Alex Gerulaitis
Forum Replies Created
-
[Matloh Zaluski] “can u use k4000 benefits and still be using decklink just to monitor video on broadcast monitor?”
Yes.
[Matloh Zaluski] “k4000 for doing the horsepower stuff like gpu acceleration and connceting computer monitors to if to see the timeline window etc while monitoring on decklink?”
Yes.
-
You answered your own question: you don’t need K4000 just for 10-bit monitoring in PP and AE if you’re using Decklink 4k Extreme for that, with a proper monitor.
[Matloh Zaluski] ” i can use blackmagic decklink 4k extreme just to have sdi i/o and k4000 quadro for doing horsepower stuff like gui and effects – right?”
Exactly.
Alex Gerulaitis
Systems Engineer
DV411 – Los Angeles, CA -
[Amy Wilson] “I imagine upgrading my raid card means formatting my raids.”
Not from R380 to R680. R680 will recognize the existing array that was created on R380.
[Amy Wilson] “Or do i just hang onto this setup until its run into the ground?”
If it does what you need and there’re no major performance or other bottlenecks – sure.
Alex Gerulaitis
Systems Engineer
DV411 – Los Angeles, CA -
As long as camera’s long GOP codec doesn’t “break” under pressure of fast motion, the picture quality will always be better vs. the same camera encoding into an i-frame-only codec. GOP codecs are simply way more efficient vs. i-frame ones.
I-frame doesn’t get choppy in fast motion – it’s the GOP codec that can “break” (fail) in extreme circumstances, when the amount of information (fast motion, lots of picture details) exceeds its capacity for processing – but like I said, this mostly applies to old, cheap, V.1 cameras. The only reasons for i-frame (vs. GOP) in cameras are: (a) camera’s brain not powerful enough to encode into GOP, (b) easier to work with in post. Both of these reasons may have been valid, say, five years ago – not so much anymore.
Alex Gerulaitis
Systems Engineer
DV411 – Los Angeles, CA -
[Stephen Pickering] “I thought I could connect my two editing systems on the same gigabit network and simply “Share” the RAID 5 drive. Would this work?”
Sure it would. If possible, make it the same drive letter on both systems.
[Stephen Pickering] “Does this setup look acceptable?”
Seems perfect for your goals: a combination of drive redundancy and backup to maximize uptime and minimize data loss. Access over LAN can be anywhere from 20MB/s to 110MB/s regardless of how fast your local storage is – something to be aware of.
Alex Gerulaitis
Systems Engineer
DV411 – Los Angeles, CA -
Thanks DRW. Would you be able to come up with a competitive comparison, on how Promax MAM solution stacks up against CatDV and others – on features and prices?
(Yes, I am looking for someone to do my job – i.e. do a research on entry level MAM solutions for Adobe CS. :))
-
I was wrong – the VCAs use Grid K2 cards (not Teslas), and each VM (virtual machine) is given 32GB RAM.
(Side note: I must have misheard and was under impression that the VMs would work directly on the GPUs and not host CPU(s); in retrospect, that was a silly thought. The VMs work on main CPU(s), with a GPU dedicated to each VM.
This may necessitate careful planning for heavy duty deployments with CPU-intensive tasks, as those are unlikely to benefit much from a VCA.)
The rest of the stuff is valid, i.e. each remote session is married directly to a specific GPU, although NVidia says that may change in the future, i.e. an 8-GPU VCA may support more than 8 users.
The concept is quite a bit different from Adobe Anywhere: VCA doesn’t require locally installed software other than a lightweight remote connection agent. Adobe Anywhere requires a full installation of the software on the client system, as far as I know. There’re other differences as well: VCAs exist in two fixed configurations; Adobe Anywhere – in a variety of them, on COTS servers with certain minimum requirements; VCAs are limited to dual 10GbE and an 8-lane PCIe slot for all connectivity and storage if I remember right; no such limitations with AA.
That said, VCAs are more flexible in some respects – they can run many apps that benefit from fast GPUs. AA – only (some) Adobe apps.
-
[Anhtu Vu] “I still don’t understand why so many software developer still recommend the Quadro 4000 when there are much cheaper alternatives that actually performs better.”
10-bit color support, single point of support in enterprise environments (important for some). Quadros also generally run cooler and are better made. There aren’t too many folks out there for whom that outweighs the vastly better performance/price ratio with GeForce cards, but there are some.
Alex Gerulaitis
Systems Engineer
DV411 – Los Angeles, CA -
[walter biscardi] “I’m going to be meeting with the Atlanta nVidia rep in the near future to really figure out exactly what the box does and does NOT do in the workflow.”
No multi-monitor support, and 1920×1200 max resolution. Other than that, working with one is the same as having a remote session into another computer that has Pr CS6 installed. Available memory may be another limitation – Tesla K20 has 5GB if I am not mistaken, and that’s all the memory you’ll get.
The biggest advantage I see is being able to open MacBook Air or even an iPad (anything, really), dial the VCA for a remote session and start editing with full GPU acceleration. It won’t replace desktops that are well configured for CS, not even iMacs.
Alex Gerulaitis
Systems Engineer
DV411 – Los Angeles, CA -
[calvin nguyen] “I really would love to edit without having to render. does my monitor resolution slow me down, I am at 2560×1600?”
With your current GeForce 9600 – probably.
I’d look at the following to run CS4 better and to have better chances with CS6+:
– Intel Core i7 CPU (not AMD)
– 8GB is fine for CS4; for CS6 you might want to up it to 12-16GB. Memory is cheap.
– GTX-650Ti graphicsHTH.
Alex Gerulaitis
Systems Engineer
DV411 – Los Angeles, CA