I have never seen your content before and now I'm addicted. Love it. Good mid-strength beer. Good swearies. Shonky camera focus. Perfect. I also am craving one of these boards now.
Hahah thank you mate, much appreciated. And yeah pretty awesome board, a LOT of potential. I did a 2nd video touching on some ideas I have for it too 🤘
I wish those clusterboards had faster on-board switches. Just 1GbE is rather limiting when you do something like Longhorn across all the NVMe drives. It definitively isn't bad, but there is a lot of performance left on the table. Really love those boards - space saving, fit into small enclosures, can totally stuff them into a 1U server case and be totally fine. Thanks for the showcase!
Yeah I definitely get that, but I think there is also a limitation of those B2B connections that may make 10GbE infeasible? I'm not 100% sure TBH but have not seen a 10Gbps and/or fibre one yet! The fact it fits in 1RU is bloody amazing, did not consider that. Definitely going to 3D print some arms/ears to fit it in the 20RU I've got mounted up on the wall!
@@PlatimaTinkers I feel like we are getting closer to a point where you can buy a small ARM board with fast networking and at least one reasonably quick storage device interface, and once you have that you start to be able to do some very, VERY interesting things with distributed storage systems like Ceph. SoftIron has already kinda done this with their big ARM servers but it really doesn't have to be that large and complicated!
@@0xKruzr Getting so very close! As soon as XGMII becomes more common in modern SoCs, then we should start seeing some real leaps and bounds I think. For now PCIe via M.2 is likely going to have to be the way (which is 10Gbps for many it looks)
@@longinus665 I'll blame Google Translate - probably originated from a foreign design and just got mis-translated haha. Like "brick nogging" as a great example!
Nice work, need to populate that board with six Radxa CM5's. The video needs a part 2 that tests the NIC interconnect performance. They probably did it for cost but the NIC really needs to be 10GbE so you can connect all the modules to ethernet based webcams and perform AI inferencing on them in a mini AI cluster.
Hah yeah I have a CM5 lite that needs to go on there, and another new secret one on the way 10Gbps interfaces would definitely be the way to go. The RTL8370N used it limited to 8x 1Gbps though 😥 BUT, it can be hacked: github.com/libc0607/Realtek_switch_hacking/blob/master/RTL8370N-SR8808M.md
I have one of these boards. Not ideal but you can put an m.2 to 10Gbe adapter in the m.2 slot on each. I haven't done it with the super6c but I have with my turingpi2. I then serve disk images over iscsi.
Hey you're very welcome - it's a great product. Currently working on 10Gbps networking using M.2 and will then do another video! (And yes, I don't use Raspberry Pi 😋) The one thing that was very challenging was the limited documentation, which resulted in a lot of confusing about HDMI1/2! Would recommend updating it perhaps? Keep up the great work! (PS just saw you added the video on the website - thank you kindly 🥰)
@@PlatimaTinkers Yes, we have also noticed this issue. The product is primarily designed to be closely integrated with the Raspberry Pi's Compute Module 4 (CM4). Our product is also based on the open-source data documentation of Raspberry Pi, such as extending the board-to-board pins of the CM4. The Raspberry Pi chip can support two HDMI interfaces, and the two HDMI interfaces on this baseboard are actually the two HDMI outputs of the #1 CM4. Other CM4s do not actually require these two HDMI interfaces. We will reorganize and update this information on the wiki, listing incompatible devices to avoid the same issue you encountered. This will allow you to better focus on your projects without being affected by such minor details. Regardless, we greatly appreciate your feedback and consideration.
@@52Pi_Maker_Education yeah that makes complete sense! The issue was however that even with RPi CM4, they are HDMI0 and HDMI1, but your board says HDMI1 and HDMI2, and I am pretty certain your traces map HDMI2 on the board to HDMI0 on the RPi? So I was plugging into HDMI1 for everything, and not getting any signal! FYI would be awesome if the M.2 ports could be upgraded to PCIe Gen 4, as PCIe Gen 2 is a tad limiting in bandwidth when many CM4 boards have Gen 3! Eg with M-key M.2 ports you get x4 lanes, so 2.0/2.1 is only 2GB/s aka 16Gbps, but PCIe 3.0/3.1 nearly doubles that 😁
@@PlatimaTinkers The Raspberry Pi CM4 supports a PCIe 1x Gen2 interface, with a theoretical maximum bandwidth of 5Gbps, rather than PCIe 4.0 or Gen 4.0. This means it supports PCIe version 2.0, not the newer versions. This information can be found in the official Raspberry Pi documentation, which mentions that the CM4 uses two 100-pin high-density connectors, providing an additional second HDMI port, PCIe, and Ethernet interface.
