Welcome to our community

Be a part of something great, join today!

Log in
  • Hey all, just changed over the backend after 15 years I figured time to give it a bit of an update, its probably gonna be a bit weird for most of you and i am sure there is a few bugs to work out but it should kinda work the same as before... hopefully :)

RED + NVIDIA Announce Realtime 8K R3D GPU Decoding

I have no doubt that everything upto 8k will work with the $2.500 RTX Titan around NAB 2019 (Playback 8k 5:1 at around 24..30 fps).
When you apply effects the same will happen as with current GPU's "no more real-time playback in full res" and with TNR applied you won't even have real-time play-back in half-res.

The $1.200 question is: how much resolution + some effects + TNR can the RTX2080ti handle without running out of memory?
 
andrewhake said:
Yes that's all well and good. But it all just comes down to RED opening up their SDK to allow this to take place on the GPU. Nvidia/CUDA isn't the only platform capable of processing 8K R3Ds on a GPU in this way, it's just the only one being given access to this at the moment. Hopefully it will be opened up beyond CUDA, because competition is good.
Click to expand...

It actually sounds like the opposite, to be honest. nVidia put their money where their mouth is, and gave resources (two engineers and a weekend/week, apparently), and hence they got there long before anyone else (including RED) could have.

Re-watch the video of the presentation. It sounds like an informal meeting (at NAB) turned into a 'we weren't joking', turned into 'nice code; we made it 400% more efficient over the weekend" (before photokina, or IBC, or whatever convention was August/Sept.)

Also, Jarred/RED have already said it'll make it's way to OpenCL some time this year (though no word if it'll be as capable as the CUDA version).

Any dev has access to the SDK (they just need apply). nVidia were the ones that said, 'pffft, child's play' and knocked it out of the park.
 
Ready for some realtime 8k...and Maya goodness
IMG-0873.jpg
 
Michael Lindsay said:
Do the 2 fans suck or is one sucking and the other blowing...?

If they suck and blow which one blows?

Trying to work out the cooling implications of 2 of these (and another 1 slot quadra for GUI) in a z8..

Ps I do realise my questions are like a 70s comic setup..;-)
Click to expand...

For best cooling allow at least one slot between them. Or put something small between them like a BMD card or NIC.
 
Michael Lindsay said:
Do the 2 fans suck or is one sucking and the other blowing...?

If they suck and blow which one blows?

Trying to work out the cooling implications of 2 of these (and another 1 slot quadra for GUI) in a z8..

Ps I do realise my questions are like a 70s comic setup..;-)
Click to expand...

Both fans blow air into cooling fins and the air gets channeled towards the rear of the card where the hot air is forced thru holes in the back.
 
andrewhake said:
Yes that's all well and good. But it all just comes down to RED opening up their SDK to allow this to take place on the GPU. Nvidia/CUDA isn't the only platform capable of processing 8K R3Ds on a GPU in this way, it's just the only one being given access to this at the moment. Hopefully it will be opened up beyond CUDA, because competition is good.
Click to expand...

We offered it to both.
 
Misha Engel said:
pic_disp.php
Click to expand...

They are both good. Common misconception is 1 fan is better than 2. If the card has 1 fan, it has to push more air to get the same thermal co-efficiency when compared to 2 fans. So, the single fan has to oscillate at a much higher speed on demand. In an industrial environment this may not be a big deal. In comparison, well designed two large fans could do the same with less noise.

Gamer's tend to stack more graphics cards in to an enclosure (up to 4 cards in some instances). Imagine the amount of noise generated if all cards had one fan and were running at their maximum speed!
 
Michael Lindsay said:
Thank you! So it is essentially a blower card but with 2fans rather than 1?
Click to expand...

You are welcome. Yes they are both blower fans. nVidia tech-support encourages end users to setup multiple cards in two configurations. For editors, VFx and other graphics professionals who are using two 8k displays (I wish), no SLI configuration. And for gaming community, use SLI. Please note all cards needs to be identical (make, model, Ti or whatever) to work properly.
 
Some early thoughts as I'm now testing the prerelease build of RCX with the new GPU Decode Features.

First, here it is running on one of my boxes with some 8K FF, 24fps, REDCODE RAW 5:1 material (currently the most demanding data we can shoot):
https://www.instagram.com/p/BrvoD_jBTk3/

It's impressive where this is. The effort began back around NAB 2018 and here we are some 8 months later and I'm already impressed. There's a few kinks to iron out, but it's already doing good things. I don't have my RTX cards yet, but I'll be testing it on other machines shortly with previous gen cards.

From what I've seen in the last couple weeks you'll be able to get 24fps playback with the modern RTX stuff and better than that with bigger RTX cards like the Titan and Quadro. From talks with engineers at the event nvLink can take that further and that's what I'll be focused on testing next.

This is all very focused on getting realtime 8K playback of REDCODE RAW material, but there's also a really welcomed boost for 1/2 and 1/4 debayer workflows. Which can translate into a few things from a practical perspective in an NLE or on more modest/older systems.

Just some open thoughts on CUDA/OpenCL/other discussion. If you recall in the past with GPU acceleration Cuda had the lead, then OpenCL took over, then a couple years back Cuda took the first speed lead again. Nvidia worked hard on adding a few things that OpenCL currently doesn't have. Eventually however this will hopefully get deployed to all directions to be had here. But the efforts by the RED Software Team and Nvidia are really, really appreciated here.

What's truly exciting for me is how this will effect workflows and perspective of workflows regarding higher resolution motion picture content.
 
You must log in or register to reply here.
Back
Top