Engineering Students being taught DSP on 20 year old processors RMIT

[Vernichtung]

Hi guys,
There isn't a dedicated processor thread so I figured I will put this in with Micros and FPGA's.

I am studying Electronic and Communications engineering at RMIT University. For the most part the experience has been reasonable albeit expensive.
Something which has kind of irked myself and my fellow students is our perceived absurdity of our Digital signal processing micro apps class.
The instructor is having us program a Motorolla DSP56800 processor using assembly language. 
This processor is older than most of my fellow classmates and as far as I can tell it isn't used in industry anymore. Learning to use a device that isn't used in industry
to me is absurd.
From an employers prospective i'm thinking why would I hire me who knows how to program on old hardware which isn't used over someone who can program current hardware....?
Moreover the device needs specific programming hardware not obtainable by students meaning we are limited to class specific lab time to have hands on with the tech.
Something like the Discovery Dave reviewed last week would be ideal as it's low cost and "all inclusive" layout means the students can purchase and learn via tinkering on the actual hardware itself rather than relying exclusively on theory and testing in class.

I am the Student Staff Consultative comity representative and have been asked by my Cohorts to raise this with the university.
We are wanting the class to teach ARM processors over this ancient tech.

Therefore I have a few questions i'm hoping someone on here can help with.
I need to demonstrate to the program manager / Dean that,

1) The hardware isn't used in industry / isn't in demand enough to warrant it's use in education.
2) The skills gained from learning on said hardware are not easily applied to learning current hardware.
3) The cost benefit of changing hardware is positive.
4) Changing the hardware will be of significant benefit to the students and the reputation of the university.
5) The fundamentals can be equally taught using modern ARM processors.   

Can anyone offer any advise on how I can demonstrate the above 5 criteria?
If you cant could you just answer the following few questions, it may be helpful to show the answers to the Dean and Program manager.

A) Do you think it is reasonable to be teaching DSP processors using this old technology or would you personally prefer to be taught on more modern hardware?

B) Would you hire a graduate engineer who was taught to use 20 year old processors over someone who was taught to to program current technology?

C) How long do you think it would take the average person to re-learn to use modern technology after learning this older tech?

D) If you were hiring staff, would you rather them know how to use C++ or assembly to a high proficiency?

E) If you are designing an average project which uses a processor, which processor would you choose? (Keeping costs to the average)


Thank you all very much and we look forward to working with you in the near future. =D

[TK]

I studied programming (IBM 370 assembly, FORTRAN, COBOL) using punched cards (years 1985 to 1990).  Ended having a successful career working for several multinational computer hardware and software companies.  The tools you use to learn is not important.

[Vernichtung]

I studied programming (IBM 370 assembly, FORTRAN, COBOL) using punched cards (years 1985 to 1990).  Ended having a successful career working for several multinational computer hardware and software companies.  The tools you use to learn is not important.

Thanking you kindly for the reply.
We don't doubt that we can transfer the skills to new hardware and as such that we will have successful careers in the industry.
We are caring about ensuring we are prepared for the industry, About our employability and ensuring the skills we are learning are relevant and position us above or on equal ground
to the thousands of other graduates applying for the same limited positions here in Australia.

[amyk]

Microcontroller/CPU classes still commonly use a 8080/8085/Z80. The goal is to teach you the basic principles which you can then apply in the real world.

DSP56k is still in use.

[NorthGuy]

A) Do you think it is reasonable to be teaching DSP processors using this old technology or would you personally prefer to be taught on more modern hardware?

I hope they teach principles, not processors. If so, it doesn't matter what they use. If not, you have bigger problems than worrying about processors.

B) Would you hire a graduate engineer who was taught to use 20 year old processors over someone who was taught to to program current technology?

Irrelevant. Modern HR machine hires by matching resumes. Resume doesn't say what processors were used by your teachers in your classes. And I don't think anyone would ask you that on the interview.

C) How long do you think it would take the average person to re-learn to use modern technology after learning this older tech?

This question is too vague to answer.

D) If you were hiring staff, would you rather them know how to use C++ or assembly to a high proficiency?

