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 :)

New Red Rocket for 8K

Jagger Christian

Well-known member
Joined
Apr 14, 2008
Messages
274
Reaction score
0
Points
16
Location
Hollywood
Anybody thinking that there might be a new Red Rocket coming sometime that is improved or made for 8K? Thoughts?
 
Take a look at Jarred's post in the Update thread. Re-posted below:

Jarred Land said:
As for the RocketX and Helium.. as you know we have been focusing our efforts on GPU decode for the last while.

So yes Rocket-X plus a good GPU is gonna get you around 24fps Full 8K on Helium.

But you can also get close to that full 8K 24fps with a new Titan-X GPU and Intel's largests core Xeon in a dual cpu box without the Rocket-X.

I love you Rocket-X but in a year or two I think we should be at the point where GPU + CPU puts you out to pasture.

Nvidia is just dominating right now and their GPU's are advancing so quickly it's the horse I would bet on.
 
Great Stacey!! :)
 
Q1) Last time I read about the Rocket-X ... there were 3 programmable 'banks', and there was at least 1 of them that had Not yet been utilised. Is that still the case?

Q2) IF the Rocket-X requires input from a GPU to achieve 24fps @ 8K,
AND one of the Banks of the Rocket-X is being used to perform Video output,
AND the RRX is more efficient than a GPU...

Perhaps there is an option here for RED to switch all the Rocket-X banks to RAW processing (ie Remove the Video output capabilities), and let the GPU to the VIDEO output.

Q3) If the Rocket-X is PCIe x 16 and the speed of a single lane is 2.5Gb ... does this mean that Thunderbolt 3 = bidirectional 40 Gb can completely max out the data ingesting / transmission of the Rocket-X?

... could a Rocket-X still have great value for chunching 8k on a 15" MBP Late 2015 if hooked up over Thunderbolt 3?

AJ
 
Q1) Last time I read about the Rocket-X ... there were 3 programmable 'banks', and there was at least 1 of them that had Not yet been utilised. Is that still the case?

Q2) IF the Rocket-X requires input from a GPU to achieve 24fps @ 8K,
AND one of the Banks of the Rocket-X is being used to perform Video output,
AND the RRX is more efficient than a GPU...

Perhaps there is an option here for RED to switch all the Rocket-X banks to RAW processing (ie Remove the Video output capabilities), and let the GPU to the VIDEO output.

Q3) If the Rocket-X is PCIe x 16 and the speed of a single lane is 2.5Gb ... does this mean that Thunderbolt 3 = bidirectional 40 Gb can completely max out the data ingesting / transmission of the Rocket-X?

... could a Rocket-X still have great value for chunching 8k on a 15" MBP Late 2015 if hooked up over Thunderbolt 3?

AJ

Yeah, I think they recently did a firmware update and I thought that was to make it TB3 compatible and work with 8k Vv footage, at the very least.
 
White paper please

White paper please

It would be great to get a "state of the solid state" report on the RR-X. Unless I missed it, there has never been a white paper or other tech document that breaks down the operational parameters of the RR-X and how it can best be leveraged in a complete system configuration. The RR-X may be a footnote in 5 years, but for those of us who have one it would be nice to have some guidance when assembling resources and designing workflows.

I take Jarred's note about the trajectory of R3D file handling NOT being one trick pony accelerator cards. He goes on to laud nVidia GPU development - does that mean that Rob and the RedTeam are looking to leverage CUDA vs OpenCL? Does it mean that AMD GPUs are essentially going to handicap Macs (assuming Apple sticks with AMD) in terms of R3D processing? What difference, if any, will there be with using TB3 for RR-X or nVidia Pascal in external enclosures vs TB2 with R3Ds? Is it more about data rates or compute cycles? Does the RR-X take load off the CPU that is then available for other processing tasks?

Would giving us more technical info provide any competitive advantage to other manufacturers? If not...

Cheers - #19
 
It would be great to get a "state of the solid state" report on the RR-X. Unless I missed it, there has never been a white paper or other tech document that breaks down the operational parameters of the RR-X and how it can best be leveraged in a complete system configuration. The RR-X may be a footnote in 5 years, but for those of us who have one it would be nice to have some guidance when assembling resources and designing workflows.

I take Jarred's note about the trajectory of R3D file handling NOT being one trick pony accelerator cards. He goes on to laud nVidia GPU development - does that mean that Rob and the RedTeam are looking to leverage CUDA vs OpenCL? Does it mean that AMD GPUs are essentially going to handicap Macs (assuming Apple sticks with AMD) in terms of R3D processing? What difference, if any, will there be with using TB3 for RR-X or nVidia Pascal in external enclosures vs TB2 with R3Ds? Is it more about data rates or compute cycles? Does the RR-X take load off the CPU that is then available for other processing tasks?

