News about the ROCK 5B Plus! ;)

#1

Hello friends!

Thanks to generous Radxa, they sent me an ROCK 5B Plus for testing :grinning:
I would like to give a quick preview of the board with you.

IMG_20240426_132417
IMG_20240426_132417.jpg3680×5408 3.6 MB
IMG_20240426_132428
IMG_20240426_132428.jpg3680×5408 2.95 MB
IMG_20240426_132504
IMG_20240426_132504.jpg1840×2704 1.5 MB
IMG_20240426_132525
IMG_20240426_132525.jpg3680×5408 3.29 MB
IMG_20240426_132622
IMG_20240426_132622.jpg5408×3680 3.02 MB
IMG_20240426_132636
IMG_20240426_132636.jpg5408×3680 3.14 MB

The board on my hand is version X1.1 and the details are subjected to changes.

For those I know,
SoC would be our old friend RK3588.
For RAM, the board on my hand is 2 x K3LKBKB0BM-MGCP(32 Gb)
LPDDR5,which give a total of 8GB.
This time we have wifi on board, which is Azurewave AW-XM548NF, which support WIFI 6 AX.
The USB Type-C power input port this time would trigger PD-12V, which should eliminate a lots of power related problems. :grin:
It also got 2 HDMI out and 1 HDMI in ports, at standrad size.
There is a raised 2242 B-key slot on front with PCIE 2.1 x1 lane and TWO m.2 2280 slots at the back! (PCIE 3.0 x2 + x2).
It would be perfect for NAS I bet :slight_smile:
In additions, I see two CSI, one DSI, and a SIM Card slot at the back for modem.
For eMMC, it has got on-board 8GB from FORESEE.

Update: RAM test is added

To see if there is material improvement in RAM speed, sysbench ram test was performed on Rock5B and Rock5B Plus with results as follow:

Rock5B Plus RAM Write: 11359.78 MiB/sec

rock@rock-5b-plus:~$ sysbench memory --memory-block-size=1G --memory-total-size=                                                                                                                                                             
20G --memory-oper=write run
sysbench 1.0.20 (using system LuaJIT 2.1.0-beta3)

Running the test with following options:
Number of threads: 1
Initializing random number generator from current time


Running memory speed test with the following options:
block size: 1048576KiB
total size: 20480MiB
operation: write
scope: global

Initializing worker threads...

Threads started!

Total operations: 20 (   11.09 per second)

20480.00 MiB transferred ( 11359.78 MiB/sec )


General statistics:
total time:                          1.8012s
total number of events:              20

Latency (ms):
     min:                                   88.65
     avg:                                   90.04
     max:                                  115.59
     95th percentile:                       89.16
     sum:                                 1800.89

Threads fairness:
events (avg/stddev):           20.0000/0.00
execution time (avg/stddev):   1.8009/0.00

Rock5B RAM Write: 11807.97 MiB/sec

rock@rock-5b:~$ sysbench memory --memory-block-size=1G --memory-total-size=20G --memory- 
oper=write run
sysbench 1.0.20 (using system LuaJIT 2.1.0-beta3)

Running the test with following options:
Number of threads: 1
Initializing random number generator from current time


Running memory speed test with the following options:
block size: 1048576KiB
total size: 20480MiB
operation: write
scope: global

Initializing worker threads...

Threads started!

Total operations: 20 (   11.53 per second)

20480.00 MiB transferred (11807.97 MiB/sec)


General statistics:
total time:                          1.7328s
total number of events:              20

Latency (ms):
     min:                                   85.44
     avg:                                   86.62
     max:                                  106.14
     95th percentile:                       86.00
     sum:                                 1732.49

Threads fairness:
events (avg/stddev):           20.0000/0.00
execution time (avg/stddev):   1.7325/0.00

Rock5B Plus RAM Read: 15481.64 MiB/sec

rock@rock-5b-plus:~$ sysbench memory --memory-block-size=1G --memory-total-size=20G --memory-oper=read run
sysbench 1.0.20 (using system LuaJIT 2.1.0-beta3)

Running the test with following options:
Number of threads: 1
Initializing random number generator from current time


Running memory speed test with the following options:
  block size: 1048576KiB
  total size: 20480MiB
  operation: read
  scope: global

Initializing worker threads...

Threads started!

Total operations: 20 (   15.12 per second)

20480.00 MiB transferred (15481.64 MiB/sec)


General statistics:
    total time:                          1.3212s
    total number of events:              20

Latency (ms):
         min:                                   63.50
         avg:                                   66.04
         max:                                  112.05
         95th percentile:                       63.32
         sum:                                 1320.79

Threads fairness:
    events (avg/stddev):           20.0000/0.00
    execution time (avg/stddev):   1.3208/0.00

Rock5B RAM Read: 10361.02 MiB/sec
rock@rock-5b:~$ sysbench memory --memory-block-size=1G --memory-total-size=20G --memory-oper=read run
sysbench 1.0.20 (using system LuaJIT 2.1.0-beta3)

Running the test with following options:
Number of threads: 1
Initializing random number generator from current time


Running memory speed test with the following options:
  block size: 1048576KiB
  total size: 20480MiB
  operation: read
  scope: global

Initializing worker threads...

Threads started!

