@mcc i continue to be baffled by analogue making and trademarking an "fpga" logo, when fpgas are something they definitely didn't invent and only have an extremely nebulous claim towards in general
@leo well, the trademark is for the word "openFPGA", and in the UNIX world there is a fine, long tradition for closed consortiums appropriating "open" (Open Group, OPENSTEP etc)
@crzwdjk Plan is to port my Amaranth rule30 CA program to Pocket ("program" may be overstating the situation) and then next is either FFTs for audio or a triangle!!
@mcc Yeah; They have been doing similarly for decades now, partly because the government requires the warnings. The gov doesn't want people hacking certain devices, including communication devices like TVs, radios, phones, CBs,etc. to interfere with radio frequencies or send protected/patented/copywritten or gov stuff out of the country. I still remember when kids took over radio frequencies to broadcast their own stuff for fun, sometimes or politics upsetting all sorts of top gov agencies.
@mcc You might be surprised how many people used all sorts of gadgets & how many types of gadgets actually did/do communicate or can be hacked into communicating across radio frequencies & more now with the internet. Hackables included atari & other games, CBs, transistors, etc. The 60s+ were filled with new, young & older techies learning on their own how to wire their devices to talk with anyone anywhere. Very political time with the early spread of binary+ and computerization.
@barbaraebj Oh, I've got some experience with RF actually! And yeah, you can do some wild stuff. I think there's still a bit of a… how shall I put it… "free experimentation" culture like that running in China.
@mcc Welp, toss it in the trash, you tried. That or 👩💻 hack 👩💻 the ever-living $#!% out of it... Then you can agree to whatever you want :) It's not technically a software lock so you violated no laws.
It's a minor thing, but I am giggling so hard right now. I just compiled one of Didier's Verilog sample programs, copied the netlist and some JSON to the Pocket's drive over USB-C, exited drive mode and selected the core, and… I just programmed an FPGA! I didn't write the code, but even just USING these things is supposed to be super hard! And all I had to do was copy files over USB and now it's running on a screen! CIRCUITS! And they're DRAWING things! That is so cool!!
Okay. This was my mod. It was supposed to be the letters L33T, the stars were supposed to be multicolor, and the "desert Chrome" colors were supposed to be different.
@Farbs No but seriously though, I am going to make a serious go at making an interesting game in pure gateware, and I predict this is going to be unbelievably gratuitous and not justify the work put into it on anything other than a conceptual/personal pride level. From what I've been told, accessing RAM (because I do have to talk to the RAM chip in pure wire language) is going to be hell
@didier@mcc@Farbs my SDRAM controller is mostly good to go, if you don’t need multiple ports. You could add multiple ports yourself without too much trouble
so here's the thing. it turns out FPGAs aren't really, compared to modern-fab CPUs anyway, that fast. it was being explained to me today the literal clock rate at which any given part of one of these FPGAs runs is in the 200MHz range. That doesn't mean it runs at the speed of a 200MHz CPU because CPUs usually can't do anything in a single cycle and FPGAs often can. But it's still a little bit surprising that number isn't higher
@Farbs So your clock rate is low. Pocket tops out at 720p because it physically can't send bits to the HDMI port fast enough. Floating point math is possible but considered an extravagance. And usually on these devices you don't have very much RAM. So if you want to do something really impressive with an FPGA, you have to find something that really, really leverages whatever advantages you do have.
And what you do have is, parallelism. Absurd, totally ridiculous parallelism.
@Farbs If I've got these numbers right, the Cyclone V in the Pocket has 50,000 "LUT"s. Which is essentially a wad of gates which can a compose an arbitrary logic function for 1 output bit.
But the thing is, all 50,000 run every cycle.
So that means, if you design things truly optimally, you can do 50,000 operations at once. 50,000 little bitty CPUs running in parallel, but each CPU is kinda slow & does the exact same thing over and over every step.
And yeah, the more I hear about these the more I think they could be awfully useful for pixel shading, convolutional audio processing, stuff like that.
@Farbs They're absolutely fantastic for audio DSP. In fact they're fantastic for DSP at speeds that makes audio rate sound very, very slow. Apparently an FFT is very easy to represent in FPGA.
"Maybe not so great for gameplay logic though" True, but! Slap a PicoRV32 in there, or ten, and you can run the gameplay logic on that. You can download a CPU and fit it into one little corner of your FPGA :D And it's RISCV, so you can just like… compile to it. With regular clang. Well, almost regular
@mcc I wonder what sort of application would be required to get FPGA hardware built into consumer grade computers. It'd have to be something equivalent to 3D cards and the way they massively outperformed software rendering back in the 90s.
