Jump to content

Stuart Naylor

Members
  • Posts

    83
  • Joined

  • Last visited

1 Follower

Recent Profile Visitors

The recent visitors block is disabled and is not being shown to other users.

  1. I still sort of think Radxa are positioning this board incorrectly as for me its the router companion to the RK3566 which is a great edge board. They seem to be positioning it as a series of multipurpose SBC's whilst ignoring its extremely strong selling points. It could be an extremely low cost native 3x SATA board that the PCIe3 could populate another 1Gbe m.2 if you wished. But as it is 3x sata could be perfect for a mini NAS on a single 1Gbe at extremely low cost. The other amazing thing is that the RK3568 has a 5Gbe MAC that would need a 5Gbe PHY chip to be added and then a M.2 sata on PCIe3x2 That at the cost would be amazing and passes a hurdle that many current SBC NAS are totally bottlenecked by ethernet bandwidth. The other thing that looks really good on paper is the G52 but in reality its a cut down MP1 that is quite different from standard the the MESA crew gave it a naming of G52L 'little' I have been looking and the Radxa Zero... really cool, that they might also do a S922x board could be cool also and the RockPi5 RK3588 a lot of cool. But when I look at the format, cost and offerings with the Rock3 range I am sort of confused as what I see that could be of huge value has been overlooked and layouts provide just another series of general purpose SBC where they don't really excel or have a unique selling point of anything... ? Each to their own but prob not likely to be interested in one.
  2. Yeah the new Radxa Zero has perked my curiosity as always loved the form factor of the Rpi0 but for me has been showing the SoCs age for some time. The performance and temps of this little thing is amazing and already tested with a small stick on heatsink 2.0Ghz seems completely stable. Is there a Radxa Zero section as have a few questions. I got the 512mb 1st because I think its more of a Pi3A+ competitor than Zero as it just blows it away and is nearer Pi4 on some benches. I can not get a non eMMC to boot in any way and wondering if we are all the same as read on the Radxa forum some have but confused and questioning how. With eMMC as long as you ``` sudo boot-g12.py rz-udisk-loader.bin wget https://dl.radxa.com/zero/images/loader/u-boot.bin sudo dd if=u-boot.bin of=/dev/sdx bs=512 seek=1 ``` Then you can boot from SD or flash an image to eMMC with no SD you can boot eMMC, put a SD in and it will boot from SD. All good but still stumped with my non eMMC version, but should try again with u-boot.bin.sd.bin again on a SD again, Toms also done a great 'Lite' image https://dl.radxa.com/users/jack/Armbian_21.08.0-trunk_Radxa-zero_focal_current_5.10.52_minimal.img but it lacks the AP6256 drivers of the Twister version. Twister is great but is loaded with so much stuff I don't need apart from how to get the AP6256 driver working. I did add the oibaf ppa for the mesa drivers as that does make a huge difference with glmark2 PS I did donate a small amount but on your subscriptions add a one off payment as a little by many always adds up. Another question is GPU overclocking where I did edit the opp table but on use unlike cpu seemed to stay the same. I was wondering if it could be pushed a little for an OC for a retro handheld and if leaning on cpu or gpu with a OC could help even though currently great for such a small low cost SBC.