Programming is, first and foremost, a problem solving skill. I would try to figure out if the candidate is capable of solving problems. It would be of no concern to me what languages he thinks he can program.

E) If you are designing an average project which uses a processor, which processor would you choose? (Keeping costs to the average)

The one which meets the requirements.

[rstofer]

Microcontroller/CPU classes still commonly use a 8080/8085/Z80. The goal is to teach you the basic principles which you can then apply in the real world.

DSP56k is still in use.

I don't know anything about the device but it's in stock at DigiKey.
The 8085 is an excellent device for learning and from what I read, the 8051 is still being used.

[KE5FX]

I studied programming (IBM 370 assembly, FORTRAN, COBOL) using punched cards (years 1985 to 1990).  Ended having a successful career working for several multinational computer hardware and software companies.  The tools you use to learn is not important.

   If you were a hiring manager, you'd be singing a very different tune.  Those graduates will be at a serious disadvantage relative to their peers who will be familiar with more modern DSP platforms.

[X]

I propose that they are actually teaching this correctly, and there is a method to the madness.

Take the original Windows NT development cycle. NT was developed on non-x86 hardware on purpose so that it would reduce the chance of it becoming exclusively tied to x86 by depending on tempting x86-only features. As a result, it became on MIPS, Alpha, Itanium, and ARM.

I would just hold out and do as you're told without getting cocky with your uni about this. It may not seem relevant now, but these old computers can teach you not to be a bloat-infested high-level abstraction-addicted developer if you are willing to learn how to properly optimise your code without getting trapped in the "oh who cares about resource management, there are plenty on modern CPUs" mentality.

If more modern developers were taught this way, I reckon the software world would be a better place and I may just be inclined to work in that field.

Also there are only so many processors a uni can teach their students while allowing them to complete a degree in a reasonable amount of time. You don't get a Bachelor's Degree in ATmega with a Xilinx devices major or a Master's Degree in Obscure Moribund Hipster Architectures as it would be rather useless and limiting.

Another explanation is that these processors were used for 20 years and clearly they helped to teach the students successfully. If the uni sees this success rate continuing then it isn't necessary for them to spend extra on new equipment. Ultimately a uni wants productive students who will be headhunted, as it gives them the ability to market themselves better. Remember, "if it ain't broke, don't fix it."

[Ian.M]

Arguably you need a teaching CPU, MCU or DSP chip that's simple enough for the top 10% of the class to fully understand in detail within the course duration, and for the next 80% to fully understand in overview.   If you pick a chip that's too complex, all you are doing is training 90% of the class to be cargo cult code monkeys stringing together manufacturer supplied library routines with little to no understanding of how and why they work.

IMHO the Microchip range of MCUs is probably a good place to start looking for educational processors, because of the wide range of cores - all the way from their 12 bit instruction word baseline core up to the PIC32 range with high performance MIPS M4K cores.  Some of the low end PICs really suck from a programmer's point of view, but learning to cope with a severely resource constrained device with an idiosyncratic architecture helps you become a more well-rounded programmer.

Many of their chips are available in easily breadboardable DIP packages that will run with minimal external components, and their low cost programmer/debugger PICkit 3 covers the whole range, and clones of it are available for less than a burger meal for two so every student can afford to work at home.  Free versions of all their compilers are available with the only restriction being limited optimisation.   The 8 bit families use the proprietary XC8 C compiler or MPASM assembler,  but the XC16 and XC32 compilers for the higher end devices are GCC based, and include the GNU assembler.

Look at the dsPIC33F range of DSP capable MCUs or possibly the dsPIC33EV range if you need 5V supply and logic level chips (so a simple USB PSU can be used to power the breadboard).

[KE5FX]

I propose that they are actually teaching this correctly, and there is a method to the madness.  Take the original Windows NT development cycle. NT was developed on non-x86 hardware on purpose so that it would reduce the chance of it becoming exclusively tied to x86 by depending on tempting x86-only features. As a result, it became on MIPS, Alpha, Itanium, and ARM.

They began development on the i860, which was a bad decision that harmed the project schedule and accomplished nothing.  And Windows NT was written in C/C++, not i860 assembly.

