As a teenager I took an intro CS class at a community college, and the instructor had us design a simple cpu instruction set, and write our own VM and assembler that worked and let me write and run assembly programs. It was shockingly easy, and it was amazing how much is demystified computers for me.
I feel like one could learn every level of computing this way- from designing a real cpu for a FPGA, to writing a simple OS and programs that run on it. This stuff is all shockingly simple if you just want it to work and don’t need all of the extra performance and security modern computing needs.
spit2wind 1 days ago [-]
That sounds like a fun class! It sounds very similar to https://www.nand2tetris.org/ or Charles Petzold's book "Code".
I think once you move from early fantasy CPUs to early CPUs in production such as 80286, the complexity immediately moves up drastically. IIRC it involves at least memory segmentation, protected mode (MMU).
sunday_serif 1 days ago [-]
True enough, but having designed a “fantasy cpu” gives you a better frame of reference for understanding the more complex features of a cpu (privilege levels, memory segmentation, virtual addresses, cache hierarchy, etc.)
I often feel like those who haven’t done the exercise of understanding the ISA of a “fantasy cpu” have a really hard time understanding those more advanced features.
I guess all I am saying is that learning the “fantasy cpu” still has value even if everything else in the real world is more complex.
Walking before running and all that.
PaulHoule 1 days ago [-]
I've been doing some planning for a 24-bit fantasy CPU, my plan is to make it pretty baroque. For instance it has some instructions to do things like unpack UTF-8 strings into chars, do alpha compositing, etc. The CPU part looks like a strange mainframe that didn't quite get built into the 1970s and it is coupled to a video system that would make a Neo-Geo blush.
notjoemama 16 hours ago [-]
“Neo Geo[a] is a brand of video game hardware developed by SNK.
It was launched with the Neo Geo, an arcade system (called MVS) with a home console counterpart (AES). Games on the Neo Geo (MVS and AES) were well received and it spawned several long-running and critically acclaimed series, mostly 2D fighters. Later, SNK released the Neo Geo CD, a more cost-effective console with games released on compact discs, which was met with limited success. A new arcade system, Hyper Neo Geo 64, was released in 1997, but it did not fare well. SNK also released a handheld console under the brand, the Neo Geo Pocket, which was quickly succeeded by the Neo Geo Pocket Color, which have been given praise despite its short lifetime.
SNK encountered various legal and financial issues resulting in a sale of the company in 2001. Despite that, the original Neo Geo arcade and console continued receiving new games under new ownership until 2004. The Neo Geo brand was revived in 2012 with the release of the Neo Geo X[1] handheld. Since then, a number of other Neo Geo products have been released based on the original Neo Geo.”
— Wikipedia
Seems to be from about 30 years ago in the 1990s.
jazzypants 9 hours ago [-]
I just realized that people like me that know all about the Neo Geo are actually in the minority in the world. It's easy to forget what is and isn't common knowledge.
jeffrallen 1 days ago [-]
Forget advanced features... Without understanding a CPU it's easy to never really understand pointers, and without pointers, it's hard to understand lots of data structures.
I was easily 12 months ahead of other students in my CS education because I learned 6502 assembly in high school. I wish all CS courses started with "make a VM".
mtzet 13 hours ago [-]
I don’t get this take. Is it so hard to understand that a computer operates on a giant array of bytes?
I think the hard thing to understand is that C’s pointer syntax is backwards (usage follows declaration is weird).
I also think understanding how arrays silently decay to pointers and how pointer arithmetic works in C is hard: ptr+1 is not address+1, but address+sizeof(*ptr)!
Pointers are not hard. C is just confusing, but happens to be the lingua franca for “high level” assembly.
wkjagt 22 hours ago [-]
This is very true. I had tried learning C multiple times but pointers were always kind of hard. I kind of understood them but not really. I later spent a lot of time making an OS in assembly for my homebrew 6502 computer, and after that pointers made so much sense. It actually took me a little while to realize that these addresses I was passing around were pointers, and I had started to understand them without realizing it.
markus_zhang 1 days ago [-]
agreed! A fantasy CPU is good for the first project.
The mostly redundant memory segmentation features that modern OS do not care or use anymore.
If OS fully utilizes memory segmentation it can be the most secure OS and makes borrow checker languages like Rust becomes unnecessary since not unlike Rust it can mitigates data races and race conditions (better at the latter).
This can considerably reduce developers' cognitive load or overload [1]. It also can contribute towards more secure computing and containers can be made as secure as VMs.
Someone can probably propose a new memory segmentation extension to the Linux kernel in the similar manner of real-time Linux. But again it can takes 20 years to be accepted into the mainline Linux kernel [2].
Or you could pick the 68000 which is easier to emulate than a 286 because it was 'properly designed' (and not a stopgap solution like the x86 line until the 386 arrived).
remexre 1 days ago [-]
the 80286 has its own problems/inessential complexity
if you look at this from the riscv angle, moving from "u-mode only vm that doesn't use paging under the hood" to "u+s-mode vm with sv39" isn't an enormous jump in complexity imo
i think i might teach it starting as like, "sv21" (page tables aren't nested), then pose real sv39 and the tree structure as the solution to making a sparse mapping over 512GiB
then moving on to the idea of having a TLB is simple, especially if students have already been introduced to hashtables
PaulHoule 1 days ago [-]
Sometimes I dream of a 24-bit generation of computers (in terms of address space and the space of index math) of which
may be the best realization. You can almost run a real operating system on that chip in the sense of a 24 bit supervisor that could run multiple 16 bit processes (if you could get CP/M to run you have one heck of a userspace) Unfortunately you can't trap the instruction that switches back to 24 bit mode.
Would be nice too if a 24 bit supervisor could contain a 24 bit application, that requires some kind of memory management and I'd imagine something a little lighter weight than the usual paging system, say something that maps (0..N) in logical space to (i+0..i+N) in physical address space. I like the "access register" concept
but would probably have a small number of memory banks such as 16 or 64. In a load-store architecture it doesn't seem like much of a burden to add a bank id to stores and loads. In return you get not just more RAM but also separation of code and data, video memory, file mmap(ing) and such.
What bugs me is how to control the mapping of memory banks to physical memory, on one hand you want the supervisor in charge so it can be used for memory protection, on the other hand some programming techniques would want the speed of changing the memory banks from user space.
The 80286 was a turkey because it didn't meet the minimal viable product level of being a target for a 24-bit OS that could virtualize DOS applications. It was already crazy fast and became affordable pretty quickly but it seemed tragic that it couldn't do that.
fentonc 10 hours ago [-]
I took a different approach by just making an FPGA-based multi-core Z80 setup. One core is dedicated to running 'supervisor' CP/NET server, and all of the applications run on CP/NET clients and can run normal CP/M software. I built a 16-core version of this, and each CPU gets its own dedicated 'terminal' window, with all of the windowing handled by the display hardware (and ultimately controlled by the supervisor CPU). It's a fun 'what-if' architecture that works way better than one might expect in practice. It would have made an amazing mid-to-late 1980s machine.
Narishma 22 hours ago [-]
> The 80286 was a turkey because it didn't meet the minimal viable product level of being a target for a 24-bit OS that could virtualize DOS applications. It was already crazy fast and became affordable pretty quickly but it seemed tragic that it couldn't do that.