Would giving us more technical info provide any competitive advantage to other manufacturers? If not...

Cheers - #19

I agree with everything you said. I bought a used ROCKET X in hopes of stopping the new computer merry-go-round.

Once it is totally enabled (if not already) then it is just the newest interface to get the ROCKET X to meet or exceed new computers with GPU cards, I would hope.
 
Why on earth would anyone by a RR-X? Even used. The GTX1080 rips it to shreds. And in a PC you run 2 or 3 of them for even more power!
You could get 6 GTX1080's for the price of one new RR-x I think.
 
Why on earth would anyone by a RR-X? Even used. The GTX1080 rips it to shreds. And in a PC you run 2 or 3 of them for even more power!
You could get 6 GTX1080's for the price of one new RR-x I think.

Because not everyone happens to have a workstation class PC with multiple CPU's. A number of tasks are done on the CPU (particularly when working with larger frame sizes), so no matter how many GPUs or Red Rocket X's you have, you will run into playback/processing limitations with anything based on a single CPU for anything larger than, say, 1/8 or 1/4 debayer. That would mean all trashcan Macs, all laptops, and most users' PCs. Jarred mentions this in his statement, but most people here seem so completely preoccupied with GPU power that they're ignoring it. He's not suggesting that a "standard" PC or a laptop is a substitute today because it isn't. He's talking to people who are using workstation class machines to do large scale processing and DI work, not to cameramen or DITs who are doing small transcoding jobs.
 
Why on earth would anyone by a RR-X? Even used. The GTX1080 rips it to shreds. And in a PC you run 2 or 3 of them for even more power!
You could get 6 GTX1080's for the price of one new RR-x I think.

The 1080's are fine, but no, they are not faster than a RR-X.
 
Because not everyone happens to have a workstation class PC with multiple CPU's. A number of tasks are done on the CPU (particularly when working with larger frame sizes), so no matter how many GPUs or Red Rocket X's you have, you will run into playback/processing limitations with anything based on a single CPU for anything larger than, say, 1/8 or 1/4 debayer. That would mean all trashcan Macs, all laptops, and most users' PCs. Jarred mentions this in his statement, but most people here seem so completely preoccupied with GPU power that they're ignoring it. He's not suggesting that a "standard" PC or a laptop is a substitute today because it isn't. He's talking to people who are using workstation class machines to do large scale processing and DI work, not to cameramen or DITs who are doing small transcoding jobs.

Any recent changes to Red Rocket-X performance on single core machines? I'm running a 7700K at 4.7GHz and am hoping to get Helium 8K 24fps playback at half or full debayer by adding a RR-X over TB3. My rig has 64GB 2400 RAM, 2x GTX1080 (1 is running at 95% performance of the other over TB3) and SSD RAIDs. I'm really happy with the system except being stuck with 1/8 res to get 24fps playback on Helium / RC-X. I think I'm maxed out lane-wise with the 7700K so a RR-X may not help much. Any advice would be much appreciated.
 
Any recent changes to Red Rocket-X performance on single core machines? I'm running a 7700K at 4.7GHz and am hoping to get Helium 8K 24fps playback at half or full debayer by adding a RR-X over TB3. My rig has 64GB 2400 RAM, 2x GTX1080 (1 is running at 95% performance of the other over TB3) and SSD RAIDs. I'm really happy with the system except being stuck with 1/8 res to get 24fps playback on Helium / RC-X. I think I'm maxed out lane-wise with the 7700K so a RR-X may not help much. Any advice would be much appreciated.

1/2 res 8K? not a chance not even anywhere close... with the 1080's and/or that CPU with RRX nope. 3 Titan XP's plus RRX with a multi core dual CPU config would get u there with room for fx at 1/2 res. But yes your number of lanes will never allow it, so you'd have to start a new build with dual xeon or a lesser build starting a single only 40 lane multi i7 for much weaker performance at 1/2 res, both of which are significantly more expensive than your 7700K. This all before talking displays, storage, etc. A lot of people have asked this type of question and the only way right now is to spend at minimum $20k with a system based on no less than the above mentioned components.
 
This is the biggest issue with 8k....decoding it. I've got a system with dual 1080Ti's, but the bottle neck is the CPU. It's an Intel® Core i7® 6950x Processor 3.0GHz, watercooled & oc'd to 4.2GHz, but maxed out when working in 8k. Machine is spec'd for 4k, so all good, but nice to know what it can and can't do.

For 8k+, does it have to be dual xenon???
 
Back
Top