Unfortunately all I can think of in this space right now is waves hands in vague manner AI.
But even then, everyone's business model is AI as a service so they absolutely don't want it running on end user hardware.
@Farbs@mcc Basically none. It's been tried many times, but it doesn't work because the perf and area gap isn't just large, it's absurdly large. FPGA doing "specific logic X" versus a custom chip doing the same logic is probably 1/100th the speed AND 100x the area AND 100x the power.
@Farbs@mcc For a very short time, people were using FPGAs to do bitcoin mining because it was such a specific task that CPUs did so abysmally badly (by design!). But then people figured out how to do them well using SIMD and/or GPUs and that advantage vanished.
It's a swiss-army-knife that plugs into a computer via USB at one end and at the other end can connect to an enormous variety of wire protocols/"plugs", such as basically every type of serial, outdated things like VGA and PS/2, and the unique protocols used by things like flash memory chips and CO2 sensors.
@TomF@Farbs The Glasgow has a very weak FPGA at its heart, but even this very weak FPGA does a great job of bitbanging any serial protocol you can imagine—which even a powerful CPU would struggle to do so consistently (because CPUs have interrupts, operating systems, other things going on), and of course you weren't ever going to put an Intel i5 into a device like this anyway, so to justify itself the FPGA really only has to outperform whatever analogous embedded chip you'd've used instead.
@mcc@Farbs A lot of FPGAs just come with an actual CPU in the corner for just this purpose - most traditionally 68k based, but I assume ARM is pretty popular these days.
@TomF@mcc@Farbs that’s one way the pocket stands out. It actually has another FPGA (a cyclone 10) instead. But same idea, that’s what runs the ‘OS’ services and interacts with most IO stuff…
@didier@mcc@Farbs The MiSTer is basically the perfect use for FPGAs. It can reconfigure to "emulate" basically any 8-bit and 16-bit platform including arcade boards and all sorts of esoterica. In this case the reprogrammability of the FPGAs is crucial! It's a neat project.
@TomF@mcc@Farbs a lot of the Pocket cores for retro stuff are actually Mister ports it seems. I plan on going the other way and porting my console to Mister once it works on the Pocket.
@pervognsen@didier@TomF@Farbs It appears they held ALL Pocket orders ALL year long until they completed a hardware revision in early August, and then blasted out an entire year's worth of orders at once. So if you order a Pocket RIGHT NOW you might be able to get one shipped immediately. Alternately, maybe at some point soon they switch over to the "nothing ships until NEXT August" batch, I have no way of knowing. And I think you may missed the chance for the glow-in-the-dark unit
@mcc@Farbs Yes - they are well-suited to being GPUs, which also have a ton of highly parallel ALUs but relatively abysmal access to memory.
Now, that's compared to an old-fashioned CPU. Modern CPUs are also massively parallel machines and the difference between a GPU and a CPU is kinda subtle these days.
@mcc@Farbs apparently simple designs can push 250Mhz thru on the Pocket. The problem becomes the time it takes to get all the gates stable when your design is complex. The higher the clock rate and the more complex the design, the more chance you have for meta-stability and then the compiler shouts at you…
Confession: After recording this video I spent like 10-15 minutes repeatedly re-reprogramming it unsuccessfully trying to make the bottom half of the text pink so I could say "look, bisexual colors!". Every attempt to make a pink just wound up with a different rusty brown. I eventually decided that my worth as a programmer does not hinge on my ability to eyeball what hex value will become pink after it's darkened an arbitrary amount. At any rate I think the current colors are pretty bisexual.
Incidentally if you look really close you'll find the Terasic JTAG "USB Blaster" has printed on it, in tiny letters, "Lead Free". I think this is supposed to make me reassured but really it's just making me worry what was going on here that I apparently ought to be worried about a JTAG connector not being lead free
Baby steps progress continues: I made a version of the Verilog sample project that contains animation and sound (an A440 tone). This actually is much less impressive than the last demo I posted, but I did more of it myself.
@acb 1. Yes, that's it. Most of the cores are MiSTer ports.
