Fullscreen HTML5Med Res (??MB)HTML4

SILICON SILIC CHIP www.siliconchip.com.au Publisher/Editor Nicholas Vinen Technical Editor John Clarke, B.E.(Elec.) Technical Staff Jim Rowe, B.A., B.Sc Bao Smith, B.Sc Tim Blythman, B.E., B.Sc Technical Contributor Duraid Madina, B.Sc, M.Sc, PhD Art Director & Production Manager Ross Tester Reader Services Ann Morris Advertising Enquiries Glyn Smith Phone (02) 9939 3295 Mobile 0431 792 293 glyn<at>siliconchip.com.au Regular Contributors Dave Thompson David Maddison B.App.Sc. (Hons 1), PhD, Grad.Dip.Entr.Innov. Geoff Graham Associate Professor Graham Parslow Ian Batty Cartoonist Brendan Akhurst Founding Editor (retired) Leo Simpson, B.Bus., FAICD Silicon Chip is published 12 times a year by Silicon Chip Publications Pty Ltd. ACN 626 922 870. ABN 20 880 526 923. All material is copyright ©. No part of this publication may be reproduced without the written consent of the publisher. Subscription rates (12 issues): $105.00 per year, post paid, in Australia. For overseas rates, see our website or email silicon<at>siliconchip.com.au Editorial office: Unit 1 (up ramp), 234 Harbord Rd, Brookvale, NSW 2100. Postal address: PO Box 139, Collaroy Beach, NSW 2097. Phone (02) 9939 3295. E-mail: silicon<at>siliconchip.com.au ISSN 1030-2662 * Recommended & maximum price only. Editorial Viewpoint Encouraging chip developments I’m thrilled to see that Advanced Micro Devices (AMD) are making a comeback in the desktop/notebook CPU market with their Ryzen 3000 series of processors. These have finally overtaken the latest Intel chips in some benchmarks, and offer outstanding value for money. I was quite concerned in 2016 when they nearly went bankrupt, since without AMD, Intel would have a virtual monopoly in the CPU market, with little incentive to innovate. Recall that it was AMD who released the first 64-bit x86-compatible CPU in April 2003. Intel quickly implemented a compatible 64-bit scheme, lifting us from the looming 4GB memory access limit. AMD has also historically helped to keep CPU prices down. Their strategy to compete has been to offer almost as much performance as Intel chips at much lower prices. They also helped to popularise multi-core computing, as their first dual-core Athlon 64 CPU was released in April 2005, a full year before Intel brought their Core 2 Duo processor series to market. The current ‘race’ seems to be to see who can jam the most cores on a single chip. The current innovation is the idea of separating the chip cores themselves and the onboard I/O controller onto separate silicon dies, and bonding them together in a single package with very fast interconnects. That brings the possibility of using multiple core dies in a single chip, which is what AMD has done with the Ryzen 9 3950X, jamming 16 cores with 32 threads into a single package with a maximum ‘boost’ clock of 4.7GHz. It’s impressive engineering and no doubt Intel is rushing to leapfrog AMD. On a different topic, Xilinx recently announced their Virtex VU19P, a huge new FPGA (Field Programmable Gate Array). It’s built on TSMC’s 16nm process and boasts nine million logic cells, more than 16GB of onboard memory and 460GB/s memory bandwidth. It’s an astounding device. This is a vast (and expensive!) chip, with a die size of around 900mm2, capable of implementing 16 ARM Cortex A9 cores simultaneously. I’m told the logic compile/synthesis time for a design that fills the chip is around two days! A device like this would be very handy for those designing moderatelysized CPUs or very large scale logic devices. It’s a lot faster to test such a device by uploading it to an FPGA and then running tests on that, compared to software simulations. And you definitely want to test your design thoroughly before spending millions of dollars on having ASICs (Application-Specific ICs) made. So you’d need the latest cutting-edge FPGA. In even more pioneering news, MIT researchers and Analog Devices recently succeeded in building a 32-bit processor called the RV16X-NANO using carbon nanotubes. It has around 14,000 individual transistors made from semiconducting nanotubes. As with many other unproven new processor technologies, I am a little sceptical as to whether this will ever catch up with traditional CMOS logic in terms of performance at commercial scales. But the fact that a working chip has been made means that the technology is a lot closer to production than many other technologies, even if it runs at a rather pathetic 10kHz clock speed. For more information, see: siliconchip.com.au/link/aauy Printing and Distribution: Nicholas Vinen 24-26 Lilian Fowler Pl, Marrickville 2204 2 Silicon Chip Australia’s electronics magazine siliconchip.com.au