Pine64 SOQuartz Blade Review

Hey Murray,

Welcome! So PCB services are a lot more advanced than this. I mean everyone has heard of PCBWay right? Look at the PCBWay order page. It’s not 2013. They will assemble the board for you. It’s not just PCBs. There’s a thousand services just like PCBWay. You wouldn’t even want to use PCBWay if you’re big. That’s just one *everyone* should know about because even YouTubers do paid promotions of it. This is easily available and cheap. There are actually paid staff that will guide you through all of this. They’re really good!

I remember the era you’re talking about. It was only a few years ago. Back then they wouldn’t assemble. If you look now though they *all* assemble. This stuff has advanced incredibly quickly. It has advanced faster than you or I or anyone could possibly keep up with that isn’t working with PCBs/boards every day. It has changed so fast that the scene is actually completely different today than when Ivan started the project. That’s just another part of my point.

I don’t believe I apologized to Ivan. I’m not prepared to do that yet. I apologize to you guys for not understanding some of the finer details of how this kickstarter camnpaign works. To be honest with you I think the whole project is trash. I don’t see it. I don’t see how this is better than just using a real BOARD service (not PCB, I shouldn’t have called it that because like I said that’s 10 years ago thinking). I do not believe this kickstarter project is the way to do things. I also definitely do not apologize for my overall criticisms of YouTubers and the way a lot of this stuff is covered on there.

I feel bad for people who have their money wrapped up in it as I’d imagine that we will get more clone releases before you get that board in your hands. At least now I understand people can pull their money out until the April deadline. That was another misconception I had.

You could probably just upload Ivan’s design to PCBWay or another one of these companies and get one a lot faster. It will definitely cost more than $100 if you just get one but it will turn out fantastic. Maybe someone could even pool a bigger order and order a bunch of them and get the price down. Which is exactly what I am suggesting Ivan should have done with the entire project in the first place.

As far as doing a background check on Ivan I was never trying to background check him. I’m not trying to figure out if he’s done shady things in the past. I wasn’t looking into his personal history. Background checking people is not what I do. I just saw a shitty project that costs a hell of a lot of money and is way, way overdue to be delivered. I started looking for reasons that was the case. I mistakenly found malice where the answer was actually incompetence. I assumed he was bad when really he’s just doing a bad job. I’m sorry if this is too harsh / mean for you guys but I don’t think you’re understanding where I’m coming from here.

We don’t have to wait for it to be delivered to judge whether Ivan’s process was more expensive than Pine64s. It was. We don’t have to wait to see if it will take longer than Pine64’s board. It did. Therefore we don’t need to wait to judge whether this was a bad way to do this project. It was. It took longer and is more expensive than if he just used a traditional process. Period. That is already a fact *today*.

I don’t care how great of a guy he is at all. It’s still not done. It should have been done by now. I’d never accept any company taking this long either and would just tell people this project is either stalled or dead.

It’s pretty rare for me to make a factual mistake on the site. I think my mistakes just show how bad and mind boggling this project is. I really don’t see it as about me but for any parts that are about me I think I’ve apologized and explained / accept that it’s not a scam. This is a bad, way overdue project. We have no idea when you guys will get your gear. None. The fact that there’s even so much room for interpretation is pure incompetence. Nobody can figure out what hell this guy is doing from the outside. If I wanted to background check him I’d have to join the Discord but I honestly don’t care. I don’t care how great of a guy he is. I only care why his project is so expensive and so late wnen this is not something that is hard to do in 2023.

Thanks so much for your response on the alternative boards. I definitely think the SOQuartz is worth a shot (or the SOQuartz Blade). There’s a lot of ways to use these. You don’t have to give up the CM4. You can use these together or interchangeably. I think it’s definitely a powerful tool for everyone to have!

I really appreciate your kickstarter comments about scams too. I don’t hate kickstarter theoretically but one does need to be careful. The more things like reward tiers and stuff I start seeing the more skeptical I get and you should definitely understand that. Once you start seeing a lot of games being played with the pledges it’s really not a good sign.

Again, I apologize for calling it a scam, but it is a terrible project. I don’t take that back. I’m not on board and I won’t be ordering one. I’d be very curious when you guys actually receive these though (legitimately). If I’m wrong about the product I’ll buy one if this just comes out and is widely available and everything is great. I just see no reason to expect things to go that smoothly from the track record. There is another way to do things and it is called literally the way everyone else does things except for Ivan.