  3. Just to say really impressed and wow you guys have done a lot with the documentation since I last looked. I was never a big fan and its been a while, but wow I have to say its really great and thanks for all your efforts. No script bloat and all seems really logical and much work most of been done.
  4. As anyone got any idea what the muxed stata/usb3 ports are? Is this basically a wired adapter as not much is said but on the Rock 3 model B 1 sata is on board and guess the other 2 are whatever you do to mux the usb 3.0 ports. Could be interesting for NAS as with a 5 port sata adapter then x8 native is quite unique for B. Still the pcie 3 is down to 1 lane (B) so overall bandwidth is same as the rockpi4 2x ver 2.1 but guess the bottleneck is the 2x 1gbe bridged maybe. Not sure if the A is the same just minus the 1st onboard and 2 lanes on the pcie.
  5. ff560000 Audio Codec PHY is the operational base address of the audio codec and the registers will be listed as an offset to that address.
  6. Hi guys, If I want to write to Address: Operational Base + offset (0x0470) ACODEC_DAC_ANA_CON12 with 55 what am I doing wrong as keeps telling me wrong address. I am root but most have the Operational base address wrong which thought that would be the codec base + 0x0479 store data there? ff560000 Audio Codec PHY fff80000 System SRAM(256K) ff800000 A35 DEBUG Tried several base addresses and just get an error as root. I thought it would be something like echo 0x60 0x40ff0168 > /sys/kernel/debug/vad/reg What I am trying to do is just see if there is register that will use the dac via I2s and even though its all cryptic thought I would just try a few registers but must have the base address wrong? PS is there anyway to put anchor points of a PDF as its in there but the right page would of been a bonus https://dl.radxa.com/rockpis/docs/hw/datasheets/Rockchip RK3308TRM V1.1 Part1-20180810.pdf But totally confused as the above is from the custom device rockchip repo https://github.com/rockchip-linux/device-custom/blob/master/rk3308/CaeDemo_VAD/RkLunch.sh The address doesn't make any sense to the ref manual as far as I can tell but maybe someone can help my stupidity
  7. Yeah I think they beefed up the audio port as it was a little flakey. I desoldered the broken sd connector and it will run but just mmc. Its just gathered dust since the raspberry 4 as the 4 lanes of pci-e are actually hard to populate. PS my fave Rockpi would be the RK3308 Pi-S if they actually did publish how to interface the x4(x1) VAD mics. Its an average general purpose SoC but for voice AI if there was more info it would be exceptional.
  8. Its been a while @Igor but yeah eventually with drivers I got great both way access speeds. I was messing around with drivers but there was a Realtek driver that seemed to do the trick but if there are Rockchip drivers and as above even better. I dropped the RockPi4 and somehow like a coin it landed edge up on the SD card and has since gathered dust, but also due to the Pi4 that mainly I play with now. I never got round to benching the RPi4 with those exceptionally good value 2.5Gb USB ethernet adapters, but the adapters from what I found are really good.
  9. There is what chwe said as the Armada 1500 is pretty mewh but I think it does have a zilog audio dsp and could make a good hack for a network speaker/mic image. I wonder if you could hack an image as embedded soundcards with dsp especially AEC are still far from cheap.
  10. Cool I will give that a try also.