Also there are only so many processors a uni can teach their students while allowing them to complete a degree in a reasonable amount of time. You don't get a Bachelor's Degree in ATmega with a Xilinx devices major as it would be rather useless.

You have to consider how far behind the curve the 56000 architecture is, though.  There is nothing portable about 56K assembly code, which means there are no analogies to be drawn with Windows NT.

Frankly, a modern DSP course shouldn't be using a dedicated processor at all.  It should be taught with native code, interfaced to the real world using something like one of the Digilent ADC/DAC boards.  The curriculum should also emphasize FPGA work.

[LabSpokane]

Worry about whether or not you are properly being taught the fundamentals, not the features of a transitory product.

The fundamentals are why you are attend university. 

[thermistor-guy]

I studied programming (IBM 370 assembly, FORTRAN, COBOL) using punched cards (years 1985 to 1990).  Ended having a successful career working for several multinational computer hardware and software companies.  The tools you use to learn is not important.

   If you were a hiring manager, you'd be singing a very different tune.  Those graduates will be at a serious disadvantage relative to their peers who will be familiar with more modern DSP platforms.

I have to agree: in a short-list (job-hire) contest between graduates who have experience with up-to-date tech, and those who don't, the ones with it will tend to win. This applies to experienced engineers as well. Keep current, or watch as your career prospects and market value fall away.

In a perfect world, hiring managers would recognize talented candidates who, for through no fault of their own, don't have the "right" resume. Breaking news: we don't live in that world. Film at eleven.

[Phoenix]

Which class is this specifically? Is this class your first introduction to CPU architectures and assembly, or a later year class specifically about signal processing in the digital domain?

(I'm avoiding the term DSP as it is used to describe both the practice of digital signal processing and a microcontroller optimised for the maths and operations involved in digital signal processing).

[rheb1026]

When I graduated in 2014 my school was teaching their undergrad DSP class with TI TMS320C6000. This will at least give you a data point to compare with

[KE5FX]

Worry about whether or not you are properly being taught the fundamentals, not the features of a transitory product.

The fundamentals are why you are attend university.

   Here on this planet, classroom time is a limited resource.  56000 programming is not a good use of it.

Those students are going to spend a lot of time learning tools and, yes, techniques, that they will never see again, unless they retire after a successful career and volunteer at a computer museum.

This is nothing but sheer laziness on the part of the faculty.  I don't blame the students for being annoyed about it.

[LabSpokane]

Worry about whether or not you are properly being taught the fundamentals, not the features of a transitory product.

The fundamentals are why you are attend university.

   Here on this planet, classroom time is a limited resource.  56000 programming is not a good use of it.

Those students are going to spend a lot of time learning tools and, yes, techniques, that they will never see again, unless they retire after a successful career and volunteer at a computer museum.

This is nothing but sheer laziness on the part of the faculty.  I don't blame the students for being annoyed about it.

Yes.  Classroom time IS a limited resource. This is why one focuses on the fundamentals. 

By this logic, college students would ignore opamps, discrete logic, etc. as so much is integrated into a micro now.   The likelihood of my using 7400 series logic in a design is low, yet there I was with everyone else, assembling state machines out of them. 

The OP is learning to program in assembly language. That's a fundamental skill which largely does not depend on the specific processor and frankly, the simpler the toolchain, the better.  The processor is merely a means to an end. 

And before you go accusing the faculty of laziness, perhaps you should actually assemble a semester's worth of technical curriculum and teach it first.  It's harder than it sounds.  It is far better to recycle and refine a good curriculum and focus on student learning than to constantly be chasing the bleeding edge whatsit. 

[EEVblog]

I am the Student Staff Consultative comity representative and have been asked by my Cohorts to raise this with the university.
We are wanting the class to teach ARM processors over this ancient tech.

No, you don't want to learn ARM for a DSP class. ARM is not a DSP.

Therefore I have a few questions i'm hoping someone on here can help with.
I need to demonstrate to the program manager / Dean that,

1) The hardware isn't used in industry / isn't in demand enough to warrant it's use in education.
2) The skills gained from learning on said hardware are not easily applied to learning current hardware.
3) The cost benefit of changing hardware is positive.
4) Changing the hardware will be of significant benefit to the students and the reputation of the university.
5) The fundamentals can be equally taught using modern ARM processors.   