That's mainly because it was designed before the IBM PC was a success and backwards compatibility with DOS applications important.
monocasa 17 hours ago [-]
Even better than that was an Z280 which did have a proper user/supervisor mode and simple MMU. Really reminiscent of a PDP-11 in terms of supervisor state, just strapped to a z80 instruction set otherwise. Also added a nice SP relative load to the instruction set.
kazinator 24 hours ago [-]
Also PIC24 family of μcontrollers.
UniverseHacker 19 hours ago [-]
There are plenty of real CPUs that are useful in the real world but are much simpler than a 286.... 8 bit RISC microcontrollers like Atmel AVRs for example. I've done a small amount of programming for those in assembly, and they are not a whole lot more complex than the 'toy' design I used for the class.
whartung 1 days ago [-]
Our CS 101 class had such a system. Simple computer/assembler written in BASIC on the PDP.
One of the assignments was to do a simple multiply (i.e. add in a loop).
Rather than do that, my friend simply altered the program and created a new MUL command.
The teacher was not amused.
JoeAltmaier 1 days ago [-]
Teachers don't want extra work.
My first 360 assembler class our first assignment: add two numbers, fault, print the crashdump and circle the answer to the ADD in the printout.
I wrote an RPN calculator, had it do a series of calculations and print the result. Turned that in.
The teacher wrote on it "You have 24 hours to turn in the required assignment"
coreyp_1 10 hours ago [-]
I was a professor. I would have given you a zero.
From the instructor's perspective, it sounds like you didn't even try to do the intentional assignment (which the fault/crashdump/find info in the crashdump was the most important part), and either did your own thing or sent in someone else's assignment because you didn't look closely enough at what it was. Yes, this is a big accusation, but it happens every semester.
The professor just wanted you to learn what you were supposed to from the assignment, so that you could build off it.
Your cynical take on it now shows that you still don't understand how to teach or the importance of that fundamental skill.
Is there a place for fun, exploratory projects? Of course, and I always incorporated them into my syllabus. But there's also a place for structure. If you want to explore along the way, then that is even better! But you still have to meet the incremental checkpoints along the way, otherwise you have not demonstrated that you have met the core competencies of the subject, as reflected by your grade.
There's a lot of shortcomings in modern university education, but I loved teaching, and would still be doing it today if it paid reasonably. My biggest headaches were the cheaters and the know-it-alls, mostly because neither of them knew enough to respect the subject matter and its importance. The students that made me the most proud, though, were those that trusted the roadmap, and in the end were able to do bigger and better things than they thought that they were capable of.
It's not about extra work. We did plenty of it, btw. We just didn't always tell you about it. The University system referrs to it as "service", and you, the student, are the beneficiary of it, either directly or indirectly.
"Teachers don't want extra work." I'm sorry, but this sounds like a toddler stomping their foot and saying "Mommy and Daddy don't love me because they went to work today instead of staying home to play with me."
I'm glad that you wrote the RPN calculator. That's cool! I'm disappointed that you use this as an opportunity to bash the educator who agreed to take a low salary so that they could help you learn important things that are fundamental to your craft and chosen field of study.
JoeAltmaier 10 hours ago [-]
I didn't bash anybody - that's words in my mouth.
I understand that it's a sensitive subject, students that stray from the boundaries of the lesson plan, so often to no good effect. Making work for an overworked teacher, I have complete sympathy.
But understand that there's also room for the teacher that doesn't stomp their own foot and petulantly demand every student stay in the herd like good little scholars.
Maybe even encourage exceptional students, maybe suggest something more appropriate for them, another class perhaps. But no, that certainly didn't happen, just the deliberate ignoring of the effort spent, because it didn't serve their tiny objective.
Just that one weary, autocratic bland statement, telegraphing as sure as a digital signal "I am the teacher, and you will not do anything I don't sanction, like learning way more than I demand for today's lesson"
rramadass 2 hours ago [-]
I am not the person you replied to, but you are wrong about this;
> "I am the teacher, and you will not do anything I don't sanction, like learning way more than I demand for today's lesson"
The person explicitly made it clear that while you can be encouraged on your own initiatives you still need to be graded/ranked on a common scale with your peers and hence the need for some sort of standardization.
For example, in a recent HN discussion on "Software Developer Productivity" lots of folks were arguing on how you can have no common scale/metrics and so you can make no comparisons at all between developers which is plainly wrong.
Education is about giving everybody a minimum standard of knowledge/tools needed for the Modern World and that is what a Teacher/Professor focuses on. Every Society needs to revere/honour (and pay well !) their Teachers/Professors because they are the ones who build the next generation.
rramadass 3 hours ago [-]
Well put! A thoughtful, balanced and nuanced explanation. I wish more Teachers/Professors/Seniors put forth their pov like you. The problem with "today's kids" is that because of their access to the Internet they think they "know" everything whereas without the structure given by the educational system (in spite of its shortcomings) all is mere data and they can form no proper mental model of the "entire thing". Education is holistic and tries to give one most of the knowledge and tools needed to deal with the Modern World.
MortyWaves 23 hours ago [-]
“Don’t let education get in the way of your learning” seems appropriate here.
Never underestimate how few fucks teachers give about anything beyond their small bubble.
fuzztester 22 hours ago [-]
I never let schooling get in the way of my education.
- Mark Twain.
aleph_minus_one 11 hours ago [-]
> My first 360 assembler class our first assignment: add two numbers, fault, print the crashdump and circle the answer to the ADD in the printout.
> I wrote an RPN calculator, had it do a series of calculations and print the result. Turned that in.
The teacher wanted you to learn:
1. how to write a (very simple) assembly program for 360 to add to numbers
2. how to create a fault
3. how to print a crashdump
4. how to read a crashdump
By using a RPN calculator, you learned none of these objectives.
JoeAltmaier 10 hours ago [-]
Not 'using'. Writing. In assembler. A full RPN calculator.
Which, by the way, required accomplishing those bullet points many times over. Obvious to anybody who's ever written anything in assembler.
kubb 21 hours ago [-]
Sounds like you needed to learn to follow basic instructions.
JoeAltmaier 17 hours ago [-]
Said like a true teacher.
chii 1 days ago [-]
> This stuff is all shockingly simple
it is, but the thing is that these simple building blocks end up quite far away from an actual production level outcome that someone looking at a computer might see or interact with. It's hundreds of levels deep.
Someone with curiosity and eager to learn will be able to easily learn these foundational layers. Someone looking to "get rich quick" and be ready and employable ASAP won't.
tralarpa 1 days ago [-]
That's what the nand2tetris course does, I think (I only looked at the first lessons)
mystified5016 15 hours ago [-]
After watching Ben Eater's breadboard CPU series, I want nothing more than to design and emulate my own CPU.
I just wish I could find the time to sit down and design the damn thing.
rramadass 24 hours ago [-]
Some books that i was pointed to;
1) Virtual Machines: Versatile Platforms for Systems and Processes by Smith and Nair - Seems to be a comprehensive subject overview book.
2) Virtual Machines by Iain Craig - Seems like a more hands-on book with languages and VMs.
3) Virtual Machine Design and Implementation in C/C++ by Bill Blunden - Seems like an hands-on implementation.
If somebody who has read any of the above can add some comments that would be helpful to everybody.
kragen 20 hours ago [-]
I'm not convinced that the subject is narrow enough to overview in a single book. The considerations involved in writing a Nintendo emulator, a hypervisor using VT-x, a conventional multitasking operating system, an interpreter for a new scripting language, a SQL query optimizer, a regular expression matcher, a security monitor for untrusted player code running on a game server, etc., would seem to have very little overlap. But these are all "virtual machines". There's even a stack-based virtual machine in the terminfo format for specifying character-cell terminal escape sequences!
