The commercialisation of the classroom

Author : Robert Owen, Imagination Technologies

03 November 2016

OK, I admit it. I’m at least partly to blame for what’s going on now…

(Click here to view article in digital issue)

In 1994 TI asked me to drive the teaching of digital signal processing (DSP) in universities, with the simple goal that future graduates would feel comfortable using the technology and have an instinctive preference to use TI’s tools. To back this up, TI donated millions of dollars’ worth of hardware and gave away the software tools. This was a great success. The teachers wanted and needed DSP materials and TI were happy to give it. 

However, as time passed I started to see the danger signs as other companies copied TI’s approach: universities were asked to put up posters or brass plates on the door crediting the sponsor, they had to put logos and “product roadmaps” on slides, there was interference in the curriculum, restrictions that excluded competing suppliers, and often very restrictive contracts. Worst of all, many inappropriate things began to be pushed on academia: “since it’s important to the company, you must push it in universities”. 

To work well, the job of a university programme manager requires respect for academic independence. You are looking for that sweet spot of convergence where the needs of the teacher coincide with a company’s offering. 

When I arrived at Imagination, I vowed to build a programme that would address fundamental educational needs first, and Imagination’s needs second.

One of the fundamental elements of a Computer Science or Computer Engineering degree, and indeed many Electronics degrees, is the understanding of processor architecture. To achieve better performance, lower power consumption, better silicon area efficiency, and shorter time to market, a good understanding of computer architecture is vital. However, there is a huge disconnect between what is used in industry and that which teachers can use in the classroom. 

Most teachers run a mile from the complexity of CISC in its x86 form. SPARC and PowerPC are interesting, but no longer contemporary and also pretty complex. RISCV is interesting, but lacks commercial traction. ARM architectures are a jealously guarded secret, and woe betide anyone who makes an “ARM-like” processor. My survey revealed a love of the simplicity of MIPS and its true RISC design, and I found a thriving activity in “MIPS-like” and “MIPS-compatible” processors. Yes dear reader, you probably made a “simple MIPS” yourself at some time in your career. It is probably the world’s most copied CPU architecture. We understand the allure. How to make a MIPS has never been a secret: it came from Stanford University and has wonderful roots in academia. Given this, I resolved that MIPS should be Imagination’s primary focus for educational support. 

We call the project “MIPSfpga”. Imagination offers a real-world, mass production, contemporary CPU core for academic use in unobfuscated form and free to use. The MIPS microAptiv core, used in Microchip Technology’s PIC32MZ family and other major products, is available globally for the teaching of Computer Architecture, Verification and System-on-Chip design. The package includes excellent teaching materials in five languages written by Sarah Harris and David Harris. The courses are backed by the benchmark textbooks: “Computer Organisation and Design” by Patterson and Hennessy, and “Digital Design & Computer Architecture” by Harris and Harris. Since its launch in 2015, take-up of MIPSfpga has been strong, with hundreds of universities adopting it worldwide. 

We haven’t stopped there though. MIPSfpga v2.0 is imminent, and will take students deeper into the core. We have also added a route to silicon. Through Europractice and MOSIS we have enabled a low-cost route to silicon implementations of a MIPS CPU for research projects in up to 100 units of silicon. Finally, we’ve created a complementary MCU & IoT course for under-grads based on the PIC32 MCU. 

So MIPS is not just an elegant architecture based on the pages written by Patterson and Hennessy; it’s a complete teaching tool for CPU architecture and SoC design, and a route to prototyping systems. And no, to use it you don’t have to put posters on the walls, a brass plate on the door, or stop using other solutions. Our respect for your independence is sacrosanct!


Contact Details and Archive...

Print this page | E-mail this page