AMD Posts Linux Driver Patches For Video Core Next 5 "VCN 5.0"

[phoronix]

Phoronix: AMD Posts Linux Driver Patches For Video Core Next 5 "VCN 5.0"

It looks like next-generation RDNA4 graphics will feature a new iteration of the Video Core Next (VCN) for accelerated video encode/decode. VCN 5.0 patches were posted today for the AMDGPU Linux kernel driver...

https://www.phoronix.com/news/AMD-VCN5-AMDGPU-Linux-Driver

[M.Bahr]

AMD got some really good professional video encoder cards based on fpgas. I wish they would go back to those instead of producing completely asic based ones like with the Alveo MA35D. Don't get me wrong. I know there are pros and cons but fpgas got the big advantage of flexibility on the hardware level. In times of hardware products with increasing complexity also errors increase. And fpgas are a very good means to debug those way deeper at the hardware level than higher level software could.

Remember fpgas are a field-programmable gate array. You can literally replicate hardware logic that is way faster than abstract programs runing on general purpose compute blocks. The doom chip was a really cool demonstration in that regard. https://hackaday.com/2020/05/13/the-doom-chip/

You can replicate complete new video codecs at the hardware level. You "burn the algorithm into wires" to put it with the words of the doom chip developer. But of course this would prevent people from buying new rigid asic based video encoders whenever a new video standard comes out.

Sooner or later AMD has to do more with the xlinix pi or a competitor may fill this gap. I really would like to see more fpgas in consumer products. The new ryzens made a good step forward but i also would like to see fpgas in gpus. This gives possibilities for flexible A.I. replication at the faster hardware level.

[Laughing1]

Not yet on https://en.wikipedia.org/wiki/Video_Core_Next

[ezst036]

Hopefully AMD gets this working in OBS. There have been a lot of generations come and gone that do not support 4:4:4 leaving NVenc the only choice for many when doing streaming.

[OneTimeShot]

Remember fpgas are a field-programmable gate array. You can literally replicate hardware logic that is way faster than abstract programs runing on general purpose compute blocks. The doom chip was a really cool demonstration in that regard. https://hackaday.com/2020/05/13/the-doom-chip/
.

FPGAs are great for small volume runs, but after you ship ~10,000 of the things, hardwired chips are much cheaper. Also the maximum speed of an FPGA is much lower than hardwired chips.

It would be interesting to have an FPGA in a PC, but I'm not sure if it has a mass-market use case. The video encoder/decoder block is basically a small CPU with a custom instruction set anyway. Only "the expensive bits" of the process are done in tailor-made hardware.

[Danny3]

Cool, but when CEC support?
I really want to use my TV's remote control to control programs on my computer.

[Serafean]

Cool, but when CEC support?
I really want to use my TV's remote control to control programs on my computer.

That might work if you use an usb-c/DisplayPort to HDMI adapter. (A very carefully selected one) https://www.phoronix.com/news/AMDGPU-Nouveau-CEC-Tunnel

[Anux]

There have been a lot of generations come and gone that do not support 4:4:4 leaving NVenc the only choice for many when doing streaming.

What does 4:4:4 have to do with streaming? If your bandwidth limited you should use every compression advantage you can get and color detail reduction with 4:2:0 is one of them.

Is there even a platform that supports streaming with 444?

[hajj_3]

maybe they will add h.266/vvc decoding support.

[Serafean]

maybe they will add h.266/vvc decoding support.

Hopefully they don't. Leave h.266 to die under a mountain of patent trolls.
Unfortunately DVB-T2 has already been extended to support it.

Companies known not to be a part of the Access Advance or Via-LA patent pools as of November 2023 are: Apple, Canon, Ericsson, Fraunhofer, Google, Huawei, Humax, Intel, LG, Interdigital, Maxell, Microsoft, Oppo, Qualcomm, Samsung, Sharp and Sony.

Big names not interested in VVC. I'm still kind of hopeful.