In a deep sense, virtual machines are what make computers computers, as we use the term today. Turing's 01936 paper about the Entscheidungsproblem hinged on virtual machines being able to emulate one another.
rramadass 18 hours ago [-]
You are thinking about "Virtual Machines" in a more philosophical/abstract sense. Here the definition is in a much more narrower and conventional sense. Surprisingly i can't find too many books on this subject. The Iain Craig book looks pretty interesting since it seems to link language to the VM, so maybe one can learn both language design and a VM to run it on.
I'm mostly talking about the sense of the term used in the title of the article we're commenting on, not a more philosophical sense, though I did draw the connection in my second paragraph. It's hard to get less philosophical than terminfo!
saithound 1 days ago [-]
Alas, educational architectures like the Brookshear Machine and Little Computer look nothing like real ones, making them worse than useless: in my experience students who take courses using these often end up with a more distorted understanding of computers than those who take no classes at all.
Most people who want to learn a bit about how their machines work would be better served by taking an operating systems course. Same goes here: if you only have time for a short tutorial, I recommend "Writing my own bootloader" instead. [1]
(This is not meant to say that the Write your own VM tutorial is a bad tutorial; only that, in my experience, most people who'd do it would be best served by a different subject)
I just did the LC-3 one and look forward to learning a bit about dynamic recompilation using LC-3 as an (inappropriate) target machine in the current project.
Can you please elaborate why it is bad to use LC-3 for studying computer architecture? I do understand it is completely different from real hardware, and too simplified. But from the perspective of writing CPU emulators, is it bad for me?
anyfoo 1 days ago [-]
When studying computer architecture, I think the implementation details of a real architecture, even a simple one, might actually slow you down if your goal is to learn about dynamic recompilation. The commenter you replied to seems to be coming from a software engineering perspective, i.e. learning how to program a computer (which someone interested in computer architecture probably knows about already).
saithound 1 days ago [-]
I answered a similar question by anyfoo downthread, the point about implementing a PIT should explain what I mean, and why I think that emulating a simple real architecture might be more useful while requiring similar effort.
Again, you are coming at this from a software engineering direction, not a computer architecture/engineering one.
saithound 1 days ago [-]
Since I have the exact same thought (that you are coming at this from a software engineering direction, not a computer architecture/engineering one), this is unlikely to be further productive :(
You say you want students to be able to write emulators/simulators etc. (you list a bunch of software.engineering tasks). I say writing an LC-3 emulator won't give you any transferable skills for writing, say, a Gameboy Emulator (still a 30yo system), because these educational arhitectures are bad and abstract away precisely the things that you have to get right to do that, leaving you with something trivial to implement: a C array for "memory" and a few bitwise operations.
Since we're largely looking at the same things and coming to orthogonal conclusions, I suggest we agree to disagree on this one.
liontwist 6 hours ago [-]
This article would directly apply to NES or game boy emulation.
Yes, there are fewer instructions.
PaulHoule 1 days ago [-]
I think of Knuth's old MIX which was a decimal machine which might have gotten built in the 1960s but which nobody has built since the 1970s. A system like that can teach you many fundamentals but not the tricks here
which mainly depend on conventional numeric representations.
kragen 19 hours ago [-]
MIX wasn't decimal, but rather abstract over the choice of base, a decision Knuth changed in later volumes of the series, even before switching to MMIX.
There have been decimal and hybrid computers built since the 01970s (especially for pocket calculators, but new decimal instructions were still being added to Intel's lineup in the 8086 with AAM and AAD, and I think decimal continued to be a consideration for the AS/400 into the 90s), but MIX is really kind of an 01950s design, with:
- word-addressed memory
- five bytes per word (plus a sign bit)
- one word per instruction
- six bits per byte (in binary realizations)
- possible decimal
- a single accumulator
- sign-magnitude
- no stack, and
- a nonrecursive subroutine call convention relying on self-modifying code.
You can find one or another of these characteristics in machines designed since 01960, but the combination looked old-fashioned already when it was introduced.
anyfoo 1 days ago [-]
Can you elaborate what you don’t like about this one, LC-3, in particular? I’m not familiar with it, but just had a look at it on Wikipedia. After your comic, I was expecting something weird, but upon a quick glance, it doesn’t seem too jarring. A bit like a mixture of s/360, some x86, and a tiny bit of ARM (or other RISCy architectures). With lots of omissions and some weirdness of course, but the goal seems to be to quickly come to a working implementation. I’m curious what exactly you think makes it “worse than useless” for teaching.
saithound 1 days ago [-]
I'm not talking about the instruction set, or teaching basic assembly (probably anything except Malbolge is suitable for that).
Let's look at just one thing every programmer has to deal with, memory.
On an LC-3, the address space is exactly 64KiB. There is no concept of missing memory, all addresses are assumed to exist, no memory detection is needed or possible, and memory mapped IO uses fixed addresses.
There are no memory management capabilities on the LC-3, no MMU, no paging, no segmentation. In turn there are no memory-related exceptions, page faults or protection faults.
When an x86 machine boots with 1MB of RAM, the 4GB address space still exists in full, but accessing certain addresses will cause bus timeouts, crashes. One must track and manage available memory. There's a BIOS, and manually probing memory locations may trash its critical structures. There's INT 0x15.
I picked memory arbitrarily but you run into the same limitations no matter what you pick. Would a students who was educated on LC-3 know how a computer keeps time? Of course not, there's no PIT, there's no CMOS clock. Would they have thought about caches? Nope.
Oh, but wouldn't a student who implements a timer emulation extension for LC-3 learn more about timers than somebody who just learned to use an x86 PIT? Alas, no. There are 20 equally easy and reasonable mathematical ways to implement a timer abstraction. A good 15 of these are physically impossible on real hardware, out of the remaining 5 two would be prohibiitively expensive due to electrical engineering reasons, one has never been implemented in real hardware due to historical accidents, and two are designs that are actually in use. So to write timer emulation that teaches you anything at all about how actual timers work, you'll have to look at and understand a real architecture anyway.
That's why educational architectures are so contraproductive. They abstract away exactly the things that make modern computers modern computers. One comes away with fundamentally wrong ideas about what computers do and how they actually work, or could work.
It's like learning to drive in GTA: in principle, there could be plenty of skills that transfer to the real thing, but in practice you'll prefer to teach how to drive to the person who didn't play GTA at all.
anyfoo 1 days ago [-]
There seems to be a misunderstanding what fields those learning architectures are geared towards.
They are usually not for learning about application or even systems programming, in assembly and otherwise. They are about CPU architectures (and some surrounding concepts) themselves. The goal is to be able to quickly build your own emulator/simulator (like the VM behind the link), and maybe an assembler. Or, coming from the other side, given a working emulator/simulator, implement a high level language compiler, such as for a simplified C language.
In both those goals the CPU architecture is geared towards quickly converging towards a working system. So of course they don't require you to implement (as someone learning architecture engineering) or even use (as someone learning to build a compiler) an entire MMU, much like your first project car likely wouldn't contain a computerized double clutch transmission. Instead, they are simplified CPU architectures that allow you to make steady progress, without being slowed down by the tedious complexity that a real architecture brings for reasons that are not relevant yet when starting out.