Thanks for your comment, I hope I wasn’t too harsh but I just wanted to explain my positions. I agree 90% with you but I think I’m still just not there with the project overall even if we can understand why the project is the way it is. Take care!

Hey Razor Burn,

Thanks for sharing that! This is the #1 article on my site now. It has blown past everything else in only a few days. In fact my traffic overall is up about 25% sitewide since I published this. Most of it is amp traffic so I must be showing in people’s Google News and other feeds. I’d say people are very interested in products like this. It’s more popular in traffic already than even my big hits over the past few months including articles with 100+ comments.

I thought this might drop off in a day or two but my site doesn’t really work that way ever (even though I know that’s the traditional publishing model that you get all your hits/clicks in the first day or so) and indeed it’s only picking up steam. This is ranked/Google traffic and today is higher than yesterday already surprisingly.

It’s hard to say if it’s because they’re looking for information on that project or if they’re looking for information on the Pine64 Blade. Either way they’re finding out about both and I think that’s a great thing! I think his project will end up being the more premium option (in both features and price) and there’s definitely a place to have a low-end option like this and a high-end option using a CM4 and that future Blade launch.

This attention is going to bring more clones as well. Count on it. This project has kind of been flying under the radar. I really didn’t know much about it outside of Jeff’s video that I’d seen a while back. Pine64 might have been the only one that noticed this was being made and then when they released their clone nobody knew what it was (nor do they seem to be marketing it because they’re working on mainline kernel support for the SOQuartz as well as the Blade). It was not just released recently or anything like that either. Just hardly anybody noticed or covered it best I can tell.

That’s why this was so easy to rank on. Jeff is the only other person who has ever covered it. Now that the other companies see the Toms Hardware interviews about the major Blade project at least 2-3 more of them are probably making one of these now. This just seems to not be a thing very many people knew was even being worked on (or already existed in the case of Pine64’s).

I’ll even name one. I’ll be surprised if Radxa doesn’t come out with one of these as well with all the compute module stuff they’re already doing. If that kickstarter isn’t starting to move *immediately* into production there’s a good chance we’ll see one from Radxa first (or at least announced and probably going up for pre-order). There’s honestly not a moment to lose. The hour is quite late already. Radxa has not been disappointing with their recent releases either. If they weren’t already going to do this it would be almost irresponsible for them not to do it now. It’s such an obvious play for them and they have a *ton* of Compute Module experience already.

It’s understandable they’d be promoting it and doing interviews as the goal was just reached. Only time will tell if that attention / promotion at this time though was in fact a gift or a curse. I’m not sure they understand how much danger they’re in and how late it already is right now. I’m not sure they understand who they are going to be competing against very shortly here even if they aren’t worried about Pine64’s version. If they aren’t worried about Radxa’s version or some of the other cloning companies they’re completely insane. Radxa would be coming for the mid and high end both. It will take them a fraction of the time anything on the kickstarter project took because they’re literally a professional board cloning company with many years of successful releases. They already have all of their partners / facilities and they know how to work with all of these parts and produce them quickly and cost effectively.

If it released today they’re only competing against Pine64. Surely nobody thinks that will be the last though. I definitely don’t think that. I again then go back to what is the lead time for the project? Because I kind of have a better idea of the lead time of the unannounced clones than I do the kickstarter still because I know how long it will take them to clone stuff like this from years of experience watching and covering them do it slowly to the Raspberry Pi and now slowly to everything else. It’s less than a year realistically even with prototyping and all of that. They’re just really, really good at cloning boards (quickly and cheaply). They weren’t always though. It took them a long time to get good at doing it with a lot of old and terrible forgotten models along the way paying for the experience that they now have.

I’ve no way of knowing how many people bought the Pine64 Blade from this article because I don’t make any commission from Pine64 so I don’t have the numbers. The traffic #’s tell me that a lot of people are for sure hungry for something like this whether they are trying the Pine64 one or they are jumping into the kickstarter project. I’d imagine the answer is that both are happening depending on people’s needs / preferences and that’s a positive outcome for sure.

Take care!

