Welcome to our community

Be a part of something great, join today!

  • 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 SDK and updated REDCINE-X now available

Since Apple settled with Qualcomm, perhaps nVidia is next... OK, not holding my breath ;)

FWIW, I find it hard to imagine a $10K+ MacPro 7,1 build without nVidia/CUDA that would merit that level of investment...

Cheers - #19

Yeh. 7,1 Mac Pro sans nVidia is DOA.

Like it or not nVidia/CUDA is dominant in a lot of market segments. Not everyone sits around running Resolve benchmarks all day.
 
Yeh. 7,1 Mac Pro sans nVidia is DOA.

Like it or not nVidia/CUDA is dominant in a lot of market segments. Not everyone sits around running Resolve benchmarks all day.

Maybe, maybe not.

I think Apple is more focused on ProResRAW and their own CPU+GPU combination.
The market for .R3D post processing is tiny compared to AVC, HVEC(consumers), ProResRAW(prosumers) and ProRes(fossils).

FCPX and Davinci Resolve both work on iMac's with every codec in realtime except... the heavy one.
 
Bravo RED team. I can playback 6k 50fps footage from my EPIC DRAGON with a Nvidia GTX1060 in a 2 years MSI 4K laptop. It uses 46% of my CPU and 5.5GB out of 16GB of the GPU. 8K begins now for me.
 
Last edited:
Perhaps there's enough demand for Apple/AMD to co-develop a hybrid chip - perhaps optimized for Metal2 by an army of coders - that could be competitive with the nVidia/CUDA solutions...

Until then: it's either OSX - or - nVidia/CUDA on Windoze? With the new code - plus the general improvements in Win10 - the situation has "evolved".

I'd still rather have nVidia/CUDA running smoothly on OSX if I can...

Cheers - #19
 
This has been extremely empowering and a revelation really on the REDCINE-X PRO side of things. It's insane playing back an 8K timeline properly now this way at full debayer.

I'm curious to see how and where this gets implemented into 3rd party NLEs. Oddly throughout this process due to some of my projects I've been looking at how each application handles various codecs playback and rendering. It feels like there's a lot of work that can be done to improve a few things for sure. This is certainly a beneficial technology to working professionals.

It would be a nice day where resource allocation is a bit more optimized for most major applications. I understand the difficulties however of making stuff compatible with "everything" and some of these packages are just stacks of hamburger patties just piled on and getting a bit wobbly higher up there. Might be time to look at the burger as a whole again team A and B.

Kudos out to the beta team and that feedback structure btw.

I'm guessing that Assimilate will have a build incorporating the new SDK in a day or two... Assimilate had realtime 8K playback with a 2080 and six cores working when it got to NAB (and the demo got hosed because the demo machine that B&H gave Mazze didn't have an RTX card).

Black Magic will most likely work this into one of the upcoming beta builds also, and I suspect that SGO will get it integrated into Mistika in a few weeks.

It's going to make life much better for people using Red cameras :)
 
Black Magic will most likely work this into one of the upcoming beta builds also, and I suspect that SGO will get it integrated into Mistika in a few weeks.

It's going to make life much better for people using Red cameras :)
We mostly use Davinci Resolve for both grading and editing. I have a slight worry that the gpu usage from Red files could impact Resolve negatively, since everything in resolve is using the GPU so much. Of course in theory one could solve it by having multiple cards.
I am a bit on the fence if I should get the Titan or the 2080 RTX, I do get out of ram sometimes currently, but the Titan is so much more expensive.

On another note I hope we will get laptop versions at a point that has more RAM. 8GB is nice, but as we know not enough for some things. When I run 8k media I am currently using 7GB in Redcine, which makes me nervous.
/Andreas
 
We mostly use Davinci Resolve for both grading and editing. I have a slight worry that the gpu usage from Red files could impact Resolve negatively, since everything in resolve is using the GPU so much. Of course in theory one could solve it by having multiple cards.
I am a bit on the fence if I should get the Titan or the 2080 RTX, I do get out of ram sometimes currently, but the Titan is so much more expensive.

On another note I hope we will get laptop versions at a point that has more RAM. 8GB is nice, but as we know not enough for some things. When I run 8k media I am currently using 7GB in Redcine, which makes me nervous.
/Andreas

Yeah it would be great with some info on different graphic cards combos that work etc. Im sure there are plenty of redusers that have a Nvidia 1080Ti and want to know if its any good to combo with another 1080Ti or if its better to run a single 2080Ti. Do users get the out of ram message with their workstation.
 
I can playback 8k 60fps footage with a 1080ti in a 4-5 year old Xeon workstation. It uses 13% of my CPU, 11% of my GPU and 6.1GB (out of 11GB) of the GPU memory as per Windows task manager. Amazing!
 
Final nail in the coffin for Rocket-X in the windows world. The trusty Rocket still makes a big difference on our iMac Pro but thats about it.
 
What is equally impressive is that the skimming is so much faster. Before it would take 3 seconds before the video would begin playing, now it is probably 0.5 second.

Feature request. When you skim in full debayer it would be nice if it drops to half debayer until you let go of the mouse, then it switches back to full debayer. That way it will skim much faster. This is how After Effects does when you move in timeline to keep up speed. As it is now skimming is super fast in half/quarter but a tad slow in full debayer.