Then, once that is achieved, you can freely add things like an MMU, a cache controller, maybe shift towards a pipelined or even superscalar architectures...
Besides, a very large class of computers don't have many of the things that you mentioned. For one, even very advanced microcontrollers explicitly don't even have an MMU, because that would destroy latency guarantees (very bad for automative controllers). For the rest, I've got to say that there is a certain irony in complaining that computer architecture students don't know about "modern computers", while in the same breath mentioning things like INT 15h, segmentation, and x86 PITs, as if we were in the 1990s.
1000100_1000101 14 hours ago [-]
> On an LC-3, the address space is exactly 64KiB. There is no concept of missing memory, all addresses are assumed to exist, no memory detection is needed or possible, and memory mapped IO uses fixed addresses.
> There are no memory management capabilities on the LC-3, no MMU, no paging, no segmentation. In turn there are no memory-related exceptions, page faults or protection faults.
Sounds an awful lot like a Commodore 64, where I got my start. There's plenty to learn before needing to worry about paging, protection, virtualization, device discovery, bus faults, etc.
It sounds like it's not teaching the wrong things like your GTA driving example, but teaching a valid subset, but not the subset you'd prefer.
upghost 1 days ago [-]
Interesting. How would you advocate actually gaining that knowledge then?
It seems like a student would need to know a significant amount of coding in order to learn those abstractions in an interactive manner.
And by learn them I mean learn them (not just following a tutorial), organizing the code for a fully working x86 architecture is no joke.
But a student with that level of skill probably doesn't need to learn the x86 architecture so intensively, they are probably already employable.
I am asking this seriously, by the way, not trying to nitpick. I'm trying to put together a free course based on the video game Turing Complete[1] but from what you're saying it sounds like it might not be very effective. (to be clear the goal is to teach programming, not Computer Engineering)
My working assumption throughout was that the people in a computer architecture class already had 1 or 2 semesters of other programming courses where they worked in a high-level language, and are looking to learn how computers work "closer to the hardware". And educational architectures create a completely false impression in this domain.
If I had to teach assembly programming to people who never programmed before, I'd _definitely_ not want to start with x86 assembly. I'd start by teaching them JavaScript so that they can program the computers that they themselves, and other people, actually use. At that point they'd be ready to learn computer architechture through an x86 deep dive, but would no longer need to learn it, since, as you said, they'd probably already be employable. But the same goes for learning LC-3, and much more so.
To be honest, my opinion is only that educational architectures are a poor way to learn what modern computers actually do, and while I think I have good reasons for holding that particular opinion, I don't have the breadth of experience to generalize this specific observation into an overarching theory about teaching programming and/or compsci. I hope your course will be a useful resource for many people, but I doubt listening to me will make it better: my experience does not generalize to the domain you're targeting.
tonyarkles 15 hours ago [-]
Many years ago I taught CMPT215 - Intro to Computer Architecture. This was a second year course and taught MIPS assembler. There’s a chain of follow on courses culminating in a 400-level course where you implement a basic OS from scratch on x86 (when I took it) or ARM (present day, I believe).
MIPS was, I think, a decent compromise between academic architecture and industrial architecture. There was a decent amount of practical stuff the students had to deal with without having to eg do the dance from x86 real mode to protected mode.
One of the most memorable lectures though was on the second last day. As it turned out I had gotten through all of my material one say early. We had a review/Q&A session scheduled for the last day but needed 1h30 of filler. I ended up putting together a lecture on JVM bytecode. Throughout the course I would often use C <-> MIPS ASM throughout examples because the 214, which was a prerequisite, was a C course. All of their first year work had been in Java. Anyway, we took real Java programs and disassembled them down to bytecode and walked through it, showing how it’s very similar to hardware assembly but with extra instructions for eg placing objects on the stack or calling methods on objects. The class ended up going almost an hour over because everyone was oddly engaged and had a ton of questions.
TuringTest 22 hours ago [-]
The thing is that you are conflating CPU architectures with computer architectures; In academia they are treated as two different educational topics, for good reason.
The first one covers the lowest level of how logic gates are physically wired to produce a Turing-equivalent computing machine. For that purpose, having the simplest possible instruction set is a pedagogical must. It may also cover more advanced topics like parallel and/or segmented instruction pipelines, but they're described in the abstract, not as current state-of-the-art industry practice.
Then, for actually learning how modern computers work you have another separate one-term course for whole machine architecture. There you learn about data and control buses, memory level abstractions, caching, networking, parallel processing... taking for granted a previous understanding of how the underlying electronics can be abstracted away.
upghost 18 hours ago [-]
I appreciate the candid response. I have noticed there is a class of very intelligent, well-educated adult learners who have nevertheless been unexposed to software education until adulthood who are now looking for a career change. I've found that there is a lot of difficulty initially with combining abstractions, i.e., "a variable holds a value, a function is also a value, a function is also a thing that sometimes takes values and sometimes returns values, therefore a variable can hold a function, a function can be passed as an argument to a function, and a function can return a function".
Reasonable adults might have reasonable questions about those facts, such as, "what does any of that have to do with a computer?"
To my embarrassment, I realized they were completely right and my early exposure to software made me overlook some extremely important context.
So for these adults, the expectation of struggling through a few semesters/years of javascript is not an optimal learning route.
My hope was that working from the logic gate level up would at least provide the intuition about the relationship between computers (Turing Machines, really, not modern computers) and software.
However, I think based on your excellent critique I will be sure to include a unit on how "educational architectures are very different from modern architectures and I may have ruined your brain by teaching you this" haha.
1 days ago [-]
hayley-patton 1 days ago [-]
The LC-3 has pretty odd addressing modes - in particular, you can do a doubly indirect load through a PC-relative word in the middle. But you still have to generate subtraction from negation, and negation from NOT and ADD ,,#-1. (I suppose NOT d,s = XOR d,s,#-1 would be a better use of the limited instruction encoding space too.)
userbinator 1 days ago [-]
My recommendation is to use an old 8-bit architecture like 6502 or Z80.
Apparently a lot of CS courses in India still use the 8086/8088 too.
whartung 1 days ago [-]
I wrote a simple 6502 simulator. Just made it a byte code interpreter.
No cycle accurate anything. Simple I/O. Since I/O on the 6502 is all memory accesses, adding console is as simple as echoing whenever a character is stored in a particular memory address. No reason the simulate a VIA chip or UART or anything that low level.
Fun project. You can take it as deep as you like.
markus_zhang 1 days ago [-]
Back in China my first comp arch class used 8051, it was fun. I remember everyone getting a development board and playing with assembly.
userbinator 1 days ago [-]
8051-core SoCs seem to be found in tons of cheap Chinese electronics. MP3 players, toys, SD card/USB drive controllers, touchscreen drivers etc.
kragen 1 days ago [-]
Yeah, the 8051 family is still pretty popular, unlike, for example, the 68HC11/08/12 etc.. I think it's not as popular as the PIC, ARM, AVR, or even MSP430, but it seems to be more popular than STM8, M16C, SH-4, MAXQ, MIPS, or 8086. I'm not sure how to assess the Z80 (8080 variant), Z8, RL78, and RX families, all of which seem to be showing disturbing signs of life. Fortunately the 8048 seems to be dead.
userbinator 22 hours ago [-]
I think it's not as popular as the PIC, ARM, AVR, or even MSP430
Far more 8051 cores are shipped than all of those combined, excluding ARM.
kragen 20 hours ago [-]
Even today? That's plausible, but I'd like to know what leads you to believe it. I was mostly looking at Digi-Key stock numbers, which are at best a very imperfect measure.
