lanefu changed the topic of #armbian-rockchip to: Armbian - Linux for ARM development boards | Rockchip SoC | www.armbian.com | This channel is relayed to the equivalent Discord channel | this channel is logged
DC-IRC has quit [Remote host closed the connection]
<DC-IRC>
<nicod_sbc> All my benchmarks show the same.
<DC-IRC>
<tenkawa42> The Rock5's are very "inconsistent" pieces of hardware..
<DC-IRC>
<tenkawa42> Got one that USB is being a royal nuisance on
<DC-IRC>
<nicod_sbc> I'm glad I'm not the only one who thinks Rock5B is the worst RK3588 that I have. Cost me an arm and a leg too.
<DC-IRC>
<efectn> Why
<DC-IRC>
<nicod_sbc> Got multiple design flaws and it's not as stable as my others.
<DC-IRC>
<efectn> Most rk3588 boards have annoying design flaws as i've seen
<DC-IRC>
<efectn> Especially about power
<DC-IRC>
<nicod_sbc> The possitioning of the heatsink mounting holes is just awful.
<DC-IRC>
<tenkawa42> They are horribly inconsistent.. poorly designed. and poorly manufactured (this is the board I've had parts fall off of one)
<DC-IRC>
<tenkawa42> The Khadas 3588s Edge2 is the only one i like that I have
<DC-IRC>
<nicod_sbc> Indeed, none are perfect, but the Rock5B is the least perfect of them all. Still a nice board, but nothing like Mekotronics boards that are stable as hell and just work.
<DC-IRC>
<tenkawa42> These Radxa ones are horrible
<DC-IRC>
<infinity_q> I like my opi5s
<DC-IRC>
<tenkawa42> They "try" on the software side... its just really hardware design issues I think
<DC-IRC>
<nicod_sbc> I reviewed the OPi5 and it was good. I just had a cap fall off but it still works.
<DC-IRC>
<efectn> opi5 also have some problems but it's the cheapest one, so it's normal to me
<DC-IRC>
<infinity_q> gosh, you must beat yours on stuff 😅
<DC-IRC>
<infinity_q> I have 2 regulars and 1 plus. haven't had any problems.
<DC-IRC>
<nicod_sbc> I'm about sure it was from shipping. The first time I touched it it just fell off. So either very bad solder or bad trip to belgium.
<DC-IRC>
<efectn> For plus, the placement is a bit bad. For example emmc port
<DC-IRC>
<tenkawa42> I had a microsd hinge just fall off of one sitting on the desk
<DC-IRC>
<efectn> Also 5v4a is really bad
<DC-IRC>
<infinity_q> It could've been. I'm in US and used Amazon, so
<DC-IRC>
<infinity_q> well, I can't say I've used the emmc at all
<DC-IRC>
<tenkawa42> I personally wish more of these would support the small form factor nvme
<DC-IRC>
<infinity_q> I guess it depends on the use case
<DC-IRC>
<tenkawa42> even the ones that do don't put the mounting holes to bolt it down
<DC-IRC>
<infinity_q> fast+much storage = bigger
<DC-IRC>
<tenkawa42> my biggest nvme drive = small form factor
<DC-IRC>
<tenkawa42> its bigger than all the rest of my other nvme drives
<DC-IRC>
<infinity_q> biggest small one I have is 256
<DC-IRC>
<tenkawa42> I have a 512 and a 1 tb
<DC-IRC>
<infinity_q> biggest large I have is 2tb
<DC-IRC>
<efectn> There are some extenders like this one
<DC-IRC>
<nicod_sbc> I now have 2 x Edge 2 devices. Khadas Edge 2 and Mixtile Edge 2. KEdge2 indeed is nice for that reason. But I'm not a fan of the plastic case they designed for it. I want a metal case that keeps it cool.
<DC-IRC>
<mariob> mixtile edge? or blade
<DC-IRC>
<tenkawa42> @nicod_sbc I just use the metal case & fan
<DC-IRC>
<tenkawa42> (heatsink & fan)
<DC-IRC>
<nicod_sbc> Yeah, but for taking it with me on my trips that's not an option. I now have the KVIM4 with awesome metal case. But only Android that's usable for now.
