC3205) Misc – 100 MHZ CMOS oscillator, 1x RGB LED […]
ODROID-N2 GPU Drivers, Linux 5.0, and Impressive glmarks-es2 Score
ODROID-N2 was announced last February for $63 (2GB RAM), and $79 (4GB RAM), but Hardkernel was not quite ready to take orders at the time. One of the good news is that the 4GB RAM is now available for pre-order with shipping scheduled to start on April 3. Another good news is on the software side with Hardkernel having released the userland Mali-G52 Wayland driver. It does not work well with Linux 4.9 due to incomplete DRM implementation, but it goes work with Linux 5.0 plus some modifications as further discussed in the aforelinked forum thread. The screenshot above, courtesy of odroid forum member memeka , shows ODROID-N2 running Ubuntu 18.04 + Gnome3 + Linux 5.0 on top of Wayland with GPU drivers providing acceleration as shown by glmark2-es2-wayland test program. The benchmark results are pretty impressive:
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 |
======================================================= glmark2 2014.03+git20150611.fa71af2d ======================================================= OpenGL Information GL_VENDOR: ARM GL_RENDERER: Mali-G52 GL_VERSION: OpenGL ES 3.2 v1.r16p0-01rel0.2943fc4ef9657d91ee32c9a58dec6cd2 ======================================================= [build] use-vbo=false: FPS: 961 FrameTime: 1.041 ms [build] use-vbo=true: FPS: 1592 FrameTime: 0.628 ms [texture] texture-filter=nearest: FPS: 1491 FrameTime: 0.671 ms [texture] texture-filter=linear: FPS: 1477 FrameTime: 0.677 ms [texture] texture-filter=mipmap: FPS: 1524 FrameTime: 0.656 ms [shading] shading=gouraud: FPS: 1151 FrameTime: 0.869 ms [shading] shading=blinn-phong-inf: FPS: 1215 FrameTime: 0.823 ms [shading] shading=phong: FPS: 1043 FrameTime: 0.959 ms [shading] shading=cel: FPS: 1126 FrameTime: 0.888 ms [bump] bump-render=high-poly: FPS: 514 FrameTime: 1.946 ms [bump] bump-render=normals: FPS: 1976 FrameTime: 0.506 ms [bump] bump-render=height: FPS: 1777 FrameTime: 0.563 ms [effect2d] kernel=0,1,0;1,-4,1;0,1,0;: FPS: 1139 FrameTime: 0.878 ms [effect2d] kernel=1,1,1,1,1;1,1,1,1,1;1,1,1,1,1;: FPS: 383 FrameTime: 2.611 ms [pulsar] light=false:quads=5:texture=false: FPS: 2096 FrameTime: 0.477 ms [desktop] blur-radius=5:effect=blur:passes=1:separable=true:windows=4: FPS: 389 FrameTime: 2.571 ms [desktop] effect=shadow:windows=4: FPS: 788 FrameTime: 1.269 ms [buffer] columns=200:interleave=false:update-dispersion=0.9:update-fraction=0.5:update-method=map: FPS: 103 FrameTime: 9.709 ms [buffer] columns=200:interleave=false:update-dispersion=0.9:update-fraction=0.5:update-method=subdata: FPS: 129 FrameTime: 7.752 ms [buffer] columns=200:interleave=true:update-dispersion=0.9:update-fraction=0.5:update-method=map: FPS: 158 FrameTime: 6.329 ms [ideas] speed=duration: FPS: 356 FrameTime: 2.809 ms [jellyfish] <default>: FPS: 979 FrameTime: 1.021 ms [terrain] <default>: FPS: 52 FrameTime: 19.231 ms [shadow] <default>: FPS: 437 FrameTime: 2.288 ms [refract] <default>: FPS: 88 FrameTime: 11.364 ms [conditionals] fragment-steps=0:vertex-steps=0: FPS: 1769 FrameTime: 0.565 ms [conditionals] fragment-steps=5:vertex-steps=0: FPS: 1769 FrameTime: 0.565 ms [conditionals] fragment-steps=0:vertex-steps=5: FPS: 1853 FrameTime: 0.540 ms [function] fragment-complexity=low:fragment-steps=5: FPS: 1783 FrameTime: 0.561 ms [function] fragment-complexity=medium:fragment-steps=5: FPS: 1726 FrameTime: 0.579 ms [loop] fragment-loop=false:fragment-steps=5:vertex-steps=5: FPS: 1715 FrameTime: 0.583 ms [loop] fragment-steps=5:fragment-uniform=false:vertex-steps=5: FPS: 1650 FrameTime: 0.606 ms [loop] fragment-steps=5:fragment-uniform=true:vertex-steps=5: FPS: 1728 FrameTime: 0.579 ms ======================================================= glmark2 Score: 1119 ======================================================= |
I’ve never seen such as high score (1,119 points) on Arm hardware. […]
Cubbit Aims to Crowdsource the Cloud for Improved Privacy (Crowdfunding)
