IT'S HAPPENING

twitter.com/IntelGraphics/status/1551945518141644804

IT'S HAPPENING

AV1 HARDWARE ENCODER FOR VIDEO STREAMING SITES

AV1 EVERYWHERE EXCEPT ON Any Forums

Attached: Arctic.png (602x569, 115.36K)

Other urls found in this thread:

phoronix.com/news/Intel-CR-22.28.23726
docs.amd.com/bundle/ROCm-Installation-Guide-v5.2.1/page/Prerequisite_Actions.html
twitter.com/AnonBabble

They can't even make a good normal GPU first

Oh boy, can't wait to pay $2500 for a 2070 equivalent gpu

>I MUST BUY [LATEST THING]

I doubt the encoder is worthwhile for bitrate-constrained use cases such as Any Forums. Software encode is likely to get a much better quality with filesize that low.

Is this Intel's equivalent to Nvidia Quadro?

No, this is server only card

Arc Pro is what you are thinking about

Is it possible to use this in a render farm? I've been considering building a render server for Blender.

Does it mean GPU reached CPU levels of virtualization or is just intel cheap attempt to take nvidia and amd market share?

Nine more years.

>They can't even make a good normal GPU first
Of course not. They hired Raja. If you weren't expecting a delayed shit consumer GPU as a trojan horse for a pro offering like in OP you haven't been paying attention.

I wish they would sell a low-cost and low power card with just the media engine, no 3d capabilities. I would buy that and put it in my media server in a heartbeat.

i will buy it for 1 dollar

you mean integrated graphics
because Iris Xe already supports that shit

Any Forums isn't a video streaming site

Attached: m0t.png (372x1179, 60.26K)

poor moot

>starting with cloud gaming and media streaming
In other words, there are no drivers for things like Tensorflow.

>intel cheap attempt
It's not cheap at all. Intel is losing shitloads of money doing this. They couldn't even fab it themselves and the design team had to be built from the ground up. They're doing this at the same time that their new CPUs and chipsets are also loss leaders. Their only hope now is that China invades Taiwan. Their last cash market, server CPUs, is getting Btfo by AMD and custom ARM chips. Something crazy is going to have to happen for this to work out for them.

Competing with Nvidia in AI is suicide.
You need to get everyone to adopt a proprietary GPU API like CUDA, OneAPI, or ROCm else your product stands no chance, because nobody's using Vulkan Kompute for some reason.

>kiketel
>AV1

Not touching it it sis.

>encoders are for Any Forums!
lol

>nobody's using Vulkan Kompute for some reason.
There's ncnn by Tencent which I think runs on Vulkan though I don't know the specifics.

>CUDA
The only one that's ubiquitously supported on hardware. Multi-generational support, every library supports it as a target.
>OneAPI
Literally disables support for perfectly good hardware - phoronix.com/news/Intel-CR-22.28.23726
>ROCm
Literally has no hardware support for their own GPUs. No consumer GPU is officially supported, only Radeon Pro and Instinct - docs.amd.com/bundle/ROCm-Installation-Guide-v5.2.1/page/Prerequisite_Actions.html
There is no contes here, CUDA is light years beyond competition.