Hey Razor Burn,

Great points for sure! So I’d rate August 2023 as pretty good if they can hit that. I don’t think they’ll be in disastrous shape if that’s the kind of timeline we’re looking at here as far as other blade clones are concerned.

It’s anyone’s guess what will be available/announced in 6 months though is the only caveat I would add. I mean at this point you have to wonder if the Compute Module 4 is about to be passed up altogether by a RK3588/RK3588S variant. I’d guess that Radxa and others are working on 8-core Compute Modules that will use the CM4 connector/form factor.

One could argue “well the CM4 is really cheap compared to what those will be”! Sure, but if people have enough money to be building clusters of blades with CM4 and the cost of the kickstarter blade then I would say it should be no problem for them to afford a Radxa / Orange Pi / Pine64 / whoever 8-core RK3588/RK3588S version. If you’re using the CM4s with high memory capacity those are not cheap or easy to get either.

What I’m saying is that I won’t be jumping in the kickstarter. I’ll buy one when it releases if it makes sense to still do so. I’m saying it may not though. This is a high-performance product and I don’t expect the CM4 to be the high-performance compute module choice for very long. This already happened with the RPI4 and nobody is trying to impress other people with a RPI4 cluster anymore because that’s a lot more well-understood to be obsolete. I don’t think the CM4 is far behind it. I think we’re a RK3588S or RK3588 variant away from the CM4 looking just as old/obsolete as the Pi 4 does today.

It seems risky to pre-order a cluster that won’t be available for 6 more months when not only will there likely be more blade clones announced/available by then but it might not even make sense to use the CM4 at all by that point given what has just happened in the SBC market with the standard non-compute boards.

Again, if they were available today all of the math I’m doing here goes away. In 6 months though with what has been happening? The Orange Pi 5 and Rock 5B launches were not even 6 months ago. Neither of them. Not even close. Think about how much has happened in that time and how much sentiment toward those types of boards have changed. Think about how the Pi 4 is viewed now vs. before those boards were released. That was only the very tail end of 2022 like November/December.

We have at least twice that long to go until this board will theoretically launch so from my perspective it just seems like all bets are off. I know I mostly mentioned the risks with the project but there’s also risk with the CM4 frankly also not making a lot of sense to use anymore by that time so I wanted to explain that in depth. The CM4 is almost as old as the Pi 4 but the Pi 4 has a year or two on it. It too will be made completely obsolete this year I’m officially predicting.

Building a high-performance cluster of these seems insane to me. They aren’t cheap or easy to get and if that changes by August the most reasonable explanation would be because nobody wants CM4s anymore because there’s a RK3588/RK3588S version with 8 cores and potentially even PCIe 3.0 if they go with the RK3588 proper. Again, 6 months ago nobody thought the Rock 5B or Orange Pi 5 would stomp the RPI4 so bad that it’s widely considered to be obsolete (I mean yes, I thought it would and I pre-ordered it for that very reason but I didn’t think that so many other people would think it so quickly). Those only launched about half that time ago.

People can keep saying the “better support” arguments and depending on what they’re doing it may still make sense for them. For an objective person who isn’t tied down really strongly into an ecosystem though or dependent on proprietary libraries/code within the Broadcom chips themselves I can’t imagine that by the end of this year that it will make any kind of logical sense to a reasonable person to build something like this with Pis/CM4s. Just like if you aren’t tied in with Raspberry Pi it’s a no-brainer to look at the Orange Pi 5 vs. a RPI4 and be like, yeah give me the 8 core monster that I can just buy today at a normal price.

If people reading this don’t think the exact same thing is going to happen to compute modules I don’t know what to tell them. The Pi ecosystem has very little hold over a reasonable person investigating the best hardware to build a cluster with. The people who are comfortable with the Pi may build one of these but I think they’ll have cluster envy pretty quick when they see a RK3588 cluster of these.

The CM4 is not new technology. We’re literally just waiting for the same thing to happen to it that happened to the Pi 4 in my view. It’s just another reason the hour is very late for this kickstarter project. The CM4 will not be the performance choice for very much longer even for doing something like this. We already know what’s coming with the RK3588/RK3588S. We already know it can’t compete. There’s not really much of an indication they’ll even try to compete for the high end. It’s going to be hard for them to even catch up at this point.