Storing data in the cloud is convenient since you have access it from anywhere with an Internet connection, but there are privacy concerns, and you may have to pay a monthly fee if you exceed you storage limit. Cubbit aims to reinvent the cloud by not storing files in corporate datacenters, but instead relying on a swarm of “Cubbit Cells” to deliver fully private and reliable cloud storage without monthly. You’d just need to pay for the boxes and potentially extra local storage, and then it’s basically free to use afterwards. Cubbit Cell hardware specifications: Processor – Dual core Arm Cortex-A53 processor @ up to 1.2 GHz (possibly Marvell Armada 3700) System Memory – 1GB DDR4 Storage – Built-in SATA drive Networking – 1x Gigabit Ethernet port USB – 1x USB 3.0 port Power Supply – 12V Dimensions – 160 x 142 x 56mm The hardware looks like a basic […]
Beelink GT1-K TV Box Features Amlogic S922X SoC, Android 9.0 OS
Amlogic S922 processor was first shown in roadmaps in early 2018, and since then we know there are at least two variants with Amlogic S922X and S922D processors both coming with Cortex A73 & A53 cores, and an Arm Mali-G52MP GPU, but the latter adds a Neural Network co-processor. So far the only hardware announced with Amlogic S922X was ODROID-N2 SBC, but since most Amlogic processors are primarily designed for TV box, it was just a question of time before we see some Amlogic S922X TV boxes hit the market, and the first model to surface is Beelink GT1-K “King” TV box running Android 9.0 OS. Beelink GT1-K specifications: SoC – Amlogic S922X hexa-core big.LITTLE processor with 4x Arm Cortex A73 cores @ up to 1.8 GHz, 2x Arm Cortex A53 cores @ 1.9 GHz, Arm Mali-G52MP GPU @ 846MHz; 12nm manufacturing process System Memory – 4GB LPDDR4 RAM Storage […]
AI Core XM2280 M.2 Card is Equipped with two Myriad X 2485 VPUs
AAEON released UP AI Core mPCIe card with a Myriad 2 VPU (Vision Processing Unit) last year. But the company also has an AI Core X family powered by the more powerful Myriad X VPU with the latest member being AI Core XM2280 M.2 card featuring not one, but two Myriad X 2485 VPUs coupled with 1GB LPDDR4 RAM (512MB x2). The card supports Intel OpenVINO toolkit v4 or greater, and is compatible with Tensorflow and Caffe AI frameworks. AI Core XM2280 M.2 specifications: VPU – 2x Intel Movidius Myriad X VPU, MA2485 System Memory – 2x 4Gbit LPDDR4 Host Interface – M.2 connector Dimensions – 80 x 22 mm (M.2 M+B key form factor) Certification – CE/FCC Class A Operating Temperature – 0~50°C Operating Humidity – 10%~80%RH, non-condensing The card works with Intel Vision Accelerator Design SW SDK available for Ubuntu 16.04, and Windows 10. Thanks to the two […]
T-bao T15A 15.6″ Portable Display Works with Phones, Computers and Game Consoles
In recent months we’ve seen several lightweight portable displays that works with phones or computers such as the 12.5″ ultra-lightweight LAPSCREEN display or TAIHE Gemini 15.6″ Full HD / 4K portable display. But AFAIK, the former is not available for purchase just yet, and the latter was launched via a crowdfunding campaign and shipping appears to be scheduled for May 2019. If you want something similar that ships today, T-bao T15A 15.6″ portable display may be an options. It comes with two USB-C ports one for power, one to connect USB-C devices like smartphones, and a mini HDMI port for laptops, computers, game consoles and so on. T-bao T15A display specifications: Screen – 15.6″ IPS display with 1920×1080 resolution, W-LED backlight, 178-degree viewing angle Ports 1x mini HDMI 2x USB type-C ports including one for host device input, one for power 1x micro USB port to connect keyboard and mouse […]
Amlogic S922D, A311D, and A311X AI Processors Feature a Neural Network Coprocessor
We already know Amlogic is preparing to launch a few new processors with Amlogic S922X Hexa-core Cortex-A73/A53 processor found in the upcoming ODROID-N2 board, as well as Amlogic A311D processor with a very similar features. At the time, I had been told A311 would include an NPU (Neural network Processing Unit) but with zero details. I still don’t have much details about the NPU inside A311D processor, except for performance of up to 5 TOPS (Trillion Operations Per Second). So it appears the same NPU will be found in Amlogic S922D processor that, based on the table above, has the exact same specifications as Amlogic S922X except for the extra Neural Network AI co-processor that delivers up to 2.5 TOPS (16-bit?) and 5.0 TOPS (8-bit?) A.I. performance. I’ve also been informed that another Amlogic A311X processor would also be introduced with the NPU. Both A311X and A311D are said to […]