32-bit designs


From 1986 to 2011


The 32-bit designs we made, evaluated or “kicked the tyres” are crammed into one page here. For the more curious folk, we do give a bit more, however, this is a burial ground for past lives and artifacts. It is a place where we come a speak to our ancestors about prior art, as the patent lawyers are unable to. It is hoped that more and more websites will become spiritual homes for prior art. When Google pays $12,5 billion for Motorola Mobile just for their patent treasure chest of 24,500 patents (and they are meant to be the good guys), or Apple, Microsoft and others paying $4,5 billion for Nortel’s 6000 patents, then you had better have very understanding ancestors. Soon you won’t be able to do the equivalent of “crossing the street in Electronics Wonderland” without standing on a land mine.


From here...

And this was not the highest tech in 1993, but we could knock it out in little over a month.

S68070 terminal board

A Philips 68070 processor on a VME pinout bus. The partially populated boards were used to drive HMIs at a tannery in 1993. Click here for more info.

 

...to here

It's just a natural progression. You don’t even have to extrapolate into the future; what we have and looking back a mere ten years is good enough for automation.

Kinetis collage of boards

The Kinetis K60 ARM device in less than twenty years later has 128kBytes of RAM, 512 kBytes of Flash, up to 180 MHz clock speed, many more peripherals and costs a few dollars. It can run the same software!

ARM — 2003 ⇔ now

We ordered a compiler for the Archimedes workstation around 1987. When we were in the UK, the VLSI ARM devices were looked into. A few databooks were ordered, but the first RISCs ran marginally faster than the 68000, and although they were launched with much hype, the CISC vendor countered by saying that the competition had only done half the job. Indeed they were correct — peripheral chips still came from the CISC vendors, and interrupt controllers were pretty awful with the user having to clean up the whole mess. Fast forward some thirty years and moving from 4MHz to close to 2GHz, cleaning up the interrupt mess is not an issue. In the microcontroller area, ARM are the undisputed leaders, however, Renesas still ships more of their own controllers than any one ARM vendor! Anyway, a couple of billion a year and over 100 vendors later, we decided to jump onto the ARM ship. A pity about a lack of 64-bits, but then a x86-64 motherboard costs less than a pair of shoes.

There are several ARM entries including the index/ overview:

Atmel AVR32 — 2006 ⇔ 2007

Atmel introduced their AVR32 as a 32-bit RISC with a low-cost evaluation board. We bought two as the vendor halved the price after we had ordered — one was sold so that there would be another developer working on similar hardware to share libraries and code.

The AVR32 was a very well designed device that could trace into its own address space. Unfortunately, its roadmap does not compare to ARM.

Freescale ColdFire — 2002 ⇔ 2004

The ColdFire is fairly similar to the M68000

Freescale M68000 — 1986 ⇔ 1994

We developed several 68008/68000 boards. This was a really nice processor that took a back seat to the PowerPC and even the ColdFire roadmaps. Perhaps there were too many second sources or Motorola could see the relative sales volumes.

MIPS — 1992 ⇔ 2005

After the M68000 projects, we tested the waters with the Transputer, but settled in the MIPS camp for a few years. They seemed to move to the high-end where we were not really able to go, as we did not have customers or funding for the test additional equipment.

National Semiconductor NS32000 — 1983 ⇔ 1984

We bought a set of NS32000 chips but never completed the hardware design. For more info click here.

PowerPC — 1991 ⇔ 2007

The PowerPC was a design that was jointly defined by Apple, IBM and Motorola (AIM). In the embedded space, Motorola (later became Freescale) seemed to offer more support. We worked on several PowerPC systems:

SPARC — 1990 ⇔ 2005

We had two SPARC stations (SLC and ELC), and were seriously considering both VME and CompactPCI based systems. We also looked at chipsets from Fujitsu, Cypress Semi and Sun Microsystems, but the spat with the clones was not good for small companies, and we got rid of our SPARC hardware. There are several papers on tracing and architectural studies that are referenced in Trace and Profile Survey (PDF) 684 kBytes. We would like to have included the following as web pages, (T1, T2, Simply RISC S1, LEON), but the links would be broken for a while and with the Sun/Oracle acquisition, the T1 and T2 are no longer available. Anyway, the embedded market is no longer of interest to Sun or Oracle.

Transputer — 1992 ⇔ 1993

The Transputer was possibly ahead of its time, or just badly marketed. The momentum was lost with the late delivery of the T9000. It was pretty easy to design with if deeply embedded was not an issue. We ended up with a B004 compatible interface and some pretty impressive data rates through ADCs.

Others — 2001 ⇔ 2008

Others include the Eltrax (not documented here), DSPs and FPGAs. Have a look along the main menu bar.