@@52Pi_Maker_Education Oh yes I have no issue with that. Your 6C documentation states that the M.2 PCIe support is 2.0 though, hence I thought there may be some hardware limitation such as a PCIe switch used in your design (not just that of the CM4 board used)! Did what I explained about the HDMI port numbering make sense? I double checked the RPi CM4 documentation for that one too! 😊
Usually I try to record them on a Friday, and edit that night or Saturday, but recently having to move my recording setup twice, and now having to setup a new recording/camera system has really thrown a spanner in it. This one I actually started recording Saturday arvo, but it took hours, and then I didn't bother editing and uploading it until Monday arvo (local time). Regardless of when I get it out, I try to cap the time I spend on UA-cam videos to 6 hours per video, as that's all I can currently budget. What I do full time though? For one of our companies I do the accounts and video editing, for another I am senior system & network engineer, for another I do web design and accounts, then there's this one where I do the content and also run the ecommerce shop, and for the last one I don't really do stuff all but do need to keep an eye on everything to make sure it's working. So yeah, a tad busy haha
Hahahha no, I wouldn't have bleeped that 😂 And yeah same. www.reddit.com/r/gameofthrones/comments/twzzy/what_stood_out_to_me_the_most_during_the_prince/?rdt=36361
It's possible that the cluster requires all the compute modules to be of the same type and spec to function correctly. Traditionally the HID devices only connect to the first unit in the cluster and the other 5 are blindly slaved to it and assigned their tasks by it. It would be interesting to see if you can use it for your purpose but I believe you might have the wrong board.. you need a KVM switching board.
Nah good thought but doesn't work like that at all. Turns out if you watch the whole thing, it was working, just with HDMI 2 not HDMI 1. That's also not how Compute Modules work or how the Super6C is designed. These are all individual compute modules that operate independently. The B2B connectors are just power and IO, which ended up working as designed, the doco / instructions were just a bit misleading!
Hahaha yeah part 2 is on my list of things to do - I've got two more boards coming to add to it before I do that, and I might do some fun networking too!
Oh my man thang you for pointing that out! I could not find the email trail for the life of me, as I was going to send something to follow up, and started to think I was going insane 😂
Yeah I am REALLY thinking about stocking the Mars CM. Most of what I focus on stocking is bespoke in some way, eg eSBC / headless / embedded, or RISC-V. I branched out with the Core3566 boards just because I really bloody like them, and now I've got the Waveshare carriers... Would make sense. You know the day I stock a RPi is the day I have sold out 😅
@@PlatimaTinkers From own experience I would say 1GB RAM per CPU core is best (4GB RAM model for 4-core JH7110); less is limiting and more is not needed. Nothing wrong with leaning into existing RPi accessories. Also reach out to Forlinx about the FET7110-C SoM?
@@niceride I'd say 1GB / core is fine for SBC single user headless operation, but if hosting multi-user / public services I usually spec servers at about 12GB / core. Eg 256GB RAM 20 (physical) core. For SBCs with a GUI, I'd go 2GB per core, as min 8GB these days seems necessary for browsing the web without swapping in most cases! Re the FET7110-C, I don't see any particular use case for it. The JH7110 is just RV64 with RVV0.7.1. Forlinx did actually reach out at the start of month offering some content, but I turned them down.
@@PlatimaTinkers Maybe you are thinking of the TH1520 with the weird RVV implementation? JH7110 so far as I'm aware has none. It would be nice to have multiple sources of JH7110 CM4 modules in the spotlight. Better to see some more options and competition even for the same CPU.
@@niceride OH yeah right you are - the JH7110 is the U74 cores. My bad. But yeah so I've got the StarFive VisionFive 2 and both Milk-V Mars and Mars CM, so got plenty of access to JH7110 to tinker with if need be! If they bring out something new or bespoke I'd be all over it, as they didn't really seem to limit my selection, just said "pick what you want to review and we'll send it to you" 😅
Next time you use OBS, I recommend setting up just a single window capture rather than a screen capture. That way, scene transitions can be gentler, and you don't have to worry about accidentally revealing any notifications.