No, other than the (reasonable, not different from MiSTer) inconvenience of using Altera Quartus, it's quite straightforward. If you want to do it fast, you have to take off the back with a screwdriver and attach a $100 proprietary JTAG debugger but to develop this demo all I did was repeatedly bump it into USB-C mode and drag the rbf_r file on the PC into a folder full of JSON
Thanks to some absolutely heroic work by @whitequark it is as of last night now possible to make Analogue Pocket cores in Amaranth, which is the Python eDSL HDL I used to make this: https://cohost.org/mcc/post/2499788-my-first-fpga-project Basically you write a Python script that describes a circuit, and you get to just skip most of Verilog's unnecessary headaches. The chances of me making something cool for Pocket just went way up.
Here's Whitequark's template Amaranth core filling the screen with magenta:
Also, reviewing this thread I'm realizing if you read the Cohost post about my Doppler/Amaranth project then work down the videos/images in this thread you can basically watch my nail polish degrading in real time. I really need to put on a new coat before I try to take any more demo shots
Check out this weird naked Game Boy cart that comes with it. This is actually the most important thing in the entire devkit package because it's got a button, an LED, and a serial port on it, exactly the things you'd want for debugging a design
Ok so this is very exciting!! I need to write a Cohost post to really explain this but. Check out this custom FPGA core I built! It's unpleasant to look at and unpleasant to listen to but it's important because:
Its synchronized video animation and audio with all "business logic" in pure Amaranth (Python). (@whitequark did the video writing code, I did the color patterns and the audio.)
The ugly color patterns (which my phone camera can't pick up even up close) test Pocket screen resolutions
For this screen test app, I wound up doing audio synthesis (the 16-note sequence of square waves playing C octave) at the bit level by generating the needed bits in real time as each bit was needed while writing those bits to a sound chip using the i2s serial protocol. This was a fascinating experience that I never, ever want to do again.
Okay, here's something a lot prettier looking (maybe not prettier sounding… depending on your tastes). After taking two whole weeks to make a "hello world", I made this little "rule 30" cellular automata scroller in about two hours (about 3/4 of which, inevitably, was debugging). This is running in pure gateware (fixed circuit in FPGA) on an Analogue Pocket. The original version was much higher res but I reduced it to take a video. Gonna try adding controls next…
@mcc@jplebreton There's probably a way by putting a corner of your phone between your eye and your glasses but depending of the size of the phone it'd be hard to balance it.
Also beware of pink eye if your phone isn't clean.
Vaulted over the final mandatory gate to making "something useful" on the Analogue Pocket: Input! I now have a demo for the Pocket buttons. This is written in Amaranth and despite not being that complicated managed to uncover bugs in both of Amaranth and Yosys! I feel proud. I think.
Been experimenting with the "developer key". I was able in my own code to talk to the button (somehow haven't got the LED working with my own code yet, I'm missing something). I also got some Verilog code from a nice person on the Discord named "Boogerman" which takes serial bitmap data over the developer key's USB jack and displays it on the screen. Here I'm using the "Boogerman code" to display my avatar. Notice the cable's going into the back— it's plugged into a fake Game Boy cart in back.
I'm having this weird problem where this app is supposed to be black and white, but often when I launch it I find it randomly displays black lines with green shadows. It looks as if my RGBRGBRGB pixel bytes sometimes come up offset by one, so that it's instead interpreting the bytes as _BRGBRGBRG. One time I even got purple and black and white, with weird line jittering similar to the Amiga core.
Still don't have the PLLs working in my cellular toy core well enough to release a beta BUT here is me successfully running a simple Rust program on the Pocket, on the beta of agg23's general purpose riscv core! On the first try uploading all it did was show an off-center picture of Rick Astley and play a loud ~200hz tone, but fortunately that was exactly what I programmed it to do
@mcc this is great because of the meta rickroll aspect. I expected to see him bop around and assure me I'd never be given up, but instead I hit play and was Toned.
The original version of the demo app written by agg23, as demoed on his Twitter https://twitter.com/iam_agg/status/1723042566235455636, displayed Rick Astley and played an mp3 of "Never Gonna Give You Up", but I did not have an interest in compiling an mp3 library and did want to test if I could generate audio such as a 200 hz tone, so.
Anyway check out the pretty blue snow pattern that the beta RISCV core displays before you load an app. It's not intentional but I'm trying to convince agg to keep it
Okay, now there's something I can be proud of as a "first program" on the Pocket! I took one of my very old experimental glitch audio programs and ported it to Rust then had it generate a Hilbert curve fractal kinda thing.
