Any way to save USB type-c and power up the board through something else (battery pins or dc in connector)?
Introduce ROCK 5B - ARM Desktop level SBC
I am hoping that the release of the Rock5b is imminent. Very exciting times for single board computer lovers! Khadas released their latest version of the Vim. That board is cool I have the Vim3 pro and it’s been great. That said , the Rock5b is the board I ordered and the one I want
1 Like
It’s nice to see some khadas upgrade, but I’m really disappointed with its price and some of specs, they still use old cores what makes it almost two times slower than rk3588:
https://browser.geekbench.com/v5/cpu/compare/13973601?baseline=14170348
and NVME is still 1x lane (vs 4x on Rock 5 or Rock 4c+). Sure - nice to see upgrade from 4GB to 8GB, but yet again we will get 16GB on same price level with Rock5 (and we may see 32GB on this chip like with mixtile blade).
I have two khadas VIM3 pro and they are small and cute. It’s perfectly sized board with best case and some unique features. I would buy new VIM at half of it’s price, but not much more. It’s far behind expected rock5
1 Like
I am really interested in what price Radxa set for the Radxa-Zero2 as its basically a Vim3 and yeah $200 seems very steep for what is essentially just 2x extra GPU cores in the Vim4.
Khadas do some really nice gear but you can never say its cheap…
The RK3588 is faster but you normally compare like for like to get meaningful results.
Your link has the RK3588 on a different version of test software, twice the memory and 64 bit Android of a different version. You call that like for like ?
Well, the people here are people who aren’t interested in VIM4 or anything.
1 Like
Back in 2018, i thought it was a good idea to buy the VIM2 MAX because the Pi 3B did not have hardware acceleration for x265. Having 8 cores, 3 GB RAM with 64 GB eMMC i would be having no worries for at least 10 years…ooh boy what was i wrong. I was running CoreElec and suddenly last year or in 2020 it was announced that CoreElec for VIM2 MAX would not receive big updates anymore because of Amlogic no longer supporting their S912 SOC. I was dumbfounded…
Khadas, has done a very good job with support etc, but if a company stops supporting their product well what can you say about it. Not much. I hope Rockchip will not do something similiar in just 1-2 years.
It’s all we have now and both boards are under active development so both can improve those results. I just don’t expect there is anything wrong with those results because new amlogic just added two old A53 cores when RK3588 upgraded all cores to A55/A76 and they are significantly faster and power efficient. I don’t think there is anything wrong about those results, they will not change much. This is one of two disappointment about VIM4 - it’s like change from 6GB RAM to 8GB RAM compared to DDR3 vs DDR4 
No no no. I own and use now two VIM3 Pro boards and when they were released it was something really exciting. VIM4 is way to expensive with slightly modified specs and double price of VIM3 Pro. In 2022 I would not buy any board with 1 lane NVMe when boards like Rock 4C+ have 4x and are available for $59.
1 Like
Amlogic never supported linux as they don’t do opensource all there products as far as I know are behind a licence paywall.
It was baylibre who provided a large amount of the support https://baylibre.com/improved-amlogic-support-mainline-linux/
https://linux-meson.com/hardware.html
Amlogic didn’t stop supporting as they never did its just the baylibre crew tried valiantly but never managed to reverse engineer hardware decoding and think that is just about the only thing that isn’t supported in opensource.
That gets you stuck with binary blobs on an old kernel tree that may of come to an end.
You mean A55/A76?
yes, sorry for typo, I’ll edit if that is possible
And here’s the BayLibre employee having done most of the work for mainlining Amlogic in the past talking about the chances for T7 getting mainline support anytime soon or at all: https://www.cnx-software.com/2022/05/07/khadas-vim4-review-part-2-android-11-preview-and-benchmarks/#comment-592259 (check the links to @chewitt posts in Khadas forum there as well).
