It is 35 years since the Acorn RISC Machine chip sipped power for the first time
- Reference: 1588002606
- News link: https://www.theregister.co.uk/2020/04/27/35_years_arm/
- Source link:
"At 1pm on April 26th 1985," [1]recalled Prof Steve Furber, "the first ARM microprocessors arrived back from the manufacturer – VTI [VLSI Technology, Inc]. They were put straight into the development system which was fired up with a tweak or two and, at 3 pm, the screen displayed: 'Hello World, I am ARM'."
[2]Furber and fellow engineer Sophie Wilson had enjoyed considerable success back in the 1980s at Acorn Computers – Britain's answer to Apple – thanks to the BBC Micro, based on the ageing MOS 6502 processor. Famously, the duo had managed to put together a prototype microcomputer in a week or so to satisfy the BBC, leading to the [3]launch of the eponymous home computer in 1981.
By 1983, recognising that the 6502 was coming to the end of the road, the [4]Acorn RISC Machine project was kicked off. Wilson had created a simulation of the 32-bit microprocessor's instruction set in BBC BASIC while Furber focused on the hardware architecture, which featured a three-stage pipeline and barrel shifter. They assembled a small team, and turned to chip fabricator VLSI to churn out [5]the first samples of the silicon. The first chips worked as expected, 35 years ago.
Acorn co-founder Hermann Hauser [6]recalled : "While IBM spent months simulating their instruction sets on large mainframes, Sophie did it all in her brain."
Incidentally, the first [7]ARM1 chips required so little power, when the first one from the factory was plugged into a development system to test it, the microprocessor immediately sprung to life up by drawing current from the IO interface – before its own power supply could be properly connected.
[8]#OnThisDay in 1985 the first ARM processor was powered up! A few hours later, after a tweak or two, the screen displayed: 'Hello World, I am ARM'. [9]https://t.co/TDIkJjTJty [10]pic.twitter.com/ZrdWBbBOZP — Computing History (@computermuseum) [11]April 26, 2020
The early chips, which executed roughly three to eight million instructions per second, outperformed Intel's 80286 while sipping less current. Performance would continue to be improved over subsequent years, though the chip was not enough to save the fortunes of Acorn, which had been snapped up by Olivetti as the first Arm silicon arrived.
The first-generation Arm chip found use as a co-processor for the BBC Micro and a follow-up, the ARM2 and subsequent iterations found their way into Acorn's Archimedes and RiscPC lines of computers.
Famously, despite the lack of wild commercial success enjoyed by Acorn's personal computers, Apple began sniffing around the technology as something that could be used to power its Newton computing platform. The result was the spinning off of the bits of Acorn's Research and Development team responsible for the RISC microprocessor into a new company, Advanced RISC Machines, in November 1990. Wilson stayed on to consult (and would later turn up in an architectural role at Broadcom in Cambridge.)
You can watch a fascinating video of Arm's pitch to Apple engineers in 1992 here, courtesy of The Centre for Computing History:
[12]Youtube Video
The Newton, which eventually used the ARM610 and shipped in 1993, was infamously [13]not Apple's most successful product , and the PDA was axed in 1998.
However, the experience taught Arm a person cannot live on Apples alone (at least, not then), and under then CEO Robin Saxby, the design house pursued its unusual (for the time) IP business model with gusto. That model served the company well and Arm's CPU and GPU blueprints have been licensed to many semiconductor companies in return for an upfront fee and per-chip royalties thereafter.
For example, even as the Newton fiasco was sputtering to a halt, Arm's designs were creeping into other devices, most notably everyone's favourite Snake game console: [14]1998's Nokia 6110 , which was the first GSM phone powered by Arm's technology.
Arm went public on April 17, 1998, at a share price of 575p. The dotcom boom inflated things briefly before Arm (and much of the tech industry) returned to Earth. However, Arm's march to dominance was well underway thanks to its licensing model that permitted its tech to be integrated into more and more system-on-chips.
The addition of the Cortex line of processors saw Arm's designs dominate the portable device arena by the time Japan's [15]SoftBank acquired the company in 2016 . SoftBank went on to [16]sell off part of Arm to a China-led investment group in 2018.
The low-powered chip designs have also found their way into more traditional laptops and computers, as well as servers. After his stint at Acorn, Furber went on become a professor at the University of Manchester in England, and most recently used thousands of ARM9 processors in 19-inch server racks, in an effort to build a computer to [17]mimic the workings of the human brain .
As for the future, Arm has seen off much of the competition over recent years or simply licensed its technology to others. However, while the likes of Intel has failed to wrench themselves free of its legacy x86 and cede the mobile market, Arm's business model faces [18]a real threat from the open-source [19]RISC-V architecture.