Warning: Very loud
Warning: Experimental glitch techno, may be confusing or at least very annoying
Warning: My nail polish looks bad again
I have now created MINIMALIST BREAKOUT. No blocks. No balls. Just the paddle. No score display, you have to remember your score (it's always zero). What's that you say, this is boring? This is INDIE GAMING, you're not supposed to have fun you're supposed to have meaningful experience in which you contemplate the lonely futility of existence. The buzzing droning sound that plays continuously? That's "anxiety". It's a metaphor. Okay, game's done. I will not be holding a question and answer period
By the way, the glitch noises app from the previous video now has controls and it's incredibly cool, assuming you have my particular tastes in art and music. Will post more when I'm ready to post a download.
I have decided to abandon my attempts at ART and COMMUNICATING EMOTIONS and have embraced CRASS COMMERCIALISM by adding the ability to both win and lose to my "break out" style game for agg23's RISC-V core for the Analogue Pocket. It now has a win condition, a lose condition, music, and a "final boss". This was supposed to be a prototype for something much more complex but it's playable enough I think I will release it as sample code when the core leaves beta (after adding RNG and maybe a score)
@mcc this looks so fun! Love following this series :)
Tempted to get an Analogue Pocket, seems really neat to dev on.
Also, the video loops so perfectly it caught me off-guard at first.
@allpurposemat It's not your easiest option for (I believe the Ambernic handhelds can just run normal linux executables) but developing for it is such an interesting learning experience, and there is so much potential (I want to make little 3D video cards in gateware that my Rust programs on the RISCV chip control…)
Here is the C program I was trying to write. It doesn't really exist for any purpose except to be written in C. We didn't have a C program that drew graphics for the platform, so I wrote one.
This was supposed to be a simple, minimal demo app. All it was supposed to be doing was demonstrating how to access the video and sound registers in C. I got carried away. I got so carried away. I don't know how to say this without it sounding like a joke but I'm not really aware of time having passed over the last eight hours, I got consumed in bringing this Thing into the world. It's probably too complicated now to be accepted upstream as sample code. I regret nothing. It had to happen.
Forgot to link into my Pocket progress thread: Using a new feature in Firmware 1.2, I can now create cores which have a different screen resolution depending on whether they are running in handheld or docked (TV) mode https://mastodon.social/@mcc/111592441309751932
So basically every step of this cellular toy project has been a process of writing bugs and then accidentally discovering the bugged version is more interesting than the thing I was trying to create on purpose
So after many months of shipping difficulties I now have this Tang Nano 9k fpga dev board stick plus hardwired screen. Here's an Amaranth demo by @whitequark which I modified to make the little square move on it's own. Goal is to get my Pocket cores (https://pocket.runhello.com) in some kinda build infrastructure where they can all cross-build for either the Pocket or for this thing (and ideally also for the Colorlight board I've got).
For the maybe one of you who is actually following my Pocket releases instead of just following my development pictures, I have a much improved version of Cellular Toy locally but I will probably wait until I've improved the sound (read: do something about the high pitched screeching it makes if you get it into the state where, by its current rules, it would make a high pitched screeching)
Offhanded end of week project: I wrote a demo drawing sprites (using art by Miguel Sternberg) in motion with double buffering on agg23's RISCV (LiteX) core for Analogue Pocket. A couple things are missing (the sprites don't individually animate and I haven't tested how many it can draw at once) but it's probably good sample code already. It's not trivial because this FPGA core was designed for utilities not games, so animation is done only by writing to framebuffers in RAM.
Current "sprite animation on Analogue Pocket RISC-V utility core" project status: can now do about 16-17 sprites simultaneously plus smooth vertical scrolling with a simple background.
Next step: smooth vertical scrolling with a complex background, horizontal scrolling (this is impossible, BUT what I CAN do is rotate the entire screen 90 degrees clockwise and simultaneously switch to a framebuffer that looks the same but 90 degrees rotated counter-clockwise, so that vertical becomes horizontal)
I'm curious how the Analogue Pocket compares (hardware wise) to the FP-GBC (a full GBC clone on a Gowin FPGA that fits in an OG GBC shell that costs ~$150)
I did this big refactor last weekend where instead of unpacking the images into memory I kept them RLE encoded and drew the RLE directly to screen at draw time, which I thought would be faster because it would be reading less from memory and could skip transparent pixels totally, I even had a cool RLE damage system based on Regions from Mac OS 1, but this was SLOWER actually. Either I introduced too many branches in the hot loop or I broke an optimization where it was blitting whole 32 bit words
(4 cores here, none are on the main core inventory… yet)
There is now a new "beta3" version of Cellular Toy, it is a big improvement. It is now fullscreen in BOTH Dock and handheld aspect ratios and there are "Core Settings" (in the Analogue menu) for color schemes and access to all 256 elementary cellular automata (in beta 2 it was limited to four).