Total operations: 20 (   10.12 per second)

20480.00 MiB transferred (10361.02 MiB/sec)


General statistics:
    total time:                          1.9674s
    total number of events:              20

Latency (ms):
         min:                                   97.41
         avg:                                   98.34
         max:                                  112.99
         95th percentile:                       97.55
         sum:                                 1966.74

Threads fairness:
    events (avg/stddev):           20.0000/0.00
    execution time (avg/stddev):   1.9667/0.00

The result look divergence. :face_with_raised_eyebrow:

5 Likes
#2

Does that chip handle Bluetooth too?

#3

Referring to the datasheet from azurewave, the wifi chip bundled with Bluetooth

1 Like
#4

Photos updated! :grin:

#5

Is the eMMC no longer swappable this time? Maybe Radxa decided to pre-install ROOBI OS in it.

#6

From the board I’ve got, yes eMMC is not swappable.
For the ROOBI OS, I think that’s not the case.
ROOBI OS support x64 CPU, while for the ROCK series, they are all arm64 SoC.
They came with RadxaOS and rsetup tools under Debian distribution.

RadxaOS | Radxa Docs

Such eMMC should be flashable I think.

#7

That document is not finished yet. In Rock 5 ITX, the onboard 8GB eMMC will be pre-installed with ROOBI OS. It reminds me of this.

#8

em… but on ROOBI OS page it just stated its support for PuER N1 and SLiM X2L…
on my board it came with RadxaOS either…

So maybe let’s wait for the board release :slight_smile:

#9

image
image.jpg1920×815 140 KB

#10

I can see there is still this square component below the nvme. Is it still blocking ‘thick’ nvmes from beeing mounted correctly? Could you take a picture from the side to see how much space is between nvme and this component?

#11

Now I only need to wait for it to be available, so I will be able to use NC550SFP (or maybe ConnectX-4) + PCIe switch for multiple SATA boards :slight_smile:

#12

Khadas has their oowoow, others also experiment with such software to allow users easily flash several things and play with them. This is business and marketing decision rather than tech. As we can see here on forum there are problems over and over again about wrong thing tested by users. On the other hand I bet that most of You don’t use eMMC socket for anything (I’m using it as much better SD card).

I expected that 5B+ will get some speed advantage over 5B as key feature on this board. As @tkaiser noted (and many others confirmed) right now RK3588 with DDR4 and DDR5 have same performance. This may change in future, maybe we will also see DDR5 cheaper, especially with big capacities. For now it’s nothing to take into account when choosing between 5B and 5B+.

For double m.2 slot, we already could have that in 5B thanks to @seni, who shared design of his board. It’s small, cheap, without active element. This should be official radxa accessory allowing users to decide if they need 4x or 2x+2x.

Built in wifi6 will be great for many uses and this will perfectly work with roobi. It’s rather good move, not for me because I’d prefer eMMC socket :slight_smile:

#13

I will miss the EMMC socket too. Should be 32 GB to be at least useful (which would still be not much). 256GB EMMC via module + M.2 Sata ports + M.2 Nic would be really nice to do

2 Likes
#14

Re: For double m.2 slot, we already could have that in 5B thanks to @seni, who shared design of his board.

Do you have a link with information to dual m.2 board? I searched the forum and didn’t find any info.

#15

I am sure your search is not deep enough :wink:
I’ve included the information from @Semi in my OMV post :slight_smile:
[NAS Image] - OMV7 image for Rock 5B with ZFS, PCIe Bifurcation and 10G NIC - ROCK 5 Series - Radxa Community

#16

The link for atrium still works,
the only thing that You need to take care about is to choose 0,8 thick PCB. Board does not contain any active elements, it’s cheap to make and works great. Also You can choose if You need 4x pcie lanes or 2x2 so in this case it’s just better than ROCK 5B+ with bit thicker design.

I still think that this should be official radxa accessory.

image
image.jpg1194×695 554 KB

For now ROCK 5B vs 5B+

  • usb-c with video so it’s possible to connect and power board via one cable vs separate video
  • m.2 e slot for wifi card of choice (or ssd, 2.5G ethernet, native sata) vs build in wifi6
  • eMMC slot vs built in Roobi OS on soldered eMMC (not accessible)
  • no sim card slot (need m.2 card with that) vs slot on board
  • 4x lanes on one m.2 pcie or 2x2 with @Semi i board vs only this option on plus board
  • DDR4 vs DDR5 with same performance

Probably many users will be happy about most of those on 5B+, mainly because of wifi6 already on board, some will love roobi os so they can quickly start and flash something. Even better version could have 64G eMMC with option to flash beyond 16GB so it’s possible to get board running without anything else and with faster than any sd card built in memory. I expected promised speed improvement with ddr5 and so far it’s not the case, so I have no reason to upgrade even there when two m.2 are in use :slight_smile:

3 Likes
#17

It turns out that searching “seni m.2” (as spelled by dominik) does not yield the same results as “semi m.2”.

I found the altium link. Thanks to you and domink for the info :slight_smile:

1 Like
#18

Is it final, that Rock5b+ only supports 2x2 and not 1x4 with disabled port? Thats indeed a good argument. Also: Is it 100 % sure that it’s not optionally possible to power via “Data-usb” port?

#19

I see the sockets for external antennas. Very nice. MHF4?

#20

i am happy with the add-on pcie bifurcation board, until i got both of the slots populated with nvme SSD :frowning:
The power supply from both of the slots on this board are coming from the single m2 slot on Rock 5B. When you try to write them simutaneoously, in my case, the disks drop out.

Therefore, Rock5B+ would be good to explore with some more power hungry application.
As the add-on board solution, I think we need to think a way to provide additional power to mitigate this problem…