Andreas
 
Single Titan RTX cruising along at 60% usage playing 8k full debayer...yawn :biggrin:
 
Single Titan RTX cruising along at 60% usage playing 8k full debayer...yawn :biggrin:

Sounds good, what kind of 8k? In Resolve with 8k.R3D (9:1) 25 fps, the Quardo RTX 6000/8000 won't go much faster than 23..25 fps(GPU limited).
Is REDCINE-X also using a different kind of de-bayering compared to Resolve? Titan RTX is kind of the prosumer version of the Quadro RTX6000.
 
Sounds good, what kind of 8k? In Resolve with 8k.R3D (9:1) 25 fps, the Quardo RTX 6000/8000 won't go much faster than 23..25 fps(GPU limited).
Is REDCINE-X also using a different kind of de-bayering compared to Resolve? Titan RTX is kind of the prosumer version of the Quadro RTX6000.

Misha,

Are you running a Quadro 6000 or 8000 GPU?

And it that all what it gives you?
 
Puget hasn't done any testing with the new REDCINE-X featuring GPU Decode as far as I am aware of.

Titan RTX can get you 8K FF 5:1 24-30fps full debayer, which is what we showed in March. That's on playback. Render speed, more variables come into play to just publish a number, but depending on codec you can/will see faster than realtime encoding to 4K and even 8K if the rest of the recipe plays nicely.
 
Puget hasn't done any testing with the new REDCINE-X featuring GPU Decode as far as I am aware of.

Titan RTX can get you 8K FF 5:1 24-30fps full debayer, which is what we showed in March. That's on playback. Render speed, more variables come into play to just publish a number, but depending on codec you can/will see faster than realtime encoding to 4K and even 8K if the rest of the recipe plays nicely.

Good. Here is what we have seen with some of our in-house PCs. These are not the most high-end performance workstations that we have. We built a few systems (just for shits-and-giggles I suppose).

Here are specs of 5 PCs:

Motherboard: ASUS ROG Maximus Hero (Wi-Fi) [I know, I know ...... we wanted to see whether these would work].
CPU: Intel Core i9-9900K (with Silicon values between 136 to 142 - got lucked out). So we OCd them to 5.54GHz.
GPU: a single ZOTAC GAMING GeForce RTX 2080 Ti AMP Extreme ZT-T20810B-10P (boost clocked to 1856 MHz - in-house modded liquid cooled).
RAM: G.SKILL TridentZ RGB Series DDR4 4400 MHz and/or Corsair Vengeance RGB Pro 64GB (4x16GB) DDR4 3600MHz (PC4-28800); 64GBs on each PC.

All systems are liquid cooled with CoolerMaster MasterLiquid ML360R radiators with Noctua NF-A12x25 PWM fans (yes these things are super quiet).

R3D files are captured with DSMC2 Helium and a couple of Monstros and 80% of captures are at 24 fps 8K FF @ 5:1 and 20% at 8K 2.4:1 @ 5:1.

So far, we get realtime playback on full debayer playback on these two formats with updated REDCINE-X and GPUs are at 46% load. So far we did not have any crashes and everything seems to be stable.
 
Good to hear Neil. I'm obviously not building a million PCs to test this stuff out, but all my main workstations are X299 and either Quadro RTX or Titan RTX based. One box has two Quadro P6000s in it and I see a difference between that and the Quadro RTX performance. That also could be due to how that specific system is setup and I am not tinkering with it if I can avoid it.

All the newer notebooks in my life from Dell, Razer, Gigabyte, and Asus have 2080 MaxQ GPUs and those do shockingly well for what they are. It will be interesting to see if we get some full sized laptops tapping into maybe multiple GPUs.
 
All the newer notebooks in my life from Dell, Razer, Gigabyte, and Asus have 2080 MaxQ GPUs and those do shockingly well for what they are. It will be interesting to see if we get some full sized laptops tapping into maybe multiple GPUs.

Has anyone tried multiple GPUs? Does it scale ok? I know MSI had an 18.4" laptop with 2x1080 GTX cards in them. Could be interesting with 2x2080 RTX since they are not the mobile versions. Only problem with the 18" was that the screen was a gaming 1080p monitor which is not what I want. I like UHD and of course better color values.

To be fair I am not sure how important it is to have 2x2080 RTX on a laptop
Andreas
 
Good to hear Neil. I'm obviously not building a million PCs to test this stuff out, but all my main workstations are X299 and either Quadro RTX or Titan RTX based. One box has two Quadro P6000s in it and I see a difference between that and the Quadro RTX performance. That also could be due to how that specific system is setup and I am not tinkering with it if I can avoid it.

All the newer notebooks in my life from Dell, Razer, Gigabyte, and Asus have 2080 MaxQ GPUs and those do shockingly well for what they are. It will be interesting to see if we get some full sized laptops tapping into maybe multiple GPUs.

Thank you Phil. You are correct, modern laptops with 2080 graphics can do wonders. I wish Jarred would some how come up with somewhat of a NLE associated with REDCine-X to complement with what they had done with nVidia partnership.

Such an endeavor does NOT have to be a fully capable or an exceptional NLE to compete with already established one's. Something that could use the already optimized OP code with nVidia to make simple EDLs to be transferred/transcoded to "Any" NLE out there.
 
Back
Top