Valid, and I considered that, but thought I might want either a web browser or a terminal, and then it's also not full screen size unless I actually fullscreen it, then switching it is harder. Else I could scale it up. Stopped caring too much this time haha. And notifications aren't a concern for this one; the computer I put together is purely for these videos, doesn't have any email or messaging apps etc :)
@@PlatimaTinkers ah, nice. I don't have the setup for capturing video from a dedicated machine and have had to rerecord things in the past. I went somewhat deep with my OBS exploration when I first started using it and set up screens with looping gifs and whatnot for video calls and streaming some Python dev work a few times. Back when I still had to do daily stand-up meetings, I would set my camera to a loop of some baby squirrels I raised climbing over my arms until I was asked to turn it off for being distracting lol. Fun stuff.
@@k98killer yeah I only put this 'bitsa' together in the last two weeks so I didn't have to do a fresh setup each time. OBS definitely has a lot of good functionality I need to explore more! This was the first time I had used scene transitions. Hahahah that video loop sounds awesome. Reminds me of one I saw where a guy made a video of himself entering the room, looking at the computer, then backing out, and used that as his virtual background for Teams/Zoom meetings, and it took ages for someone to clock that there were two of him 😂
The switch on the Waveshare development board is the BOOT switch. When the switch is set to "on," the development board is in flashing mode, and it cannot boot into the system.
@@PlatimaTinkers Oh, okay, I really enjoy watching your videos; they've taught me a lot of new things. They're fantastic, and I'm looking forward to your next video.
@@52Pi_Maker_Education Thanks mate, glad to hear! This weeks one is going to be about a shitty $50 AliExpress camera that I'm sending back 😂 Note I just patched up my reply that was in haste above, and I also noted that the Radxa CM5 lists the DeskPi Super6C as a supported IO board!
Probably, though then you're virtualizing which *can* be different than true hardware. Also you can't virtualize every hardware you might want to test. And third virtualizing may or may not be slower and requires a lot of RAM. Also behavior of cluster software may be different with real hardware with real network latency. I think VMs are often times a good solution and you're probably right in this case, but it's not always the better option. Also real hardware is just more fun
1) Cheaper, maybe. 2) Easier, maybe. 3) ESXi? Eww (and that's coming from someone who is VCP certified haha) The power consumption would be higher, you'd not have GPIOs to test/work with, you cannot test DTCs and boot optimisation, and in any virtualisation there are inherent inaccuracies with how some code may run. ESX(i) also cannot do ARM/RISC-V, which is more my use case here! I do use Docker instances for testing OSes etc, but the final tests usually require hardware 😊
@@Crftbt Oh yes so very true. Technically that board was running 16 or so cores with 16GB RAM and presumably drawing less than 45W. I'm definitely going to have to do a 'part 2' and test that!
@@ChrisJackson-js8rd Yeah not sure if I left that in or cut it out, but that's a bit of my goal. Eg something very powerful on CM1#1 for IO etc, and otherwise make sure I have a good mix of arm32, armhf, arm64, RV32, RV64 and x86-64, or something to that effect!
I have never seen your content before and now I'm addicted. Love it. Good mid-strength beer. Good swearies. Shonky camera focus. Perfect. I also am craving one of these boards now.
Hahah thank you mate, much appreciated.
And yeah pretty awesome board, a LOT of potential. I did a 2nd video touching on some ideas I have for it too 🤘
A like and a comment for "this is for me not you"
Appreciated 😊
Suck it up 😊
@HansdeRooij That's the spirit!
I wish those clusterboards had faster on-board switches. Just 1GbE is rather limiting when you do something like Longhorn across all the NVMe drives. It definitively isn't bad, but there is a lot of performance left on the table. Really love those boards - space saving, fit into small enclosures, can totally stuff them into a 1U server case and be totally fine. Thanks for the showcase!
Yeah I definitely get that, but I think there is also a limitation of those B2B connections that may make 10GbE infeasible? I'm not 100% sure TBH but have not seen a 10Gbps and/or fibre one yet!
The fact it fits in 1RU is bloody amazing, did not consider that. Definitely going to 3D print some arms/ears to fit it in the 20RU I've got mounted up on the wall!
@@PlatimaTinkers I feel like we are getting closer to a point where you can buy a small ARM board with fast networking and at least one reasonably quick storage device interface, and once you have that you start to be able to do some very, VERY interesting things with distributed storage systems like Ceph. SoftIron has already kinda done this with their big ARM servers but it really doesn't have to be that large and complicated!
@@0xKruzr Getting so very close!