Thanks, I have heard a lot of good words about this book.
rramadass 6 hours ago [-]
I have all of Michael Pont's books. The reason? Everything is in C (lots of code) and not a line of assembly in sight! And his focus is on time-triggered architectures suitable for real-time in addition to "standard" event-triggered architectures. Thus one gets to learn both the control paradigms in a unified manner. His earlier books use 8051 while the later books deal with ARM. And all of the code/teaching is for "industrial strength" apps and not merely for "hobby" purposes. So here too you get a broad spectrum.
To be That Guy: this is an emulator, not a VM. While the term can apply in a descriptive sense, and in the pre-hardware-virtualization past there was some ambiguity, the overwhelmingly common usage of "Virtual Machine" in the modern world refers to an environment making use of hardware virtualization features like VT-x et. al.
kragen 1 days ago [-]
I don't think that terminology is "overwhelmingly common", and I'd argue that it isn't even entirely correct. The JVM is widely deployed, the Ethereum VM is called the "EVM", https://www.linuxfoundation.org/hubfs/LF%20Research/The_Stat... describes BPF and eBPF repeatedly as "virtual machines", https://webassembly.org/ begins by saying, "WebAssembly (abbreviated Wasm) is a binary instruction format for a stack-based virtual machine," etc. "Virtual machine" is still the most common term for fictional machines. (Myself, I think I'd prefer "fictive machine", "fictious machine", "imaginary computer", or "fantastic automaton", but I doubt these terms will gain adoption.)
You can't always use the term "emulator" instead of "virtual machine", because while you could say wasmtime was an "emulator", you can't correctly say that WebAssembly itself is an "emulator". Rather, WebAssembly is the virtual machine which wasmtime emulates. It's also common to call emulators "virtual machines". (The Wasmtime web site mostly calls wasmtime a "runtime" and WebAssembly an "instruction format", FWIW.) And of course a running instance of an emulator is also a "virtual machine" in a different sense.
I think it's also reasonable to use "virtual machine" in the way you are describing, and it has some overlap with this last sense of "virtual machine". Perhaps in your current environment that is the overwhelmingly most common usage, but that is definitely not true elsewhere.
dataflow 23 hours ago [-]
Disagree with both of you. JVM was always a misnomer, it just stuck because that's just the name they stuck with. Putting "virtual machine" in its name in defiance of the meaning doesn't suddenly make it a virtual machine any more than calling your dog "my son" makes your dog your son. And virtual machines have used software virtialization for the longest time, they aren't always hardware based.
Really, a virtual machine is literally what it says on the tin: something that isn't a physical computer (machine), but presents itself as one. You know, with a virtual CPU or a virtual disk or a virtual keyboard etc., whatever a computer would have. If it doesn't present itself as a machine, it's not a virtual machine.
Calling JVM a virtual machine just because it interprets some bytecode the same way a machine does is like calling me a virtual doctor just because I read text the same way a doctor does.
kragen 20 hours ago [-]
The JVM is a VM in that sense. It isn't a physical computer, but presents itself as one. It's true that that computer doesn't have a disk or a keyboard, but neither do the vast majority of physical computers, including, specifically, the set-top boxes the JVM was originally designed to run on.
dataflow 18 hours ago [-]
> The JVM is a VM in that sense. It isn't a physical computer, but presents itself as one.
No, it isn't and doesn't. What in the world does it have that makes you think it pretends to be a computer? A bunch of pretend registers and a memory limit?
kragen 16 hours ago [-]
It doesn't have registers, pretend or otherwise, or a memory limit. It sounds like you aren't very familiar with the JVM or computers in general. On the JVM in particular, I suggest reading https://docs.oracle.com/javase/specs/jvms/se23/html/index.ht....
dataflow 13 hours ago [-]
What are you talking about? It literally has a program counter register, it's right there in you own link?
I just mixed it up with Dalvik having multiple registers, but that makes it even worse. At least normal machines have more registers.
And literally here:
The following section provides general guidelines for tuning VM heap sizes: The heap sizes should be set to values such that the maximum amount of memory used by the VM does not exceed the amount of available physical RAM.
Maybe instead of telling me I'm clueless, tell me how you're right? What do you see in it that is actually virtializing a computer in your mind? Because I'm pretty sure that if someone sold you a VM service and it turned out to be just them running your programs under the JVM, you'd be kind of mad.
I have to politely disagree. In the most pure sense a VM is a made up computer, without any implications being expressed to what it’s intended to be used for or how it works.
In the article the author even states as one example emulation a classic console, but it’s clear from the article that there are many other possible VM’s from the definition they provide.
Anyways the point is VM is abstract, and there are many many types of them. Simulators, emulators, hypervisors, etc… are all VMs, but also there are VMs that don’t quite already have a name for them yet because they’re strange.
I don’t mean to be rude at all - quite the contrary, I want to be respectful - but I also want to be clear about these terms for people who are interested in learning.
troad 1 days ago [-]
> To be That Guy: this is an emulator, not a VM. While the term can apply in a descriptive sense, and in the pre-hardware-virtualization past there was some ambiguity, the overwhelmingly common usage of "Virtual Machine" in the modern world refers to an environment making use of hardware virtualization features like VT-x et. al.
I don't think the distinction you're advocating for actually exists. 'Virtual machine' is commonly used for any software that executes machine- or bytecode, irrespective of the reason. This can include virtualisation, but you also commonly see the term used for language runtimes: e.g. Java's JVM, Ruby's YARV (Yet Another Ruby VM).
The one area you don't actually hear the term that often is in emulation, and this is in part because most modern emulators have tended away from emulating entire systems, and towards techniques like dynamic recompilation (dynarec) of emulated software.
asimpletune 20 hours ago [-]
Exactly, emulation is “make sure the input and output of what you’re emulating are the same”, whereas simulation is “model the actual internal mechanisms of the thing you’re simulating, so as to achieve the same inputs and outputs”.
LeFantome 24 hours ago [-]
This is a VM like in JVM ( Java Virtual Machine ). I think you will find Java qualifies for “overwhelmingly common usage”.
ajross 12 hours ago [-]
Not in this context, no. If you tell your boss you're deploying the solution "in a VM" and it turns out to be a .jar file, you're likely to be fired.
Context and convention are important. This subthread is filled with people who are technically right, but choosing to cheer for an unclear and idiosyncratic terminology. A VM means hardware virtualization. Choosing other meanings is at best confusing and at worst a terrible security bug.
jazzypants 9 hours ago [-]
Is it not the exact same concept underneath? An environment that provides a layer of abstraction by translating instructions into a standardized, transportable form that can be executed consistently, regardless of the underlying hardware or operating system?
I always thought that it made sense to use the word for both things, and I never even thought twice about it when people used it for programming languages.
It would have to be a very strange conversation for someone to misunderstand someone else in the way you mentioned.
ajross 9 hours ago [-]
> Is it not the exact same concept underneath?
Sure, in precisely the same way that a kitchen knife and a sword are the "exact same concept underneath". But if you use the words interchangeably you'll confuse people.
And I'll just repeat the point, if you tell your boss your web app is "running in a VM" when you just wrote it in Java, you'll be fired for some really terrible security analysis.
1 days ago [-]
Rendered at 08:24:18 GMT+0000 (UTC) with Wasmer Edge.