  11. Dunno what I was doing with the driver!!? Finally got rid of the cdc-ncd driver and used the r8152 driver [rock@rockpi4 ~]$ sudo lshw -C network [sudo] password for rock: *-network:0 description: Ethernet interface physical id: 9 logical name: eth0 serial: da:19:c8:7a:6d:f4 capacity: 1Gbit/s capabilities: ethernet physical tp mii 10bt 10bt-fd 100bt 100bt-fd 1000bt 1000bt-fd autonegotiation configuration: autonegotiation=on broadcast=yes driver=st_gmac driverversion=Jan_2016 link=no multicast=yes port=MII *-network:1 description: Ethernet interface physical id: a logical name: eth1 serial: a0:ce:c8:c8:2b:31 capacity: 1Gbit/s capabilities: ethernet physical mii 10bt 10bt-fd 100bt 100bt-fd 1000bt-fd autonegotiation configuration: autonegotiation=on broadcast=yes driver=r8152 driverversion=v2.12.0 (2019/04/29) duplex=full ip=192.168.1.12 link=yes multicast=yes port=MII [rock@rockpi4 ~]$ iperf3 -c 192.168.1.9 -R Connecting to host 192.168.1.9, port 5201 Reverse mode, remote host 192.168.1.9 is sending [ 5] local 192.168.1.12 port 38496 connected to 192.168.1.9 port 5201 [ ID] Interval Transfer Bitrate [ 5] 0.00-1.00 sec 266 MBytes 2.23 Gbits/sec [ 5] 1.00-2.00 sec 239 MBytes 2.01 Gbits/sec [ 5] 2.00-3.00 sec 244 MBytes 2.05 Gbits/sec [ 5] 3.00-4.00 sec 237 MBytes 1.99 Gbits/sec [ 5] 4.00-5.00 sec 239 MBytes 2.01 Gbits/sec [ 5] 5.00-6.00 sec 250 MBytes 2.10 Gbits/sec [ 5] 6.00-7.00 sec 248 MBytes 2.08 Gbits/sec [ 5] 7.00-8.00 sec 244 MBytes 2.05 Gbits/sec [ 5] 8.00-9.00 sec 249 MBytes 2.09 Gbits/sec [ 5] 9.00-10.00 sec 248 MBytes 2.08 Gbits/sec - - - - - - - - - - - - - - - - - - - - - - - - - [ ID] Interval Transfer Bitrate [ 5] 0.00-10.00 sec 2.41 GBytes 2.07 Gbits/sec sender [ 5] 0.00-10.00 sec 2.41 GBytes 2.07 Gbits/sec receiver Bit of big core affinity [root@rockpi4 ~]# echo 5 > /proc/irq/221/smp_affinity_list [root@rockpi4 ~]# iperf3 -c 192.168.1.9 -R Connecting to host 192.168.1.9, port 5201 Reverse mode, remote host 192.168.1.9 is sending [ 5] local 192.168.1.12 port 38502 connected to 192.168.1.9 port 5201 [ ID] Interval Transfer Bitrate [ 5] 0.00-1.00 sec 274 MBytes 2.30 Gbits/sec [ 5] 1.00-2.00 sec 276 MBytes 2.32 Gbits/sec [ 5] 2.00-3.00 sec 276 MBytes 2.32 Gbits/sec [ 5] 3.00-4.00 sec 278 MBytes 2.33 Gbits/sec [ 5] 4.00-5.00 sec 277 MBytes 2.33 Gbits/sec [ 5] 5.00-6.00 sec 276 MBytes 2.31 Gbits/sec [ 5] 6.00-7.00 sec 277 MBytes 2.32 Gbits/sec [ 5] 7.00-8.00 sec 277 MBytes 2.32 Gbits/sec [ 5] 8.00-9.00 sec 277 MBytes 2.33 Gbits/sec [ 5] 9.00-10.00 sec 277 MBytes 2.32 Gbits/sec - - - - - - - - - - - - - - - - - - - - - - - - - [ ID] Interval Transfer Bitrate [ 5] 0.00-10.00 sec 2.70 GBytes 2.32 Gbits/sec sender [ 5] 0.00-10.00 sec 2.70 GBytes 2.32 Gbits/sec receiver iperf Done. I am using Manjaro 5.2 as latest kernel was on that and had it on an SD but yeah they will work on Arm. I need to go back to Armbian and work out why when the driver was suggested it failed and didn't work. I am not that great with custom driver install and know manjaro a little bit better and new the kernel was much newer. I will give it a go again but no kernel ring flood and near as damn it 2.5gb on the rockpi4 aint bad.
  12. I didn't get much further as my scrapyard spare x86 for some reason just dies when I try to install ubuntu. Blingdows installed ok and if I go from ubuntu server to windows client iperf both x86 then same as before. Think I will grab Manjaro Accepted connection from 192.168.1.9, port 41180 [ 5] local 192.168.1.12 port 5201 connected to 192.168.1.9 port 41182 [ ID] Interval Transfer Bitrate [ 5] 0.00-1.00 sec 251 MBytes 2.11 Gbits/sec [ 5] 1.00-2.00 sec 281 MBytes 2.35 Gbits/sec [ 5] 2.00-3.00 sec 281 MBytes 2.35 Gbits/sec [ 5] 3.00-4.00 sec 281 MBytes 2.35 Gbits/sec [ 5] 4.00-5.00 sec 281 MBytes 2.35 Gbits/sec [ 5] 5.00-6.00 sec 281 MBytes 2.35 Gbits/sec [ 5] 6.00-7.00 sec 281 MBytes 2.35 Gbits/sec [ 5] 7.00-8.00 sec 