As soon as XGMII becomes more common in modern SoCs, then we should start seeing some real leaps and bounds I think.
For now PCIe via M.2 is likely going to have to be the way (which is 10Gbps for many it looks)
It is very exciting to see the Milk-V compute module boot on the Waveshare board. I got one for testing.
Oh awesome, glad to hear! And yeah I'm glad it works like that too.
The Boot switch on the Waveshare carrier board is to put the RPI CM4 in flash mode (for the emmc). So you normally want it off.
Aaah yep okay that makes sense! I never understood why it's always labelled 'boot' on boards, when it could be 'flash'. So be it though!
@@PlatimaTinkers Yeahh it's definitely confusing!
@@longinus665 I'll blame Google Translate - probably originated from a foreign design and just got mis-translated haha. Like "brick nogging" as a great example!
Nice work, need to populate that board with six Radxa CM5's. The video needs a part 2 that tests the NIC interconnect performance. They probably did it for cost but the NIC really needs to be 10GbE so you can connect all the modules to ethernet based webcams and perform AI inferencing on them in a mini AI cluster.
Hah yeah I have a CM5 lite that needs to go on there, and another new secret one on the way
10Gbps interfaces would definitely be the way to go. The RTL8370N used it limited to 8x 1Gbps though 😥 BUT, it can be hacked: github.com/libc0607/Realtek_switch_hacking/blob/master/RTL8370N-SR8808M.md
I have one of these boards. Not ideal but you can put an m.2 to 10Gbe adapter in the m.2 slot on each. I haven't done it with the super6c but I have with my turingpi2. I then serve disk images over iscsi.
OH that is a hell of an idea, VERY interesting. Thank you kindly!
Lol, thanks for your video review our product, and that's cool in trying different SBC on it. :P
Hey you're very welcome - it's a great product. Currently working on 10Gbps networking using M.2 and will then do another video! (And yes, I don't use Raspberry Pi 😋)
The one thing that was very challenging was the limited documentation, which resulted in a lot of confusing about HDMI1/2! Would recommend updating it perhaps?
Keep up the great work! (PS just saw you added the video on the website - thank you kindly 🥰)
@@PlatimaTinkers Yes, we have also noticed this issue. The product is primarily designed to be closely integrated with the Raspberry Pi's Compute Module 4 (CM4). Our product is also based on the open-source data documentation of Raspberry Pi, such as extending the board-to-board pins of the CM4. The Raspberry Pi chip can support two HDMI interfaces, and the two HDMI interfaces on this baseboard are actually the two HDMI outputs of the #1 CM4. Other CM4s do not actually require these two HDMI interfaces. We will reorganize and update this information on the wiki, listing incompatible devices to avoid the same issue you encountered. This will allow you to better focus on your projects without being affected by such minor details. Regardless, we greatly appreciate your feedback and consideration.
@@52Pi_Maker_Education yeah that makes complete sense! The issue was however that even with RPi CM4, they are HDMI0 and HDMI1, but your board says HDMI1 and HDMI2, and I am pretty certain your traces map HDMI2 on the board to HDMI0 on the RPi? So I was plugging into HDMI1 for everything, and not getting any signal!
FYI would be awesome if the M.2 ports could be upgraded to PCIe Gen 4, as PCIe Gen 2 is a tad limiting in bandwidth when many CM4 boards have Gen 3! Eg with M-key M.2 ports you get x4 lanes, so 2.0/2.1 is only 2GB/s aka 16Gbps, but PCIe 3.0/3.1 nearly doubles that 😁
@@PlatimaTinkers The Raspberry Pi CM4 supports a PCIe 1x Gen2 interface, with a theoretical maximum bandwidth of 5Gbps, rather than PCIe 4.0 or Gen 4.0. This means it supports PCIe version 2.0, not the newer versions. This information can be found in the official Raspberry Pi documentation, which mentions that the CM4 uses two 100-pin high-density connectors, providing an additional second HDMI port, PCIe, and Ethernet interface.
@@52Pi_Maker_Education Oh yes I have no issue with that. Your 6C documentation states that the M.2 PCIe support is 2.0 though, hence I thought there may be some hardware limitation such as a PCIe switch used in your design (not just that of the CM4 board used)!
Did what I explained about the HDMI port numbering make sense? I double checked the RPi CM4 documentation for that one too! 😊
Curious what you do full time. Since these videos seem to come out in a rush at the end of your day.