<DC-IRC>
<Tonymac32> it's literally an rk3328 with some extras unless I've had a mini stroke
<DC-IRC>
<tenkawa42> so.. its a 3528 video box
<DC-IRC>
<tenkawa42> so.. its a 3528 tv box
<DC-IRC>
<Tonymac32> I mean, the 3328 is a tv box SoC too
<DC-IRC>
<tenkawa42> from the way I read those specs
<DC-IRC>
<Tonymac32> like 90% of the parts we get
<DC-IRC>
<Tonymac32> the s912 and s922 are unnecessarily buff TV box processors 😄
<DC-IRC>
<tenkawa42> Yeah sounds like they want to tout a "8k" label
<DC-IRC>
<Tonymac32> yeah probably only vdec/enc is different
<DC-IRC>
<tenkawa42> Notice none of these have audio dsp worth anything?
<DC-IRC>
<Tonymac32> tkaiser linked this in a cnx comments section
<DC-IRC>
<Tonymac32> 2 gmacs, opp to 2 GHz w/lower power than 3328 (as tkaiser commented)
<DC-IRC>
<Tonymac32> pcie2
<DC-IRC>
<Tonymac32> gpu opps to 800 MHz 🤔
<DC-IRC>
<Tonymac32> rkvdek/venc V2
<DC-IRC>
<Tonymac32> 4 CAN bus
<DC-IRC>
<Tonymac32> so infotainment unit processor 😄
<DC-IRC>
<efectn> Did rk3328 have pcie2
<DC-IRC>
<Tonymac32> nope
<DC-IRC>
<Tonymac32> so it has new stuff in there and some better process, as tkaiser mentioned the 2 GHz opp shows 1.1V while RK3328 needs 1.4 or something to hit top clocks
<DC-IRC>
<efectn> Sounds not bad
<DC-IRC>
<efectn> I hope fake tvbox manufecturers will add m2 m slot 😂
<DC-IRC>
<Tonymac32> 😄
<DC-IRC>
<Tonymac32> The PCIe 3 patches got merged to mainline, the PCIe 2 were spun up for v4, hoping they get past all the confusion this time
<DC-IRC>
<Tonymac32> for rk3588
<DC-IRC>
<Tonymac32> well, they're in for-next, at least 🙂
<DC-IRC>
<efectn> I hope they will also send series for usbdp before 6.6-rc1 merge window closes
<DC-IRC>
<Tonymac32> I submitted the NanoPC-T6, maybe I'll get lucky 😄
<DC-IRC>
<Tonymac32> even with 1000 pages of documentation that workflow is not actually very straightforward 😄 😄 😄
<DC-IRC>
<Tonymac32> if you git format-patch and clean them up, then git send-email completely ignores --cover-letter
<DC-IRC>
<Tonymac32> but if you make the cover letter via git format-patch, then you can't automate the to and cc with the maintainers script because the coverletter isn't a patch 😄 😄 😄
<DC-IRC>
<tenkawa42> Am I completely misremembering (entirely possible) or should I be able to get sensor voltage numbers off the RK3588 SoCs?
<DC-IRC>
<tenkawa42> If so.. anyone remember which sensor it uses in the Rock5 for example?
<DC-IRC>
<tenkawa42> nevermind... just missed it in the 15 lines of output
<DC-IRC>
<tenkawa42> (I added nvme and it added a lot of new lines)
<DC-IRC>
<tenkawa42> Now to figure out why it shows up on one unit but not the other with identical setups
<prahal>
I have instable ram with blob less u-boot on helios64 rk3399 with lpddr4. Could one with an rk3398 with lpddr4 and blobless u-boot check that `for i in $(seq 1 100);do python3 -c "import pkg_resources" || break;done` does not error out with invalid free or segmentation fault? (Here python 3.11).
<prahal>
I can workaround this bug with building BOOT_SCENARIO=tpl-blob-atf-mainline see https://github.com/armbian/build/issues/4761 . But I really do not see how this could be helios64 specific. Maybe users with other rk3399 lpddr4 boards never upgraded their u-boot.to the armbian blobless one.