I feel like one could learn every level of computing this way- from designing a real cpu for a FPGA, to writing a simple OS and programs that run on it. This stuff is all shockingly simple if you just want it to work and don’t need all of the extra performance and security modern computing needs.
[0] https://store.steampowered.com/app/1444480/Turing_Complete/
I often feel like those who haven’t done the exercise of understanding the ISA of a “fantasy cpu” have a really hard time understanding those more advanced features.
I guess all I am saying is that learning the “fantasy cpu” still has value even if everything else in the real world is more complex.
Walking before running and all that.
It was launched with the Neo Geo, an arcade system (called MVS) with a home console counterpart (AES). Games on the Neo Geo (MVS and AES) were well received and it spawned several long-running and critically acclaimed series, mostly 2D fighters. Later, SNK released the Neo Geo CD, a more cost-effective console with games released on compact discs, which was met with limited success. A new arcade system, Hyper Neo Geo 64, was released in 1997, but it did not fare well. SNK also released a handheld console under the brand, the Neo Geo Pocket, which was quickly succeeded by the Neo Geo Pocket Color, which have been given praise despite its short lifetime.
SNK encountered various legal and financial issues resulting in a sale of the company in 2001. Despite that, the original Neo Geo arcade and console continued receiving new games under new ownership until 2004. The Neo Geo brand was revived in 2012 with the release of the Neo Geo X[1] handheld. Since then, a number of other Neo Geo products have been released based on the original Neo Geo.”
— Wikipedia
Seems to be from about 30 years ago in the 1990s.
I was easily 12 months ahead of other students in my CS education because I learned 6502 assembly in high school. I wish all CS courses started with "make a VM".
I think the hard thing to understand is that C’s pointer syntax is backwards (usage follows declaration is weird).
I also think understanding how arrays silently decay to pointers and how pointer arithmetic works in C is hard: ptr+1 is not address+1, but address+sizeof(*ptr)!
Pointers are not hard. C is just confusing, but happens to be the lingua franca for “high level” assembly.
A wonderfully different early CPU with plenty of existing software is the https://en.wikipedia.org/wiki/RCA_1802 which was a target of the https://en.wikipedia.org/wiki/CHIP-8 interpreter.
If OS fully utilizes memory segmentation it can be the most secure OS and makes borrow checker languages like Rust becomes unnecessary since not unlike Rust it can mitigates data races and race conditions (better at the latter).
This can considerably reduce developers' cognitive load or overload [1]. It also can contribute towards more secure computing and containers can be made as secure as VMs.
Someone can probably propose a new memory segmentation extension to the Linux kernel in the similar manner of real-time Linux. But again it can takes 20 years to be accepted into the mainline Linux kernel [2].
[1] Cognitive load is what matters:
https://news.ycombinator.com/item?id=42489645
[2] 20 years later, real-time Linux makes it to the kernel:
https://news.ycombinator.com/item?id=41584907
if you look at this from the riscv angle, moving from "u-mode only vm that doesn't use paging under the hood" to "u+s-mode vm with sv39" isn't an enormous jump in complexity imo
i think i might teach it starting as like, "sv21" (page tables aren't nested), then pose real sv39 and the tree structure as the solution to making a sparse mapping over 512GiB
then moving on to the idea of having a TLB is simple, especially if students have already been introduced to hashtables
https://en.wikipedia.org/wiki/Zilog_eZ80
may be the best realization. You can almost run a real operating system on that chip in the sense of a 24 bit supervisor that could run multiple 16 bit processes (if you could get CP/M to run you have one heck of a userspace) Unfortunately you can't trap the instruction that switches back to 24 bit mode.
Would be nice too if a 24 bit supervisor could contain a 24 bit application, that requires some kind of memory management and I'd imagine something a little lighter weight than the usual paging system, say something that maps (0..N) in logical space to (i+0..i+N) in physical address space. I like the "access register" concept
https://en.wikipedia.org/wiki/Access_register
but would probably have a small number of memory banks such as 16 or 64. In a load-store architecture it doesn't seem like much of a burden to add a bank id to stores and loads. In return you get not just more RAM but also separation of code and data, video memory, file mmap(ing) and such.
What bugs me is how to control the mapping of memory banks to physical memory, on one hand you want the supervisor in charge so it can be used for memory protection, on the other hand some programming techniques would want the speed of changing the memory banks from user space.
The 80286 was a turkey because it didn't meet the minimal viable product level of being a target for a 24-bit OS that could virtualize DOS applications. It was already crazy fast and became affordable pretty quickly but it seemed tragic that it couldn't do that.
That's mainly because it was designed before the IBM PC was a success and backwards compatibility with DOS applications important.
One of the assignments was to do a simple multiply (i.e. add in a loop).
Rather than do that, my friend simply altered the program and created a new MUL command.
The teacher was not amused.
My first 360 assembler class our first assignment: add two numbers, fault, print the crashdump and circle the answer to the ADD in the printout.
I wrote an RPN calculator, had it do a series of calculations and print the result. Turned that in.
The teacher wrote on it "You have 24 hours to turn in the required assignment"
From the instructor's perspective, it sounds like you didn't even try to do the intentional assignment (which the fault/crashdump/find info in the crashdump was the most important part), and either did your own thing or sent in someone else's assignment because you didn't look closely enough at what it was. Yes, this is a big accusation, but it happens every semester.
The professor just wanted you to learn what you were supposed to from the assignment, so that you could build off it.
Your cynical take on it now shows that you still don't understand how to teach or the importance of that fundamental skill.
Is there a place for fun, exploratory projects? Of course, and I always incorporated them into my syllabus. But there's also a place for structure. If you want to explore along the way, then that is even better! But you still have to meet the incremental checkpoints along the way, otherwise you have not demonstrated that you have met the core competencies of the subject, as reflected by your grade.
There's a lot of shortcomings in modern university education, but I loved teaching, and would still be doing it today if it paid reasonably. My biggest headaches were the cheaters and the know-it-alls, mostly because neither of them knew enough to respect the subject matter and its importance. The students that made me the most proud, though, were those that trusted the roadmap, and in the end were able to do bigger and better things than they thought that they were capable of.
It's not about extra work. We did plenty of it, btw. We just didn't always tell you about it. The University system referrs to it as "service", and you, the student, are the beneficiary of it, either directly or indirectly.
"Teachers don't want extra work." I'm sorry, but this sounds like a toddler stomping their foot and saying "Mommy and Daddy don't love me because they went to work today instead of staying home to play with me."
I'm glad that you wrote the RPN calculator. That's cool! I'm disappointed that you use this as an opportunity to bash the educator who agreed to take a low salary so that they could help you learn important things that are fundamental to your craft and chosen field of study.
I understand that it's a sensitive subject, students that stray from the boundaries of the lesson plan, so often to no good effect. Making work for an overworked teacher, I have complete sympathy.
But understand that there's also room for the teacher that doesn't stomp their own foot and petulantly demand every student stay in the herd like good little scholars.
Maybe even encourage exceptional students, maybe suggest something more appropriate for them, another class perhaps. But no, that certainly didn't happen, just the deliberate ignoring of the effort spent, because it didn't serve their tiny objective.
Just that one weary, autocratic bland statement, telegraphing as sure as a digital signal "I am the teacher, and you will not do anything I don't sanction, like learning way more than I demand for today's lesson"
> "I am the teacher, and you will not do anything I don't sanction, like learning way more than I demand for today's lesson"
The person explicitly made it clear that while you can be encouraged on your own initiatives you still need to be graded/ranked on a common scale with your peers and hence the need for some sort of standardization.