For this hack, however, the "A" in arm will forever be "Acorn". Thirty-five years on, please join us in raising a glass to the boffins behind that first bit of silicon. ®
Sponsored: [20]Legacy Modernization: Finding Your Way With Low-Code
[1] http://www.computinghistory.org.uk/det/5440/First-ARM-Processor-Powered-Up/
[2] https://www.theregister.co.uk/2012/05/02/unsung_heroes_of_tech_arm_creators_sophie_wilson_and_steve_furber/
[3] https://www.theregister.co.uk/2011/11/30/bbc_micro_model_b_30th_anniversary/
[4] https://www.theregister.co.uk/2012/05/03/unsung_heroes_of_tech_arm_creators_sophie_wilson_and_steve_furber/
[5] http://www.computinghistory.org.uk/det/57310/Engineering-Sample-of-ARM-1-Processor/
[6] http://www.computinghistory.org.uk/det/6615/Sophie-Wilson/
[7] https://en.wikichip.org/wiki/acorn/microarchitectures/arm1
[8] https://twitter.com/hashtag/OnThisDay?src=hash&ref_src=twsrc%5Etfw
[9] https://t.co/TDIkJjTJty
[10] https://t.co/ZrdWBbBOZP
[11] https://twitter.com/computermuseum/status/1254337940890816512?ref_src=twsrc%5Etfw
[12] https://www.youtube.com/watch?v=ZV1NdS_w4As
[13] https://www.theregister.co.uk/2007/06/29/ft_newton_messagepad/
[14] https://eandt.theiet.org/content/articles/2017/11/classic-project-nokia-6110-mobile-phone/
[15] https://www.theregister.co.uk/2016/09/06/softbank_arm_buy_is_done/
[16] https://www.theregister.co.uk/2018/06/06/softbank_offloads_51_per_cent_of_arm_china_for_a_bargain_7752m/
[17] https://www.theregister.co.uk/2017/10/19/steve_furber_arm_brain_interview/
[18] https://www.theregister.co.uk/2020/03/09/risc_v_intel_amd_arm/
[19] https://riscv.org/specifications/
[20] https://go.theregister.co.uk/tl/1936/-8553/legacy-modernization-finding-your-way-with-low-code?td=wptl1936
Somewhere in your phone, something is humming 'Happy Birthday to me'
Presumably it's washing the hands on the end of its Arms
Zarch
I recall playing this many times on an borrowed Archimedes, very impressive graphics and game performance for its day.
I'd think the threat from RISC-V is lower than that from OpenPOWER, but then again the threat from both combined is more significant than from either one alone.
Not sure why El Reg keeps only highlighting RISC-V as the One True ARM/x86 Killer, it's not the only open ISA out there, and in point of fact it's currently the one with the least amount of usable open-friendly silicon available for purchase, and also the one where most of the existing open source software simply won't run. And with those trends not changing very quickly, if at all, it's not currently a serious threat to all but the smallest ARM designs in reality.
"Not sure why El Reg keeps only highlighting RISC-V as the One True ARM/x86 Killer"
It's not going to kill x86, let's be honest nor is it going to outright kill Arm. It poses a threat to the latter's dominance, though.
We mention RISC-V because: Arm's CEO has said, in a meeting in which this hack was present, that RISC-V is keeps Arm "on its toes." Arm has responded to RISC-V with various licensing programs that reduce the upfront cost, and also briefly tried to smear RV with a weird attack website. That, to us, signals it's a problem for Arm.
There are other open CPU architectures, sure, but look, OpenRISC for whatever reason didn't excite the industry nor did OpenPower.
RISC-V is backed by Google, Nvidia, Samsung, Western Digital, and more. They are all using it in chips where they could have used Arm. That's why we mention RISC-V. And I speak as someone who is fond of all open architectures, not just RV.
"it's not currently a serious threat to all but the smallest ARM designs in reality."
The RISC-V implementations coming out of China, at least, are Cortex-5x or Cortex-A7x-grade, if the numbers are to be believed.
C.
My Risc PC still gets powered up every day
It does have a StrongARM but it started life with an ARM610. One day it'll retire, but not just yet.
Just playing with my first ARM-based "Arduino"-alike. The 48MHz Cortex M0 it uses could be compared with the 30MHz 610 originally in the RiscPC, though it uses a different instruction set (those condition codes were such a revelation) and would probably beat it in a head-to-head, though 32k RAM doesn't even come close :-)
M.
Cheers...
I'll drink to that.
If I recall correctly the first one appeared to draw no power at all, and it turned out the power supply wasn't wired to the chip correctly. However it could draw enough power through inputs that were driven high to operate.