Usually I try to record them on a Friday, and edit that night or Saturday, but recently having to move my recording setup twice, and now having to setup a new recording/camera system has really thrown a spanner in it.
This one I actually started recording Saturday arvo, but it took hours, and then I didn't bother editing and uploading it until Monday arvo (local time). Regardless of when I get it out, I try to cap the time I spend on UA-cam videos to 6 hours per video, as that's all I can currently budget.
What I do full time though? For one of our companies I do the accounts and video editing, for another I am senior system & network engineer, for another I do web design and accounts, then there's this one where I do the content and also run the ecommerce shop, and for the last one I don't really do stuff all but do need to keep an eye on everything to make sure it's working. So yeah, a tad busy haha
42:15 Guess they forgot to put a "quit" in when they made you! 😂Glad you got it working.
Yeah pretty much 😂 Nothing a few cans can't fix haha
Did the bleeped out word rhyme with hunt? As someone that is enjoying an RDO today, it's a word that I use quite regularly at work.
Hahahha no, I wouldn't have bleeped that 😂
And yeah same. www.reddit.com/r/gameofthrones/comments/twzzy/what_stood_out_to_me_the_most_during_the_prince/?rdt=36361
It's possible that the cluster requires all the compute modules to be of the same type and spec to function correctly. Traditionally the HID devices only connect to the first unit in the cluster and the other 5 are blindly slaved to it and assigned their tasks by it. It would be interesting to see if you can use it for your purpose but I believe you might have the wrong board.. you need a KVM switching board.
Nah good thought but doesn't work like that at all. Turns out if you watch the whole thing, it was working, just with HDMI 2 not HDMI 1. That's also not how Compute Modules work or how the Super6C is designed.
These are all individual compute modules that operate independently. The B2B connectors are just power and IO, which ended up working as designed, the doco / instructions were just a bit misleading!
Make a PT 2 plz but eat something first 😂
Hahaha yeah part 2 is on my list of things to do - I've got two more boards coming to add to it before I do that, and I might do some fun networking too!
30:55 I that was me talking to you about the computeblade, you did definitely reply :)
Oh my man thang you for pointing that out! I could not find the email trail for the life of me, as I was going to send something to follow up, and started to think I was going insane 😂
8:50 is it powder coated?
Yuuuup. Eventually I got there (@16:14) 😅
Have you considered to stock the Milk-V Mars CM/Lite ?
Yeah I am REALLY thinking about stocking the Mars CM.
Most of what I focus on stocking is bespoke in some way, eg eSBC / headless / embedded, or RISC-V. I branched out with the Core3566 boards just because I really bloody like them, and now I've got the Waveshare carriers... Would make sense.
You know the day I stock a RPi is the day I have sold out 😅
@@PlatimaTinkers From own experience I would say 1GB RAM per CPU core is best (4GB RAM model for 4-core JH7110); less is limiting and more is not needed. Nothing wrong with leaning into existing RPi accessories. Also reach out to Forlinx about the FET7110-C SoM?
@@niceride I'd say 1GB / core is fine for SBC single user headless operation, but if hosting multi-user / public services I usually spec servers at about 12GB / core. Eg 256GB RAM 20 (physical) core.
For SBCs with a GUI, I'd go 2GB per core, as min 8GB these days seems necessary for browsing the web without swapping in most cases!
Re the FET7110-C, I don't see any particular use case for it. The JH7110 is just RV64 with RVV0.7.1. Forlinx did actually reach out at the start of month offering some content, but I turned them down.
@@PlatimaTinkers Maybe you are thinking of the TH1520 with the weird RVV implementation? JH7110 so far as I'm aware has none. It would be nice to have multiple sources of JH7110 CM4 modules in the spotlight. Better to see some more options and competition even for the same CPU.
@@niceride OH yeah right you are - the JH7110 is the U74 cores. My bad. But yeah so I've got the StarFive VisionFive 2 and both Milk-V Mars and Mars CM, so got plenty of access to JH7110 to tinker with if need be!
If they bring out something new or bespoke I'd be all over it, as they didn't really seem to limit my selection, just said "pick what you want to review and we'll send it to you" 😅
Next time you use OBS, I recommend setting up just a single window capture rather than a screen capture. That way, scene transitions can be gentler, and you don't have to worry about accidentally revealing any notifications.
Valid, and I considered that, but thought I might want either a web browser or a terminal, and then it's also not full screen size unless I actually fullscreen it, then switching it is harder. Else I could scale it up. Stopped caring too much this time haha.