As for 5), no, non-DSP chips defeat the purpose of teaching hardware DSP.
Something like the TI 6000 might fit the bill though:
https://www.ti.com/processors/dsp/c6000-dsp-arm/overview.html

Can anyone offer any advise on how I can demonstrate the above 5 criteria?
If you cant could you just answer the following few questions, it may be helpful to show the answers to the Dean and Program manager.
A) Do you think it is reasonable to be teaching DSP processors using this old technology or would you personally prefer to be taught on more modern hardware?
B) Would you hire a graduate engineer who was taught to use 20 year old processors over someone who was taught to to program current technology?
C) How long do you think it would take the average person to re-learn to use modern technology after learning this older tech?
D) If you were hiring staff, would you rather them know how to use C++ or assembly to a high proficiency?
E) If you are designing an average project which uses a processor, which processor would you choose? (Keeping costs to the average)

You need to go to them with a solution, not just a complaint.
Who will write the new course material for the new hardware? This alone could dictate everything. If some DSP manufacturer has DSP kit that targets courses with pre-prepared class material and project example, then there is your solution.

[EEVblog]

And before you go accusing the faculty of laziness, perhaps you should actually assemble a semester's worth of technical curriculum and teach it first.  It's harder than it sounds.  It is far better to recycle and refine a good curriculum and focus on student learning than to constantly be chasing the bleeding edge whatsit.

There should be a compulsory class called "dabbling with new shit" where you just throw everyone in a room with a bunch of the latest dev boards and a weeks worth of pizza.

[IanB]

A) Do you think it is reasonable to be teaching DSP processors using this old technology or would you personally prefer to be taught on more modern hardware?

I think learning should be done on hardware that is simple enough that its operating principles can be understood.

B) Would you hire a graduate engineer who was taught to use 20 year old processors over someone who was taught to to program current technology?

I do not select candidates on their technical skills, I select on their thinking and problem solving skills and on the depth of their insight and understanding (as best as can be learned from interview). In particular, I am less concerned with what someone has been taught, and more concerned with what they have learned.

C) How long do you think it would take the average person to re-learn to use modern technology after learning this older tech?

This is not my field, but I do not expect professional engineers to have a hard time learning how to use new tools.

D) If you were hiring staff, would you rather them know how to use C++ or assembly to a high proficiency?

Someone who knows how to use assembly might stand out. How many people would have had the interest and curiosity to do that?

E) If you are designing an average project which uses a processor, which processor would you choose? (Keeping costs to the average)

Again, not my field, but the answer to this question will depend entirely on the purpose. A commercial product in industry will have an entirely different answer to a teaching project in a university.

[EEVblog]

I see that  ARM are pushing the DSP education market:
https://developer.arm.com/academia/arm-university-program/for-educators/digital-signal-processing

[EEVblog]

D) If you were hiring staff, would you rather them know how to use C++ or assembly to a high proficiency?

Someone who knows how to use assembly might stand out. How many people would have had the interest and curiosity to do that?

Yes I would assume any student has learned C/C++ as given.
Assembly stands out.

[EEVblog]

This could be potent:
Altera FPGA+ARM DSP educational stuff:
https://www.altera.com/support/training/university/overview.html

[X]

They began development on the i860, which was a bad decision that harmed the project schedule and accomplished nothing.  And Windows NT was written in C/C++, not i860 assembly.

And yet, Windows NT (including Win2K/XP/7/10/etc) is the world's most successful operating system. My point is not about what language the OS is written in, but that being flexible with your platforms does you a world of good. They could have restricted themselves to i860, x86, Alpha and MIPS, but they didn't. Now it's even available on ARM-based Windows devices. The point here is that even though a uni may teach you an old museum architecture, the skills and algorithms can be transferred to different environments if the fundamentals are taught right.

You have to consider how far behind the curve the 56000 architecture is, though.  There is nothing portable about 56K assembly code, which means there are no analogies to be drawn with Windows NT.