The CM4 is the last holdout for Pi performance fans. I’d have to imagine that is who is buying this blade. I’d just advise them to be prepared for it not to look very attractive as a solution to other people by the end of the year. The CM4 is going to end up being the cheap option just like the Pi 4 will be only considered the cheap option SBC (when it becomes available in sufficient quantities, which it still isn’t, and won’t be any time soon). It will not be a serious performance option just like the Pi 4 is already not a serious performance option today. That will happen this year.

Given that it seems like a tough sell to build a cluster of like a dozen of these to me. That’s what like $100 each blade if you use the Pro version and then what, $100 for a 8GB RAM CM4? Or are people using lesser CM4s? That would be $1200 in blades + $1200 in CM4s for $2400.

Well if you had a RK3588 version you’d only need half as many to have the same # of CPU cores. If someone had to buy 6 RK3588 versions (let’s say they’re $120 to give the CM4 the benefit of the doubt) and let’s say Radxa also somehow managed to make their blade cost $100. That still only comes out to $600 + $720 for a total of $1320.

It would only take $1320 to build a cluster with the same # of CPU cores using a more modern CPU. That’s vs. $2400 for CM4s + the pro blade. That’s what I mean by “if this still makes sense to buy at that time”. You can also what I see what I mean by saying be prepared for this solution to look *very* unattractive to other people by the end of the year. Those are really really ugly numbers if you just think through it. This will be something that is embarrassing to say you invested that much money in when you see how cheap you can do it with the RK3588/RK3588S and the upcoming competition.

If anyone doesn’t see it coming they haven’t been paying attention to the rest of the SBC market. It will also require twice as many blades and compute modules to make a cluster with the same # of CPU cores and that is *definitely* not greener to use twice as much hardware for the same power so not even the green part will be too impressive when you only get 4 CPU cores per entire blade/module out of it. I don’t think there will be much to be proud of as far as being green when people build a cluster as powerful as yours that uses half the modules/blades (and cost less money).

Even if people take exception to the values I calculated you can change them to whatever you want. Make the Radxa blade and the Radxa CM8 (or whatever you want to call the RK3588 version with 8 cores) cost more than my calculations. It won’t make any difference because you just need a lot less blades and a lot less hardware to hit the same #’s using a more modern CPU. It won’t change the outcome / conclusion even if people quibble with me putting the CM4 at $100 for the 8GB version or whatever their issue was with my numbers. Don’t they see? It doesn’t matter. The cores outnumber the CM4 2 to 1. There’s no version of this where the CM4 isn’t the way more expensive solution than what everyone should already know by now who remotely pays attention to SBCs is coming.

Just so nobody mentions it yes I understand the RK3588 CPU cores are 4 big/4 little and it’s not a 1:1 comparison with the CM4 cores. Again, it doesn’t matter. Break that down however you’d like and add / subtract some CM4 blades or Radxa blades and it will still cost a lot more. If you look at the benchmarks for the CPU though even taking all of that into account it’s undeniable how much more performance they are capable of. I just simplified it for the explanation in this post but again it will not change the outcome if you tweak or quibble with these.

Their answer to these issues would be “but the Pi ecosystem / support!”. Watch how little that means with numbers like that. Watch how quickly it changes. We’re already watching it in the non-compute module world. I remember the Pi ecosystem being built and if people think the “support” it enjoys today always existed they either weren’t there or they have very short memories. I helped build it and I can assure people that it wasn’t just as bad as the major alternative boards today, it was *worse*. The alternative ecosystems are rising fast and I’m helping to build them as are countless other people around the world. The support is catching up real quick on these alternative boards as more people buy and use them.

The reality is that Raspberry Pi just hasn’t done anything. For years. They can’t even produce their old models in sufficient quantities. There’s no new Pi 5 coming they say in 2023. They’re pretty much just flailing around and we’re waiting to see if they can even manage to produce their *old* boards in sufficient quantities that anyone even notices. Meanwhile the competition is moving so fast that they have already released their “Pi 5” generation (and people like it!).

I know my post makes me the villain to a lot of Raspberry Pi fans. I covered Raspberry Pis because they were cheap and easily available for everyone to buy and participate in technology with. I cover Orange Pi and Radxa boards and other boards now for that exact same reason. That’s called being consistent. I’m doing the same thing I’ve always done.