For example, in a recent HN discussion on "Software Developer Productivity" lots of folks were arguing on how you can have no common scale/metrics and so you can make no comparisons at all between developers which is plainly wrong.
Education is about giving everybody a minimum standard of knowledge/tools needed for the Modern World and that is what a Teacher/Professor focuses on. Every Society needs to revere/honour (and pay well !) their Teachers/Professors because they are the ones who build the next generation.
Never underestimate how few fucks teachers give about anything beyond their small bubble.
- Mark Twain.
> I wrote an RPN calculator, had it do a series of calculations and print the result. Turned that in.
The teacher wanted you to learn:
1. how to write a (very simple) assembly program for 360 to add to numbers
2. how to create a fault
3. how to print a crashdump
4. how to read a crashdump
By using a RPN calculator, you learned none of these objectives.
Which, by the way, required accomplishing those bullet points many times over. Obvious to anybody who's ever written anything in assembler.
it is, but the thing is that these simple building blocks end up quite far away from an actual production level outcome that someone looking at a computer might see or interact with. It's hundreds of levels deep.
Someone with curiosity and eager to learn will be able to easily learn these foundational layers. Someone looking to "get rich quick" and be ready and employable ASAP won't.
I just wish I could find the time to sit down and design the damn thing.
1) Virtual Machines: Versatile Platforms for Systems and Processes by Smith and Nair - Seems to be a comprehensive subject overview book.
2) Virtual Machines by Iain Craig - Seems like a more hands-on book with languages and VMs.
3) Virtual Machine Design and Implementation in C/C++ by Bill Blunden - Seems like an hands-on implementation.
If somebody who has read any of the above can add some comments that would be helpful to everybody.
In a deep sense, virtual machines are what make computers computers, as we use the term today. Turing's 01936 paper about the Entscheidungsproblem hinged on virtual machines being able to emulate one another.
Other than the above i can only think of studying "QEMU Internals" for more knowledge - https://airbus-seclab.github.io/qemu_blog/
Most people who want to learn a bit about how their machines work would be better served by taking an operating systems course. Same goes here: if you only have time for a short tutorial, I recommend "Writing my own bootloader" instead. [1]
(This is not meant to say that the Write your own VM tutorial is a bad tutorial; only that, in my experience, most people who'd do it would be best served by a different subject)
[1] https://dev.to/frosnerd/writing-my-own-boot-loader-3mld
Can you please elaborate why it is bad to use LC-3 for studying computer architecture? I do understand it is completely different from real hardware, and too simplified. But from the perspective of writing CPU emulators, is it bad for me?
https://news.ycombinator.com/item?id=42519788
You say you want students to be able to write emulators/simulators etc. (you list a bunch of software.engineering tasks). I say writing an LC-3 emulator won't give you any transferable skills for writing, say, a Gameboy Emulator (still a 30yo system), because these educational arhitectures are bad and abstract away precisely the things that you have to get right to do that, leaving you with something trivial to implement: a C array for "memory" and a few bitwise operations.
Since we're largely looking at the same things and coming to orthogonal conclusions, I suggest we agree to disagree on this one.
Yes, there are fewer instructions.
https://en.wikipedia.org/wiki/Hacker%27s_Delight
which mainly depend on conventional numeric representations.
There have been decimal and hybrid computers built since the 01970s (especially for pocket calculators, but new decimal instructions were still being added to Intel's lineup in the 8086 with AAM and AAD, and I think decimal continued to be a consideration for the AS/400 into the 90s), but MIX is really kind of an 01950s design, with:
- word-addressed memory
- five bytes per word (plus a sign bit)
- one word per instruction
- six bits per byte (in binary realizations)
- possible decimal
- a single accumulator
- sign-magnitude
- no stack, and
- a nonrecursive subroutine call convention relying on self-modifying code.
You can find one or another of these characteristics in machines designed since 01960, but the combination looked old-fashioned already when it was introduced.
Let's look at just one thing every programmer has to deal with, memory.
On an LC-3, the address space is exactly 64KiB. There is no concept of missing memory, all addresses are assumed to exist, no memory detection is needed or possible, and memory mapped IO uses fixed addresses.
There are no memory management capabilities on the LC-3, no MMU, no paging, no segmentation. In turn there are no memory-related exceptions, page faults or protection faults.
When an x86 machine boots with 1MB of RAM, the 4GB address space still exists in full, but accessing certain addresses will cause bus timeouts, crashes. One must track and manage available memory. There's a BIOS, and manually probing memory locations may trash its critical structures. There's INT 0x15.
I picked memory arbitrarily but you run into the same limitations no matter what you pick. Would a students who was educated on LC-3 know how a computer keeps time? Of course not, there's no PIT, there's no CMOS clock. Would they have thought about caches? Nope.
Oh, but wouldn't a student who implements a timer emulation extension for LC-3 learn more about timers than somebody who just learned to use an x86 PIT? Alas, no. There are 20 equally easy and reasonable mathematical ways to implement a timer abstraction. A good 15 of these are physically impossible on real hardware, out of the remaining 5 two would be prohibiitively expensive due to electrical engineering reasons, one has never been implemented in real hardware due to historical accidents, and two are designs that are actually in use. So to write timer emulation that teaches you anything at all about how actual timers work, you'll have to look at and understand a real architecture anyway.
That's why educational architectures are so contraproductive. They abstract away exactly the things that make modern computers modern computers. One comes away with fundamentally wrong ideas about what computers do and how they actually work, or could work.
It's like learning to drive in GTA: in principle, there could be plenty of skills that transfer to the real thing, but in practice you'll prefer to teach how to drive to the person who didn't play GTA at all.
They are usually not for learning about application or even systems programming, in assembly and otherwise. They are about CPU architectures (and some surrounding concepts) themselves. The goal is to be able to quickly build your own emulator/simulator (like the VM behind the link), and maybe an assembler. Or, coming from the other side, given a working emulator/simulator, implement a high level language compiler, such as for a simplified C language.
In both those goals the CPU architecture is geared towards quickly converging towards a working system. So of course they don't require you to implement (as someone learning architecture engineering) or even use (as someone learning to build a compiler) an entire MMU, much like your first project car likely wouldn't contain a computerized double clutch transmission. Instead, they are simplified CPU architectures that allow you to make steady progress, without being slowed down by the tedious complexity that a real architecture brings for reasons that are not relevant yet when starting out.
Then, once that is achieved, you can freely add things like an MMU, a cache controller, maybe shift towards a pipelined or even superscalar architectures...
Besides, a very large class of computers don't have many of the things that you mentioned. For one, even very advanced microcontrollers explicitly don't even have an MMU, because that would destroy latency guarantees (very bad for automative controllers). For the rest, I've got to say that there is a certain irony in complaining that computer architecture students don't know about "modern computers", while in the same breath mentioning things like INT 15h, segmentation, and x86 PITs, as if we were in the 1990s.
> There are no memory management capabilities on the LC-3, no MMU, no paging, no segmentation. In turn there are no memory-related exceptions, page faults or protection faults.
Sounds an awful lot like a Commodore 64, where I got my start. There's plenty to learn before needing to worry about paging, protection, virtualization, device discovery, bus faults, etc.
It sounds like it's not teaching the wrong things like your GTA driving example, but teaching a valid subset, but not the subset you'd prefer.
