Banana Pi BPI M2 Zero drop in replacement for Raspberry Pi Zero for RaSCSI?

[avadondragon]

I did a little digging but couldn't find a definitive answer. All Raspberry Pi's have become ridiculously expensive unobtanium recently which has me looking for alternatives for various Mac related projects. MacIPgw, RaSCSI, etc. Of course the Orange Pi has been suggested as a good alternative for a MacIPpi but I figured it makes more sense to use an old a$$ intel based tiny PC since they're dirt cheap and even one from almost 10 years ago blows a R Pi 4 out of the water performance wise. Unless TashTalk becomes readily available I don't think I need the GPIO for anything for that application. Am I wrong there?

Anyways the point of my post is that I grabbed a couple of differn't types of RaSCSI about a year ago but I very stupidly only grabbed 1 R Pi Zero 2 W at their normal price when they were available. Now I'd like to get my second RaSCSI usable but Pi prices have me livid. $15 board for $150?!

The Banana Pi Zero BPIO is pin compatible with its Raspberry counterpart but I wasn't sure how the programming side of things worked. Doesn't anyone happen to have any insight about this?

 

[avadondragon]

Oh also I've been out of the retro mac computing game for quite some time. What's up with the PiSCSI - RaSCSI name change? I noticed I couldn't find the original project and it was under a new name it seems.

 

[MarcTremblay]

I will have to agree on that. We need PiSCSI to work with Pi alternatives. I mean WTF am I looking at on AliExpress? This used to cost 10$ three years ago! It's almost retro technology by now, the prices are jacked up as F.

 

[avadondragon]

So with persistent digging I found that I'm not the only person who's had this thought AND I answered my own question in the process.

Looking through the git repo for the project I found that someone had opened an issue with this EXACT feature request a couple of months ago. It has already been implemented and that is the reason for the name change - It is no longer Raspberry Pi exclusive. Strange that it was so hard to find that info. It hasn't been mentioned anywhere that I can find.

It was hard to tell from the comments but it may not be very efficient yet. Perhaps that is why it hasn't been advertised?

Edit: Yep it looks like that is the reason.

 

[MarcTremblay]

Yes, well from what I could find last fall, it seems like Banana Pi is too slow (or perhaps it's just the current state of the code that's not optimized enough for that platform) so they ditched the idea? I don't know. Does anybody know what the state of Banana Pi support is? Or Orange Pi? Or Mango Pi? Or any F Pi that doesn't cost an arm and a leg? PLEASE?

 

[avadondragon]

The silver lining to the obnoxious Raspberry Pi shortage is that it is forcing an across the board push towards universal Pi compatibility. It looks like a lot of people are working on this right now. Many alternatives offer higher performance specs than Raspberry Pi's. Unfortunately a lot of code was optimized to use direct addressing of the GPIO pins on the Raspberry Pi for the sake of performance but that makes cross compatibility impossible. I have no idea how the code for this project was written or how easy it will be to port. It seems like it shouldn't be too long before they work out a standard for all this were you just say I want to talk to GPIO pin 1 and the compiler checks to see what board it is running on and plugs in the right address.

 

[Skate323k137]

The silver lining to the obnoxious Raspberry Pi shortage is that it is forcing an across the board push towards universal Pi compatibility. It looks like a lot of people are working on this right now. Many alternatives offer higher performance specs than Raspberry Pi's. Unfortunately a lot of code was optimized to use direct addressing of the GPIO pins on the Raspberry Pi for the sake of performance but that makes cross compatibility impossible. I have no idea how the code for this project was written or how easy it will be to port. It seems like it shouldn't be too long before they work out a standard for all this were you just say I want to talk to GPIO pin 1 and the compiler checks to see what board it is running on and plugs in the right address.

Sure would be nice. I have a project right now meant for a Pi 3, but the only one I have is my DNS server right now. And I'm not in a hurry to take a Pi 4 away from my RaSCSI just so I can steal the 3 for my other project (BMC64, C64 case and keyboard with R Pi C64 Emulator).

 

[avadondragon]

Sure would be nice. I have a project right now meant for a Pi 3, but the only one I have is my DNS server right now. And I'm not in a hurry to take a Pi 4 away from my RaSCSI just so I can steal the 3 for my other project (BMC64, C64 case and keyboard with R Pi C64 Emulator).

For any project that doesn't need the GPIO I try to find an alternative to using a Pi. Right not I'm playing with using obsolete cell phones for such projects. There are ways to run Linux along side Android and even a few ways to run it bare metal if you've got the right phone. A lot of phones people are throwing out are monstrously powerful even compared to a Raspberry Pi 4. If you want a challenging project try getting your DNS server running on a phone!

 

[Skate323k137]

For any project that doesn't need the GPIO I try to find an alternative to using a Pi. Right not I'm playing with using obsolete cell phones for such projects. There are ways to run Linux along side Android and even a few ways to run it bare metal if you've got the right phone. A lot of phones people are throwing out are monstrously powerful even compared to a Raspberry Pi 4. If you want a challenging project try getting your DNS server running on a phone!

Haha I like your thinking! Very resourceful

 

[MarcTremblay]

It looks like a lot of people are working on this right now.

Where?

 

[avadondragon]

Ugh. I'm having a really hard time finding any of the links I previously found. I spent a few hours looking into this the other day. There are already drop in replacement libraries for use with Python so you don't have to change your code. That's not particularly helpful for something like PiSCSI though. The thing that looked promising was something like a Linux kernel api implementation that was being actively developed but I have no idea how to find it again.

 

[MarcTremblay]

From what I can tell, nobody is publicly working on an alternative so I'm really unsure where you read that "lot of people are working on this right now". Seemingly, nobody is. The issue is that it's not as simple as one would think: Even if the LIBRARIES could be recompiled and ultimately would be compatible, we would still have to change the code and the physical adapter at the very least since most other Pi alternatives don't even have the same number of GPIO pins and often not even in the same order either. Looks like a major pain to me. Pi prices will be going down in about 1 year from now (but will permanently be approximately 20% higher than before the pandemic, a price upgrade that is coming from the Raspberry foundation directly, to face inflation primarily). Meanwhile, this year, I'm pretty sure we'll see new Chinese SBC that will even exceed Pi 3B+ specs for 30$ or 40$ but too bad they won't have the same GPIO header layout, so in this application it would be useless (to me).

The closest match to that would be a LE POTATO (30$ USD) as it features slightly increased specs compared to the Pi 3B+ plus, it can boot Raspbian OS and it has an ALMOST fully compatible GPIO header, but the downside is that it's very chinese and quality insurance is really hit and miss. Lots of DOA units. https://www.amazon.com/Libre-Computer-AML-S905X-CC-Potato-64-bit/dp/B074N5B8KZ

 

[avadondragon]

Given how hard it is to find anything regarding cross pi GPIO compatibility I agree with you. I guess it's just my wishful thinking that given the current market conditions this SHOULD be happening. There are a definitely a few Raspberry Pi alternatives that have been intentionally designed to be pin compatible. I mentioned Banana Pi zero because it was the first Pi zero clone that I found which had a compatible pin layout. Looks like the GPIO Linux kernel API has been around for a couple of years now. I have no idea how much performance impact using a user space API would have on things but it would greatly ease the compatibility issue if it works right. My only somewhat relevant experience is with Arduino though which is a completely different animal so I'm really talking about stuff I know little about except in theory.

 

[MarcTremblay]

For all it's worth, there is a new BlueSCSI (v2) now which works with a 10$ Pi Pico, but it doesn't support WiFi