281 MBytes 2.35 Gbits/sec [ 5] 8.00-9.00 sec 281 MBytes 2.35 Gbits/sec [ 5] 9.00-10.00 sec 281 MBytes 2.35 Gbits/sec [ 5] 10.00-10.04 sec 11.7 MBytes 2.35 Gbits/sec - - - - - - - - - - - - - - - - - - - - - - - - - [ ID] Interval Transfer Bitrate [ 5] 0.00-10.04 sec 2.72 GBytes 2.33 Gbits/sec receiver ----------------------------------------------------------- Server listening on 5201 Manjaro Kernel 5.1 so yeah either a kernel or firmware thing. There was a lot of politics about 2.5 that was coming and at first then oh no your not then to actually come in. So adapter works fine but which kernel/firmware might be a bit of a minefield. The above is Manjaro as a server (Accepted connection from 192.168.1.9) Still don't get why not in reverse though with some kernel versions / drivers but the above is a perfect 95% efficiency that you should expect.
  13. Net rockpi4 client 100MBs sort of 1gbe speed? root@rockpi:~# dd if=/dev/zero bs=1024k count=1024 | nc -v 192.168.1.9 2222 Connection to 192.168.1.9 2222 port [tcp/*] succeeded! 1024+0 records in 1024+0 records out 1073741824 bytes (1.1 GB, 1.0 GiB) copied, 9.98256 s, 108 MB/s x86 client rockpi server stuart@stuart-pc:~$ dd if=/dev/zero bs=1024K count=1024 | nc -v 192.168.1.12 2222 Connection to 192.168.1.12 2222 port [tcp/*] succeeded! 1024+0 records in 1024+0 records out 1073741824 bytes (1.1 GB, 1.0 GiB) copied, 23.158 s, 46.4 MB/s Somehow auto negiotated to 1gbe me thinks and will have to check nmcli
  14. Nope I forgot to be honest its confusing and totally forgot as been trying many things and it just seems the rockchip code depending on slave or master. I will do now though as its not just this 2.5gb USB all are useless but strange thing is from memory my plugable 1gbe usb gives exactly same results, I mean exact same figures if that 300+ level is a max or same with g_ether OTG For £20 2.5gb which actually it isn't true they will run on existing as they are pretty picky with quality and length but reasonable cables rather than my usual battered spagetti 2.5 runs great. I am going to dual boot this windows PC with bionic and try x86 linux to linux which I am expecting will work fine. I am not sure if its rockchip / rk3399 or something but prob. I shall be at netcat whilst you try out your new eth adaptors I am sure you will find them OK the overly bright white led drives me nuts, they get warm compared to some hot usb 1gb I have had. Windows wise they work great so its not the adapters.
  15. Apols about that can not even work out on the forum how to get horizontal scrolls even though managed it with one somehow? So struggling with this, but its not the adapters as we have seen in windows they works near max both ways. Without affinity on the rockpi4 it doesn't cope with the load as well as windows but can as a client manage over 2gbs But why as a server it plummets to 400gbs with affinity I am not really sure. What I don't know is how other SBC react the same as thinking its something to do with the Rockchip code as the rk3328 has a similar performance profile and wonder if its the same with speed being asymmetrical depending on client or server? There seems to be zero load loads of entropy and no reason apart from maybe rockchip code. I am also mightily confused to the CPU arrangement as the default affinity is all cores apart from big, maybe that makes sense. I am going to play with affinity as maybe as a client the task has a higher affinity and runs on both big cores as client? I will just remember its a bit map and not core number
×
×
  • Create New...

Important Information

Terms of Use - Privacy Policy - Guidelines