I'm not really sure how this happened but a couple hours ago instead of going to bed I started working on the scroll demo again.
This is… wrong… but it's actually pretty close to right… but it's far enough away from right it's convinced me I should go to bed instead of trying carefully thinking out the modular arithmetic here.
You might recognize this video as a less evocative version of the images from my thread here https://mastodon.social/@mcc/112141103375777088 "less evocative" coincides with "closer to correct"
OK, I don't know why I seem to be unable to work on this project except at times I specifically should be sleeping, but
I did it!! Using nothing but an underpowered CPU and the ability to relocate the framebuffer pointer, I can do infinite scrolling in 1 dimension through an endlessly autogenerated space (or like, I could pull a "level design" from a file, too). If you look upthread I stated this as my "next step" 42 days ago. (I only poke at this occasionally.) Next: 2 dimensions (alternating)
Update: 2 dimensional scrolling was a bust because it turns out Analogue Pocket currently draws the screen fully black for two full frames every time you change scaler modes, so rotating the screen can't be done as part of a smooth animation.
So, 1 month belated, here is the open source release of the demo in the previous post (plus an experimental cargo feature to set scaler modes):
I think I can still get 2D scrolling if I can convince litex to use a separate number for "byte distance between framebuffer rows" than it does from "width of each row", but (like the vsync bug separately making me miserable) to make this simple change I'm going to have to make sense of the dense, uncommented blob that is the litex codebase.
So that said uh if anyone reading this knows anything about the framebuffer/"vga" code in Litex mimes a "call me" motion text me
Spaceship can free fly with scrolling backgrounds and other sprites in the environment. Would you believe this core as configured can only scroll within an area 24 pixels wider and taller than the visible area? 100% of the fun in writing this is that everything is so much more difficult than it needs to be…
Please excuse the messy desk. And the horrifying vsync tear issue, that is LiteX's fault.
Current status: Added semitransparent shadows and I really like how they look. This'll look great once I have a better tile texture.
A very satisfying moment was when I realized you can cheaply darken an RGB565 pixel by just saying screen[idx] = (screen[idx] >> 1) & 0b0111101111101111;. Is this SIMD lol
Very excited as I've just figured out the reason why previous attempts to optimize my Pocket sprite drawing code actually made it slower, is because modulo is extremely slow on this vexriscv softCPU. This is SO unsurprising in retrospect but I hadn't realized it until now. Already getting significant speedups with minor tweaks and I think I can get even more.
I got too used to the "nothing is slow except branches and cache misses" era of CPUs. I was seduced by the decadence of the modern world!
@TomF I was excited that I would be able to outperform other cores on the platform by drawing NPOT sized sprites and textures but maybe I'm gonna get locked into power of two sizes after all.
Okay so I'm not going to bother to take a video, I'm not even going to bother to take a proper screenshot, but:
So I've now implemented spawning bullets when you press Y, but I haven't yet implemented the code to make the bullets move at their current velocity from frame to frame, so you're just zooming around dropping bullets which hang there endlessly, which you'll have to take my word for it is pretty funny
You know I got me one of them FPGAboys, but I haven't delved into it yet. I was arrested by an idea though. I had read about posits as an alternative to floats and thought, "Cool. Maybe in ten years they'll be available in hardware," then put it out of my mind. But with FPGAs it would seem possible to build your own posits.
Anyway just an idle thought. Love seeing your actual work running on the Analogue Pocket.
@shanecelis Right now I'm in this awkward position where I can EITHER
Make my own FPGA code which is 100% by me and 100% all as I want it
Run real Rust code on a full-stack CPU core somebody else made
But I'm not at the point yet where I can write Rust code and have it running on a CPU with extension instructions (like posit support) I invented or registers with fully custom behaviors I define (like a FPGA sprite drawing unit). I'm really looking forward to that point!
@shanecelis Well at some point I will have to address the problem of "I have no path to connecting this to anything I can make money from, but it's time consuming enough eventually it may get hard to do much of it unless I can find such a path" which is less fun. But for now I'm ignoring it!
