BBB vs PocketBeagle 2

[Juddge]

looking at Specs, the pocketbeagle 2 looks to be faster than the BBB
would this be a better alternative to the BBB ? or does the BBB still have more fucntions that the pocketbeagle 2 cannot handle ?

looking mainly at performance ?
web access to the BBB can be slow when the CPU is at 100%
but a PI4 or 5 is fast , so was wondering if changing to a faster BB Pocketbeagle 2 would be better than the BB Black for performance on a cape

 

[Skymaster]

What are you using it for? Just as a controller with cape, or as a full show player?

 

[Juddge]

i use xschedueler to run my shows all my equipment are just to run capes only

 

[Skymaster]

By going the PB2, you'll be limiting yourself to the capes that support it - which as far as I'm aware is only the Kulp PocketBeagle range.

From a purely pixel driving perspective, there's nothing wrong with the BBB; it's rare that you ever need to go and play in the webui. And it should only be at 100% usage whilst using the WebUI - most of the pixel functions shouldnt have any issues running at full whack due to it using the PRU Coprocessors for doing the actual pixel driving work.

 

[dkulp]

The PB2 is definitely an 'upgrade" over the BeagleBone Black/PB1 in many ways. The CPU cores are roughly equivalent to the Pi 3B+ so processing speed is at about that level. The PRU cores are up from 200Mhz to 250Mhz, but that doesn't really matter when outputting pixels at 800Khz (ws2811 protocol). It does help somewhat with driving panels. It also has a MUCH MUCH faster SD card (75MB/sec compared to 20MB/s on the BBB, 40MB/s on a Pi4). For the most part, the PB2 will definitely boot faster than the BBB. It boots about the same as the Pi4's. With multiple cores, it can definitely keep the UI fully responsive even while processing a ton of pixels (more on that in a second). Unfortunately , hitting the "Upgrade FPP" button is NOT much faster on the PB2 compared to the BBB so if you think that's going to be a huge help, think again. The two main reasons are:

1) compiling optimized for 64bit is a bit slower due to all the additional instructions (NEON, etc..) and size (64bit binaries are much larger)
2) The PB2 only has 512M of RAM so we cannot use the multiple cores to build in parallel. It ends up swapping heavily and slowing down to slower than a single threaded make

The other downside of the PB2 is the SD card slot location. It's UNDER the PB2 so hard to swap SD cards. That said, it's still a ton better than completely under the controller and everything like on my K16/K32's. Another downside compared to the BBB is the lack of ethernet, but if you use WIFI, that's not really a huge deal.

That all said, for my own designs, over the next two years, I expect to move all of my designs over to the PB2. It will take time as I have a bunch of different boards, but my goal is to completely move to the PB2. The first to move over is the PocketScroller and it will likely replace the existing PocketScroller and OctoScrollers. The new PB2 allows handling over 64 P5 panels (the BBB would max the CPU out at around 24). It also has the "E" line so can run P2.5 panels. By leveraging the additional CPU's to process the data and using BOTH 250Mhz PRU's, we can handle much larger panel counts compared to the BBB. It's also using the direct connect GPIO pins which allow much finer control so we'll be able to support some of the new "PWM" based panels. I have the Your Pixel Store outdoor flexible P4 panels (FM6363 PWM chip) working with them right now and may be able to add more if I get more sent to me. Will it be able to replace the ColorLight in all cases? No. Will it work great for a majority of our cases? Likely yes with SIGNIFICANTLY less headache to setup, particularly for Mac/Linux users that cannot run LEDVision. The second board I'll introduce will be the K32-Max which will be a PB2 based version of the K32, but with eFuses. Again, I plan to move over to PB2's, but it will take time. Re-designing a board is a long process and will take time to get all of my various BBB designs moved over. For the most part, other than for panels, the PB2 won't drive any more pixels than the BBB. The ws281x protocol pretty much dictates the number of pixels per port and the PB2 and BBB controllers can both drive up to 64 ports without any issues/limits.

I did mention the lack of ethernet on the PB2. For my new boards, it will use the USB capabilities of the PB2 to add a gigabit ethernet. There will be a powered USB hub (so no soundblaster power issues) on the board with one port wired to the ethernet and the other ports to USB ports. The PocketScroller v3 will have three USB ports in addition to the Ethernet. That said, it will connect at gigabit speed, but it's only connected via USB2 (all the PB2 has) so will max out at around 300Mbit/sec. Pretty much the same as the Raspberry Pi 3B+.

Again, it will take time to migrate an entire product line over. It won't happen fully this year. In addition to the work required to migrate, the tariff situation with China also throws a bit of uncertainty in the equation as well.

Anyway, my EXISTING PocketBeagle1 based designs do work with the PB2. They did an EXCELLENT job of designing the PB2 with pin compatibility in mind. The main issue is that I used to tuck the USB port under the PB1, but that's now where the SD card is. It fits, but I recommend a simple USB extender pcb thing to bring the SD card out above the USB port. (and I ship them that way now).

Dan

 

[Skymaster]

Again, it will take time to migrate an entire product line over. It won't happen fully this year. In addition to the work required to migrate, the tariff situation with China also throws a bit of uncertainty in the equation as well.

You need to move manufacturing to Australia ;-)

For my new boards, it will use the USB capabilities of the PB2 to add a gigabit ethernet

This is a great idea. My biggest gripe with the Pockets is the lack of wired connectivity.

 

[dkulp]

You need to move manufacturing to Australia ;-)

I've spent a lot of time the last 3 weeks sending requests out to various manufacturers all over the globe. I'm running low on one of the boards so I wanted to order another batch. Since it's an existing design, a direct comparison is easy. I sent RFQ's to manufactures in India, Singapore, Vietnam, Canada, Mexico, etc... The BEST I got back was still 40% higher than what my China manufacturer quoted. When tariffs are 145%, that's acceptable, but the announcement this morning of scaling back to 30% means I'm back to China for the next 90 days. I spent a lot of time emailing back and forth with manufacturers. That's my biggest issue. Moving supply chains takes a lot of time and effort that is then not spent doing actual interesting/fun work.