I have no loyalty whatsoever to Raspberry Pi and in my eyes it is them who has betrayed us (the maker community) by prioritizing IoT manufacturing operations over us (which is where all the Pis went, a lot of them are in some of my older Helium crypto miners, you’d be sick if you knew where they went). I really don’t think it’s going to get any better for Raspberry Pi fans at least not in 2023. They’re less attractive every single day.

Here’s where a lot of Pis went. The Pi 4s all went into RAK helium miners (google it). There were *millions* of these sold (it was the last cryptocurrency bubble/gold rush). They sold you out for crypto miners (me). I still got the Pis this whole time because the Raspberry Pi foundation does not give a shit about any of you. They sold them to miners / crypto companies and anyone with a big stack of cash. That’s the truth. The people with money (namely people ordering thousands+ Pis at a time) are more important than you. Have you been able to buy Pis? Look up when those Helium miners were manufactured (hint, it’s the “shortage”).

Hell no I’m not a Raspberry Pi fan because I know where those Pis have been going this entire time unlike any of their fans. I watched it with my own eyes as MILLIONS of these miners (there were 3.5M backordered at one point) were shipped out while no normal person could get them (and still can’t!). They’ve been doing this for *years*. This is not a guess. Those are the “commercial customers” being prioritized. Giant mining companies (and industrial-size scalpers with the means to set up their own companies and order huge quantities to scalp you with).

Hilariously even the mining companies are switching away from Raspberry Pi now. Here’s a updated RAK miner with a Rock Pi powering it instead. That’s because they literally exhausted all the supply and had to find alternatives. There were no more Pi 4s to be sold to them. If there would have been though they would have got all of those too. Even if it becomes available though they won’t go back because there’s better (and cheaper) options available now. That was a one time thing because the Pi was in a unique position at the time in affordability/availability which they’ve completely squandered and lost at this point.

Bringing the old thing back at the old price (or even with some price raises in some cases like the Pi Zero) is not going to do the trick. They’re way behind at this point. They’ve wasted their past several years on top while their competition has been working furiously. It won’t work with most of the commercial companies and I don’t think it will work with most people as they continue to be exposed to things like the Orange Pi 5 and Radxa Rock 5B. In the Orange Pi 5 review someone just posted they upgraded their Home Assistant instance from the Pi 4 to the Orange Pi 5 and they can’t believe how incredible of an upgrade it is.

Now I just showed where the Pis have really been going for the past couple of years. Does that look consistent with their mission of promoting open source and everything they pretend to be standing for? Do you think they should have sold millions of those to miners instead of the creator community? I don’t and I actually financially benefited from everyone getting screwed by having them run my mining equipment this entire shortage. I still don’t like it even though I benefited because I actually care about the missions they pretend to care about and know they’re faking it and just moving as many boards as they can to whoever has a big enough pile of cash to be treated as a “commercial customer”.

The shortage didn’t just come from nowhere. The miners started the shortage frankly just like with GPUs. They just went into different types of miners (mostly Helium). Their response to this shortage was to stop selling them to individuals altogether and to prioritize commercial customers with the idea that their livelihood was at stake. If they had not changed their policies the shortage wouldn’t have been *nearly* as painful for anyone. You still would have had your shot. The decision was made that your Pis would be diverted though and would never make it to an official reseller. They went straight to companies like the ones making all of those miners. That’s who they gave up on the creator/maker community for. You be the judge on whether that lives up to your expectations of what you have of Raspberry Pi in your head. It didn’t for me.

It’s an act at this point. It wasn’t always an act though. I don’t know when they sold out but they did. Please, anyone who doesn’t believe it google what is powering these Helium miners and when everyone bought them. Good luck trying to explain those Pis in miners away. That’s real and you won’t believe how many of them were made (even though a lot of them aren’t being used now because Helium miners make less than 10 cents a day for the most part, what a waste of Pis). It’s honestly undeniable and I’ve known this for a long time even though I don’t talk about it much. It honestly makes me sick when I see they are true fans and really believe in the mission and all that when I know the truth and how unworthy they are of the reputation they hold to some in the open source community (a community I’m a proud member of). It’s really gross and I’m sure for a lot of Raspberry Pi insiders it must be really gross too. This is just another business trying to make $$$ (or at least that’s how it’s being operated).