It seems like a student would need to know a significant amount of coding in order to learn those abstractions in an interactive manner.
And by learn them I mean learn them (not just following a tutorial), organizing the code for a fully working x86 architecture is no joke.
But a student with that level of skill probably doesn't need to learn the x86 architecture so intensively, they are probably already employable.
I am asking this seriously, by the way, not trying to nitpick. I'm trying to put together a free course based on the video game Turing Complete[1] but from what you're saying it sounds like it might not be very effective. (to be clear the goal is to teach programming, not Computer Engineering)
[1]: https://store.steampowered.com/app/1444480/Turing_Complete/
My working assumption throughout was that the people in a computer architecture class already had 1 or 2 semesters of other programming courses where they worked in a high-level language, and are looking to learn how computers work "closer to the hardware". And educational architectures create a completely false impression in this domain.
If I had to teach assembly programming to people who never programmed before, I'd _definitely_ not want to start with x86 assembly. I'd start by teaching them JavaScript so that they can program the computers that they themselves, and other people, actually use. At that point they'd be ready to learn computer architechture through an x86 deep dive, but would no longer need to learn it, since, as you said, they'd probably already be employable. But the same goes for learning LC-3, and much more so.
To be honest, my opinion is only that educational architectures are a poor way to learn what modern computers actually do, and while I think I have good reasons for holding that particular opinion, I don't have the breadth of experience to generalize this specific observation into an overarching theory about teaching programming and/or compsci. I hope your course will be a useful resource for many people, but I doubt listening to me will make it better: my experience does not generalize to the domain you're targeting.
MIPS was, I think, a decent compromise between academic architecture and industrial architecture. There was a decent amount of practical stuff the students had to deal with without having to eg do the dance from x86 real mode to protected mode.
One of the most memorable lectures though was on the second last day. As it turned out I had gotten through all of my material one say early. We had a review/Q&A session scheduled for the last day but needed 1h30 of filler. I ended up putting together a lecture on JVM bytecode. Throughout the course I would often use C <-> MIPS ASM throughout examples because the 214, which was a prerequisite, was a C course. All of their first year work had been in Java. Anyway, we took real Java programs and disassembled them down to bytecode and walked through it, showing how it’s very similar to hardware assembly but with extra instructions for eg placing objects on the stack or calling methods on objects. The class ended up going almost an hour over because everyone was oddly engaged and had a ton of questions.
The first one covers the lowest level of how logic gates are physically wired to produce a Turing-equivalent computing machine. For that purpose, having the simplest possible instruction set is a pedagogical must. It may also cover more advanced topics like parallel and/or segmented instruction pipelines, but they're described in the abstract, not as current state-of-the-art industry practice.
Then, for actually learning how modern computers work you have another separate one-term course for whole machine architecture. There you learn about data and control buses, memory level abstractions, caching, networking, parallel processing... taking for granted a previous understanding of how the underlying electronics can be abstracted away.
Reasonable adults might have reasonable questions about those facts, such as, "what does any of that have to do with a computer?"
To my embarrassment, I realized they were completely right and my early exposure to software made me overlook some extremely important context.
So for these adults, the expectation of struggling through a few semesters/years of javascript is not an optimal learning route.
My hope was that working from the logic gate level up would at least provide the intuition about the relationship between computers (Turing Machines, really, not modern computers) and software.
However, I think based on your excellent critique I will be sure to include a unit on how "educational architectures are very different from modern architectures and I may have ruined your brain by teaching you this" haha.
Apparently a lot of CS courses in India still use the 8086/8088 too.
No cycle accurate anything. Simple I/O. Since I/O on the 6502 is all memory accesses, adding console is as simple as echoing whenever a character is stored in a particular memory address. No reason the simulate a VIA chip or UART or anything that low level.
Fun project. You can take it as deep as you like.
Far more 8051 cores are shipped than all of those combined, excluding ARM.
Books - https://www.safetty.net/publications
Video Lectures - https://www.youtube.com/user/SafeTTynet/videos
You can't always use the term "emulator" instead of "virtual machine", because while you could say wasmtime was an "emulator", you can't correctly say that WebAssembly itself is an "emulator". Rather, WebAssembly is the virtual machine which wasmtime emulates. It's also common to call emulators "virtual machines". (The Wasmtime web site mostly calls wasmtime a "runtime" and WebAssembly an "instruction format", FWIW.) And of course a running instance of an emulator is also a "virtual machine" in a different sense.
I think it's also reasonable to use "virtual machine" in the way you are describing, and it has some overlap with this last sense of "virtual machine". Perhaps in your current environment that is the overwhelmingly most common usage, but that is definitely not true elsewhere.
Really, a virtual machine is literally what it says on the tin: something that isn't a physical computer (machine), but presents itself as one. You know, with a virtual CPU or a virtual disk or a virtual keyboard etc., whatever a computer would have. If it doesn't present itself as a machine, it's not a virtual machine.
Calling JVM a virtual machine just because it interprets some bytecode the same way a machine does is like calling me a virtual doctor just because I read text the same way a doctor does.
No, it isn't and doesn't. What in the world does it have that makes you think it pretends to be a computer? A bunch of pretend registers and a memory limit?
https://docs.oracle.com/javase/specs/jvms/se23/html/jvms-2.h...
I just mixed it up with Dalvik having multiple registers, but that makes it even worse. At least normal machines have more registers.
And literally here:
The following section provides general guidelines for tuning VM heap sizes: The heap sizes should be set to values such that the maximum amount of memory used by the VM does not exceed the amount of available physical RAM.
https://docs.oracle.com/cd/E21764_01/web.1111/e13814/jvm_tun...
Maybe instead of telling me I'm clueless, tell me how you're right? What do you see in it that is actually virtializing a computer in your mind? Because I'm pretty sure that if someone sold you a VM service and it turned out to be just them running your programs under the JVM, you'd be kind of mad.
In the article the author even states as one example emulation a classic console, but it’s clear from the article that there are many other possible VM’s from the definition they provide.
Anyways the point is VM is abstract, and there are many many types of them. Simulators, emulators, hypervisors, etc… are all VMs, but also there are VMs that don’t quite already have a name for them yet because they’re strange.
I don’t mean to be rude at all - quite the contrary, I want to be respectful - but I also want to be clear about these terms for people who are interested in learning.
I don't think the distinction you're advocating for actually exists. 'Virtual machine' is commonly used for any software that executes machine- or bytecode, irrespective of the reason. This can include virtualisation, but you also commonly see the term used for language runtimes: e.g. Java's JVM, Ruby's YARV (Yet Another Ruby VM).
The one area you don't actually hear the term that often is in emulation, and this is in part because most modern emulators have tended away from emulating entire systems, and towards techniques like dynamic recompilation (dynarec) of emulated software.
Context and convention are important. This subthread is filled with people who are technically right, but choosing to cheer for an unclear and idiosyncratic terminology. A VM means hardware virtualization. Choosing other meanings is at best confusing and at worst a terrible security bug.
I always thought that it made sense to use the word for both things, and I never even thought twice about it when people used it for programming languages.
It would have to be a very strange conversation for someone to misunderstand someone else in the way you mentioned.
Sure, in precisely the same way that a kitchen knife and a sword are the "exact same concept underneath". But if you use the words interchangeably you'll confuse people.
And I'll just repeat the point, if you tell your boss your web app is "running in a VM" when you just wrote it in Java, you'll be fired for some really terrible security analysis.