@mcc unrelated, but I genuinely always love seeing your signature weeks old nail polish! I, too, dislike acetone. It's nice to see some representation in the wild 🥰
and I just got that from another user, so there's Knowledge around. And I believe you can script quartus via tcl just like you can with Vivado. I recommend joining the FPGAming discord, they have Mame and Pocket sections and the Pocket section has several hardcore quartus users
@mcc “no, I didn’t program it to display Rick Astley. There’s no data for a Rick Astley image anywhere in the rom. I tried reducing the program to a single return statement and it’s still displaying Rick Astley”
Yes, it is easy to build and compile your own cores with nothing but the retail Pocket unit and the same USB-C cable you use to install any third party core. Before I answer though can you indicate what language you believe you'll be writing the code in, and do you understand the difference between C and Verilog?
@mcc Oh to be honest i got one pocket to test and learn stuff you are doing in Amaranth 😅 and eventually be able to help. and to answer your question yeah i know my way around with verilog and c. did some basic stuff in Amaranth before also, want to make it a permanent tool at my job, but dont feel confident enough yet.
This is all in flux, whitequark had a template but she temporarily took it down, my template's main branch is out of date. Soon there will be a better template and I will have a semi-tutorial about amaranth on pocket.
Note no one on FPGAming knows about amaranth, the only people who have gone that route are me and whitequark.
@mcc I haven't been paying any attention to the openFPGA scene, I'm mostly using my Pocket to play carts I already own, but I'm excited to see development on it beyond just emulating other consoles
@mcc did they also send the JTAG easy access rear cover (or a new Pocket)? I want to play with my Pocket FPGA but don’t want to unscrew the back all the time (plus worry about bricking it with a bad Core, although I would imagine that’s very unlikely)
@crzwdjk@mcc we decided to not reimplement the Verilog harness because it had a bunch of weird code with no testbenches or debug interfaces that the console requires to run
in practice, it's sliced in two and written to the scaler by a ddio megafunction
@mcc ohh... from your firs posts i though it was just a retro-thing emulating the gameboy. that would be quite neat already, but it's a nicely packaged FPGA? 😲😳
@mcc thats the first sample that truly blew my mind when I started learning. The star field expecially, all generated from one clock signal… hardware is crazy… enjoy and don’t hesitate to ask I can help with anything.
@mcc funny you mention it. It’s on my todo list (https://didier.malenfant.net/plan) I’ll see if I can put something out today because it’s been bugging me on Windows too.
@didier Cool! I'm not sure yet if I'm going to be primarily using this (I'm also looking at another toolchain), but it would definitely be useful to have pf-dev-tools in my toolbox.
@didier I'll keep that in mind, thank you! And thanks for the super cool example projects :) Don't know if you saw but I made a minor mod of the starfield demo to prove I could: https://mastodon.social/@mcc/111021749800302023
@didier Actually! I just remembered, I had a second question.
I tried running your 04_fullres example, and it looks like this (1)
This was with me having to create the json file by hand (due to the pf-dev-tools problem); I converted (2) in the toml to (3) in the json
Does any of this look surprising?
Do you think something went wrong because pf-dev-tools didn't finish running?
Could there be an error in the sample as checked in?
@didier Okay. Great. Then what I'm seeing is exactly what I should expect.
I expected the square to fill the screen, but not for any specific reason, I just guessed.
And the problem I was having with pf-dev-tools was that when I ran pf make it would copy the files into the quartus project, but couldn't run it because it couldn't contact docker (because as described above I don't use docker for quartus on this machine)
@Tathar You don't strictly need the dev kit! I don't have one. It's not needed for installing custom cores, though I think it would save significant time and inconvenience when casually debugging one.
@mcc wait wait you didn't have to bootup a Win7 machine and run some Win32 program with one of those 20th century grey 3d UIs that is like 5 buttons 20 check boxes and a multi line textbox "output window" with interactions disabled to prevent input, but also that doesn't let you select text to copy to a clipboard?
@lambdageek Also some of the ancillary third party tools don't work because they assume you're invoking them on Mac OS , and running the Linux version of the Windows 10 program inside Docker
@mcc nice! I was looking at these devices a month or two ago. the screen on the analogue pocket looks amazing, but eventually I decided to cheap out and get a miyoo mini plus. FPGAs are much cooler though.
Add comment