I’m not saying that one should switch their allegiance to Orange Pi or Radxa. I’m saying stop thinking that way. It’s not real. Don’t be a fan. They don’t know your name or care. They’re a business and their job is not to be your friend. Buy boards appropriate for whatever project you are doing that are being sold at the time for a reasonable price and offer good performance. You don’t have to start wearing a Radxa t-shirt. Please don’t. Just take the “fan” stuff out of it and look objectively at the scene and get what you need. Don’t be so afraid of the “support” anymore on these other major boards. It will be better than you think / are being told. If Radxa stats putting out bad or overpriced products or starts fumbling on their availability I will turn on them instantly. On a dime.

Thanks for sharing your thoughts! It’s not an easy question for sure. A lot can still change before this will be available and that’s why I’m advising some caution here. We’re still 6 months out from when this might be available too and this is what I see *today* having absolutely no insider information or knowledge of these companies or their operations. It’s really just common sense and paying attention realistically and I hope my lengthy post explained why. Take care!

Hey Razor Burn,

Thanks, I’d heard of the Radxa CM5 before but didn’t know if it had this! I did look up a reference here. It does indeed have it:

Eight-core Rockchip RK3588 really packs in the power, offering up to 16GB of RAM and a 6 TOPS neural network coprocessor

Yep, that thing is nasty! I don’t see it available for purchase yet but this article is from 5 months ago. It’s definitely going to be ready before too long if they announced it 5 months ago. It’s not just my speculation that it’s coming. That’s definitely happening for sure.

This is exactly what I was talking about but the details you gave helped me find sources. I knew without seeing this that this would be the case because we already know from their flagship board launches such as the Rock 5B and Orange Pi 5 that this is basically their current-gen CPU across the lineup of major RockChip partners. In other words they’ve already showed us their hand with their flagship boards this generation. This is their current generation CPU for sure and we’ll see it across the lineup (whether it’s RK3588 or RK3588S which it sounds like the Radxa CM5 will be the S variant).

The Radxa CM5 is going to take the performance crown for compute just as the Rock 5B / Orange Pi 5 have in the more traditional SBC form-factor. It’s just a question of when it launches but since it was announced 5 months ago I’d say that likely before the kickstarter ships this will be available for at least pre-order if not outright.

I’d honestly love to get the kickstarter Blade and review it still. It would be interesting to test it with the Radxa CM5 as well as the Pine64 SOQuartz (which Jeff did test that one on his pre-release one he has with mixed results). It’s another reason to be cautious though because compatibility testing with other boards with the kickstarter one has not been promising thus far. It doesn’t mean the Radxa CM5 wouldn’t work better with it but it’s a concern and there’s reason to believe from the previous SOQuartz compatibility testing that there may be issues.

I see every reason to wait and see though. He has enough money so you don’t have to worry about that sinking his project if you wait until the deadline. People could just wait and see what the landscape looks like. The truth is in 6 months or even 3 months this may look very unattractive even as inconceivable as that might seem today. This stuff has changed very, very quickly. I honestly think anyone who is cautious and takes this approach will probably find with each passing day it seems like a worse and worse idea and that there are more and more options popping up that look better every single day. If things are going fantastic then go ahead and grab one before the deadline closes.

It’s nothing personal against Ivan or the project. It’s just way too late. They literally couldn’t have entered a more dangerous market at a more dangerous time to be stuck in prototyping for 3 years while promoting it so all the clones who know how to do this quickly can beat them to the punch and be more than ready for them. They could have had an empire if they launched this a year or two ago but they won’t even be the first to release this now. Pine64 gets that honor.

It’s not clear to me it will launch before the CM4 is obsolete (at least for high performance compute module work). That’s up to Radxa and how fast they’re moving (probably fast). From what I can tell this is due to inexperience but there has never been a more unforgiving time to be flashing all of your plans like that and have your production still be this far out. I know that’s kind of the kickstarter model but that’s why you really can’t use it for an idea like this. It needs to be weirder / more unique and not in a market absolutely dominated by some of the best cloning companies in the world or this is going to happen (and not just with SBCs either). You’re begging to get cloned before you can launch doing that (and they were). Now we’re at the point where the CM4 itself is about to be cloned/obsoleted and is several years into it’s lifecycle. Not a good outlook in my view.

Thanks for sharing that and everything else you’ve said here!