While A311D (VIM3) and A311D2 (VIM4) sounds familiar they’re from entirely different Amlogic SoC families (G12B vs. T7) and the latter has a much better internal design, especially much better memory performance which directly translates into more CPU horsepower even today when Amlogic boot BLOBs limit the A73 cores to 2.2 GHz (usually this will be lifted over time, with G12B the A73 started at 1.8 GHz and are now at 2.4GHz).
Doesn’t change much wrt RK3588 which is much faster due to A76 cores being so much more modern and powerful compared to Amlogic’s A73 cores.
1 Like
I do agree. My vim3 pro was amazing when I first got it. I mean I was using a overclocked Rpi4 so it was definitely an upgrade. That 32 bit android image is obsolete in 2022. Absolutely nonsensical for khadas to release their upgraded vim4 but still have a 32 bit os! Insane
They can only release what Amlogic provides, and what they provide is a 32 bit Android for unknown reasons…
1 Like
Good to know. So i will boycott every SBC that ships with Amlogic SOC. I mean having a SBC, 90% if not more is or will be that you will be running something that is somehow associated with Linux. Be it Android or just a Linux distro (LibreElec, CoreElec, Ubuntu, RetroPi, Retroarch…etc)
Once you learn a bit more you will boycott also SoCs from Allwinner, Broadcom, MediaTek, Samsung and Rockchip. Since none of these vendors gives a sh*t about Linux and they all are only focused on Android e-waste products which is quite understandable since they want to sell their SoCs in the millions and not thousands.
Better get an SBC with a SoC from those vendors who care somewhat about Linux: Renesas, NXP or TI for example!
1 Like
Most probably for the same reason every other SoC vendor relies on a 32-bit ARM userland? Since it allows for way lower RAM requirements and doesn’t harm performance?
The ‘requirement’ for a 64-bit Android is mostly stupid consumers being trained into thinking ‘the more the better’ and naively thinking a 64-bit version would be twice as fast/better/whatever since ‘64’ is twice as big as ‘32’ while in reality providing no other ‘benefit’ than binaries being more fat and RAM requirements higher.
Thats sad to hear to be honest. From MediaTek i have experienced it myself in the past. Such a headache to get their chips to work with Linux. However was not aware of those other vendors.
These days if i see a product that i want to buy, i first try to find out which chip it is and search the web if there are any Linux drivers.
1 Like
The thing that I don’t understand with Amlogic, Rockchip etc… is the effort put in the harware side (probably the most expansive and most difficult aspect), but for the software/drivers, it looks like they don’t master their own code/technology. They provide a SDK one-time, with an old frankenstein kernel with no support.
Maybe they’re outsourcing this to a lowcost company, or it is a recycled example code from ARM when they pay the Arm licence …
I don’t see the point of not be more open and to not put their code in the linux mainstream, or atleast, provide a documentation. If they don’t know how to do it, they can pay Baylibre, Bootlin, Collabora, to do it, it would be beneficial for them to have a more secure et future-proof kernel 
For which reason? To make the few ‘I buy only as cheap as possible’ consumers on this planet happy who want to turn some Android e-waste product into a fully-fledged x86 PC replacement?
Have you ever tried to submit code to mainline Linux? Ever realized how long this takes? Ever thought about releasing a HW product at some point and telling your customers that ‘upstreaming drivers will be done in 2-4 years’?
Ever thought about frameworks and ABIs that do not even exist in Linux at the time you want to sell your nice and shiny new hardware so there’s no alternative than hiring some coders to hack proprietary drivers for your proprietary hardware so you are able to ship your cheap Android e-waste product with somehow working software at launch day. Nobody will care about software quality or openness since all that’s important in this Android e-waste world is ‘selling as cheap as possible’.
There is no market called ‘Linux on ARM’ (or at least this market is not part of this stupid SBC world here since those SoCs that are accompanied by a Linux SDK are for automotive/embedded markets and way too expensive to be used in the ‘Android e-waste’ and SBC world). Why would any SoC vendor right in his mind think only a second about Linux if all that’s important is shipping an Android product that needs drivers now and not in a few years?
And that pretty much is the reason why Linux on ARM is such a sh*t show. Since it’s completely irrelevant.