And notifications aren't a concern for this one; the computer I put together is purely for these videos, doesn't have any email or messaging apps etc :)
@@PlatimaTinkers ah, nice. I don't have the setup for capturing video from a dedicated machine and have had to rerecord things in the past. I went somewhat deep with my OBS exploration when I first started using it and set up screens with looping gifs and whatnot for video calls and streaming some Python dev work a few times. Back when I still had to do daily stand-up meetings, I would set my camera to a loop of some baby squirrels I raised climbing over my arms until I was asked to turn it off for being distracting lol. Fun stuff.
@@k98killer yeah I only put this 'bitsa' together in the last two weeks so I didn't have to do a fresh setup each time.
OBS definitely has a lot of good functionality I need to explore more! This was the first time I had used scene transitions.
Hahahah that video loop sounds awesome. Reminds me of one I saw where a guy made a video of himself entering the room, looking at the computer, then backing out, and used that as his virtual background for Teams/Zoom meetings, and it took ages for someone to clock that there were two of him 😂
@@PlatimaTinkers lol that is brilliant
@@k98killer x.com/itsdancrowd/status/1245939558899802113?lang=en
The switch on the Waveshare development board is the BOOT switch. When the switch is set to "on," the development board is in flashing mode, and it cannot boot into the system.
Yeah figured that out after haha. I was clearly having a slow day!
I've looked at getting one of these for a while but I just think they are priced a bit too high :/
Yeah I considered that, but then if you look at the cost of 6x IO boards, 6x cases, 6x USB-C PSUs, and an 8-port switch, it's actually not bad!
Dingus.
Yuuuuup. Had you not heard that word before? (did I even use it?) or you calling me that? Both valid 😂
😂First heard that on the EEVblog. Quite used to it now.
@@turanamo Hahah yeah I am not sure if it's aussie, but it gets thrown around a fair bit in my circles.
Sometimes at me, sometimes from me.
You may need 6 pcs Raspberry Pi CM4 module for Super6C....
"Raspberry Pi CM4"
@@PlatimaTinkers
Oh, okay, I really enjoy watching your videos; they've taught me a lot of new things. They're fantastic, and I'm looking forward to your next video.
@@52Pi_Maker_Education Thanks mate, glad to hear! This weeks one is going to be about a shitty $50 AliExpress camera that I'm sending back 😂
Note I just patched up my reply that was in haste above, and I also noted that the Radxa CM5 lists the DeskPi Super6C as a supported IO board!
i really want to test this but i didnt have the money
Yeah the money bit hurts!
Anything in particular you'd like to test? I might be able to try some things for you!
Isn't it easier and cheaper to get a proper AMD/intel machine and just run ESXI on it for whatever compute you need to do?
Probably, though then you're virtualizing which *can* be different than true hardware. Also you can't virtualize every hardware you might want to test. And third virtualizing may or may not be slower and requires a lot of RAM. Also behavior of cluster software may be different with real hardware with real network latency.
I think VMs are often times a good solution and you're probably right in this case, but it's not always the better option. Also real hardware is just more fun
@@thevayudev also power usage would be horrific on AMD/intel in comparison.
1) Cheaper, maybe. 2) Easier, maybe. 3) ESXi? Eww (and that's coming from someone who is VCP certified haha)
The power consumption would be higher, you'd not have GPIOs to test/work with, you cannot test DTCs and boot optimisation, and in any virtualisation there are inherent inaccuracies with how some code may run.
ESX(i) also cannot do ARM/RISC-V, which is more my use case here!
I do use Docker instances for testing OSes etc, but the final tests usually require hardware 😊
@@thevayudev Fun is a good point! And yeah, slightly different behavior then too!
@@Crftbt Oh yes so very true. Technically that board was running 16 or so cores with 16GB RAM and presumably drawing less than 45W.
I'm definitely going to have to do a 'part 2' and test that!
I am an idiot. I also like the content. *subscribe*
We can be idiots together 😂
I don't know.
No one ever does friend. That's the story of life
im pretty sure im asleep lol
Very very fair! 😅
@@PlatimaTinkers i like the idea of populating one of these cluster boards with a bunch of different architectures. versus just 6 of the same thing.
@@ChrisJackson-js8rd Yeah not sure if I left that in or cut it out, but that's a bit of my goal. Eg something very powerful on CM1#1 for IO etc, and otherwise make sure I have a good mix of arm32, armhf, arm64, RV32, RV64 and x86-64, or something to that effect!