When will we see a VIM4?

Are there any plans for a VIM4, and if so, what might be some of its new/unique features ?

Hi Silver and welcome.
I can’t speak for Khadas but there are very few viable candidate chipsets with a significant increase in capability as wellas linux mainline kernel support as solid as the Amlogic A311D as found in the Vim3 Pro.

The SoC I’d like to see in a SBC is the Mediatek 1300T with it’s Cortex A78 cores giving ~ 50% single threaded improvement over A311D, along with its Mali G77 MP9 which should be around twice the performance of our G52 MP5.

Unfortunately Mediatek is promoting it as android-only at the moment.

I see no current or upcoming alternative ARM sbc SOCs with Linux support with significantly better GPU than ours. The MT8195 includes too few execution units to put it into another performance class over our GPU.

To conclude, I don’t see that any system-on-chip is ready to ship to small manufacturers that offers 1) good free software support and 2) significant performance gains at 3) a still affordable price.

My advice is to buy a Vim3 Pro and do cool things with it :slight_smile:

Would Khadas consider the RK3588 for a VIM4 when it is released?

I believe that is the current roadmap :slight_smile:


Idk how true is this.


one thing i would like to see in the vim 4 would be a UFS card slot as it supports higher speeds and random access should be better as well. if the vim 4 is already set in stone, maybe ufs external card support can be added to another future product

hopefully they release the 16GB version as well.

Great news overall, but sad that the eMMC storage will be limited to 32GB.

I presume this is just the press release, the sooner before release I’m sure khadas team will have a vote pitch of sort, like they did with the VIM3 to see what is the majorities wish for things like ports and storage options. I too personally felt 16GB version might be nice to have for certain server applications and slightly modified board layout that wouldn’t be so cramped,

hope to see the best :slightly_smiling_face:


Most importantly I wonder what hashrate I can get out of it :rofl:


waiting for more info about the VIM4, release more info about the new beast! :joy:

I still can’t get how did cnx get this information without amlogic releasing anything themselve.
They also got the leaked block diagram.

emmc needs to be modular. otherwise 32gb is pretty insufficient. And if khadas doesn’t develop its own linux distribution, it has no chance against Raspberry Pi 5. Because even FHD videos do not open in current downloads.

what do you mean ? RPi5 might release somewhere near end of 2022
and if its as expected to come with a broadcom chip it will never be close to as efficient or powerful as the A311D2

consider using an SSD over eMMC, you will have much better endurance and drive speeds :slightly_smiling_face:

also what FHD videos are you facing issues with ?

1 Like

Media support in the upstream codebase is incomplete so if you need media capabilities you will need to use the vendor Linux 4.9 codebase; which becomes a 5.4 codebase with the A311D2, though the good news of a newer kernel is dulled by Amlogic simply forward-porting their old code. If you don’t need the media capabilities the A311D2 looks sufficiently similar to the current A311D(1) that support upstream should be simple to add and I expect users will be able to run modern u-boot and the latest or recent LTS kernels where software support is much cleaner, without too much effort.

Fixing media support in the upstream kernel needs some USD investment. I’d love for “the community” to solve this problem, but the gene pool of talent with the skills to do this is limited to a few professional developers who need to be compensated for the large amount of time and effort involved. The current (staging) kernel drivers need to be heavily refactored rather than “fixed” with a few patches, and the userspace v4l2m2m APIs are not published yet which complicates adding support to userspace apps like G-Streamer and FFMpeg. Fixing the Amlogic media situation is IMHO too much $$ for a single benefactor or even a couple of benefactors who agree to share the costs. In the past Google funded work but strategies and priorities changed and the work dried up. It’s a real shame, but it is what it is.

NB: IMHO The Pi Foundation are unlikely to release a new generation of chip so fast. We might see a bump in speed or a minor chip change elsewhere on the board in response to component shortages, but not an entirely new SoC. I’ll also point out that the success of RPi boards is all about software. Their hardware has always been a compromise to keep costs down, and there have always been a ton of better spec boards with faster and more powerful SoCs available in the market. However if you need to maintain and long-term support a product based on an SBC board; rather average hardware with A1+ software support always trumps A1+ hardware with rather average software; because your long-term cost is about support not the initial purchase of components.


with a 5.4 Kernel as BSP, does that mean panfrost would work out of the box ?

Panfrost might technically be present in an earlier iteration but it’s not stable until more recent kernels (perhaps 5.10 as a minimum) and since Amlogic uses mali_kbase in their BSP sources i’d have low expectations for that patching being done (or being sensible). Panfrost is also written around upstream kernel DRM drivers and the Amlogic BSP has it’s own completely separate DRM layer, so I’d say it’s a definite NO to OOB support.

NB: I’ve also been told that support for older chipsets like GXL/GXM was not forward-ported (which is normal behaviour for Amlogic) so manufacturers like Khadas who need to support a range of hardware generations will be forced to support multiple BSP kernels in their images. It’s never impossible to do, but it all adds to the background workload and complexity of support. It’s one area where the upstream kernel (where backwards compatbilitiy is important) makes life a lot easier for everyone.


also what FHD videos are you facing issues with ?

stock ubuntu versions have video playback issues. I tried all of them. Only android works stable in this sense. The thing is have you tried youtube 1080p videos? not playing etc.

consider using an SSD over eMMC, you will have much better endurance and drive speeds :slightly_smiling_face:

I do not understand what you mean. If what you’re talking about is expanding with m2x, why would I do that?

its already been done :slightly_smiling_face:

I do understand 32 GB can feel a little claustrophobic, maybe something up to 64GB would have been a nice touch but, as discussed many times before, having large quantities of eMMC doesn’t make sense, its not a very fast and is quite expensive, better than simple SD cards, but when it comes to something that is much more effective cost wise and reliable, SSDs are much better in this area

1 Like

months ago I watched this video and bought vim 3. But I didn’t find what I expected. Ubuntu does not play smoothly as in the video.

breaks are evident in the video, but not too much! There is a lot of breaking in real life.