The assembly code itself is not portable across the board, but the techniques used to craft that code are. Besides, it will do a student well to understand how historic architectures like the 56K worked, and how modern computers aren't actually all that much different in terms of their basic architecture (memory, CPU, I/O). When you go down to the basics, most of the technological advancements to this area have been with increasing the resources (CPU clock/bus speed, more memory and non-volatile storage, reduced I/O latency, caches, etc).

Frankly, a modern DSP course shouldn't be using a dedicated processor at all.  It should be taught with native code, interfaced to the real world using something like one of the Digilent ADC/DAC boards.  The curriculum should also emphasize FPGA work.

I think it's fine for it to be used as a supplementary teaching aid, but yes FPGAs are quite the norm these days, and it's going to be like that for the forseeable future.

There should be a compulsory class called "dabbling with new shit" where you just throw everyone in a room with a bunch of the latest dev boards and a weeks worth of pizza.

There were non-compulsory non-official variations of exactly this back when I was at uni, all thanks to security not giving a shit about students sleeping in certain buildings overnight.

[forrestc]

So let me throw some random thoughts out here which may or may not help:

An ARM is not a DSP.   Some newer ones have DSP-friendly instructions or DSP coprocessors, but a DSP is generally a different beast than a general purpose processor.  Let me quote a section from the DSP56800 datasheet:

Each on-chip execution unit, memory, and peripheral operates independently and in
parallel with the other units through a sophisticated bus system. The data ALU, AGU, and program
controller operate in parallel so that the following can be executed in a single instruction:
— An instruction pre-fetch
— A 16-bit x 16-bit multiplication
— A 36-bit addition
— Two data moves
— Two address-pointer updates using one of two types of arithmetic (linear or modulo)
— Sending and receiving full-duplex data by the serial ports
— Timers continuing to count in parallel

To paraphrase:  That chip can, in a single instruction cycle, fetch an instruction, do a multiplication, an addition, move data to and from memory, determine the next address to grab data from, AND stream that data in and out of the serial ports.   All at the same time.

By comparison, an ARM can often only do one of the above.   Plus often each of those takes multiple cycles.   I'm ignoring multicore here...  since that is an entirely different beast which just complicates processing.

I can understand the desire to move to a more modern platform.  But if this is really a DSP class in the context of learning about digital signal processors then you need to select one of the multiple DSP's out there, not a general purpose processor like ARM.     If I was designing a DSP Class from scratch, I'd probably use a dsPIC33 - they're modern, dirt cheap, and slow enough that you have to learn how to work in a constrained environment.   They're also widely used in places like motor control and power supplies where they are doing very quick analog processing at very low cost.   Interesting dev boards are also available for not a lot of money (around the same cost as a typical textbook).

In relation to the age of the processor:  don't forget that the ARM architecture is actually over 30 years old at this point. 

In relation to programming in Assembly vs C:   There is only one area which I know people still use assembly.... you guessed it, Digital signal processing.   The reason?   DSP work is often cycle-dependent.  Even there, assembly is rarely needed.   I'm skeptical that there is a valid reason for assembly here.   But note that I said C, not C++.   If you're working at the hardware level, generally you're going to use a C compiler, not a c++ compiler.   Many of the things people like about C++ are incredibly inefficient, and on a DSP you care about every cycle.     

BUT, the real advantage of Assembly vs C is that when writing in assembly, you learn quickly how much the underlying architecture matters.  In a DSP, structuring things so that the processor can actually do all of the operations in parallel is important.   You don't learn this without knowing assembly.   I don't think that the goal would be to have you proficient in assembly language, but instead to be proficient with a typical DSP architecture.

One final thought:  The question I have is what is the end goal of this course?  If it's to familiarize you with DSP hardware architecture then hardware matters, and the teacher will want to pick a good, illustrative processor which is easy to understand.   This might not be the fastest, newest processor.  If it's an algorithms class, then the hardware matters less and there shouldn't be any assembly involved.

[KE5FX]

I think it's fine for it to be used as a supplementary teaching aid, but yes FPGAs are quite the norm these days, and it's going to be like that for the forseeable future.

I think that also addresses the points made about assembly language.  Someone who is reasonably proficient in VHDL or Verilog has nothing to fear from any assembly language on Earth.  The reverse isn't true (speaking from experience).