- Moderator
- #21
Jeff Kilgroe
Well-known member
Only available as Tesla compute cards, not incorporated into GeForce or Quadro display cards.... yet.
What lack?
Working fine here, although ProRes support on Windows is a crap shoot at times and direct output is impossible without going through hoops in an external app. I guess my question would be, what issues do you speak of?
Resolve is not a second tier alternative with a DaVinci stamp on it. It's the real deal, it's the full DaVinci Resolve software as it always has been. Actually better because it continues to improve and become more capable. IMO, it's the best color correction system out there for the money. If you want to do better, you had better start shopping for Lustre and Quantel solutions. At any rate, you could answer all these questions yourself by giving it a try. The Lite version is [almost] fully functional and it's free. It is locked to single GPU support and can't work in or output resolutions greater than 1080p. I think those are the only restrictions, but I'm not 100% sure -- I don't use the lite/free version.
Not sure what you're getting at here. But your comments are right on about "moving over" to LGA 2011 and replacing components if you're looking at it from an upgrade point of view. For a new workstation build, there isn't anything I'd build for myself right now that isn't using the LGA2011 socket. Desktop i7 or Xeon. When working on any of this stuff we talk about around here -- high resolution video, motion graphics, R3D workflow, etc.. I don't think there's any such thing as too much computer. Moving into 6K, the 3960X (or 'K version) should be serviceable. There are too many unanswered questions for now about 6K, though... Something tells me we're going to see more than just increased pixels. I'm expecting another jump like we had from the RED One to Scarlet and Epic where the R3D decoding is more complex and demanding, we moved from a 12bit to a 16bit format, that sort of thing. So there's a lot yet we don't know... I wouldn't be surprised if they move 6K Dragon R3D's to and 18 or 20 bit container.
Not just research, but real-world experience here... And I concur on the RAM. With 64bit applications these days, my *MINIMUM* ideal RAM amount for most common workstations is 4GB per core. Better yet, I like to extend that to 4GB per thread. Beyond that it's mostly just gravy, yet that gravy can be very useful if your system is continuously choking on large data sets. R3D workflow is a great example. For playback and scrubbing, even with a Rocket card, it helps to cache as many frames as you can. If you're working with 5KFF frames in half-resolution, that's still just under 20MB per frame in RAM. At 24fps, it takes about 480MB to cache one second. If you're scrubbing back and forth on a 10 second shot, thats almost 4.8GB in RAM. If you're doing any compositing or layering, that amount starts to compound on itself.
I'm running 128GB RAM in my Z820 - 16 core, 32 thread... So right on the 4GB per thread thing. I would actually like MORE. I did run briefly with 256GB, but had to run Win8 to make use of it all. Not a fan of Windows 8 on this system yet. That and the 16GB module performance, or at least of the modules I tried, was lacking compared to what the 8GB modules were giving me.
Too bad about your Lian-Li experience. I would be curious if it was the case itself or if you used one of their bundled power supplies? Support from most any of the case manufacturers is always very lacking or even non-existant. Best luck I've had for actual support has been from Chieftec and Thermaltake. Enermax isn't too bad.. Chieftec probably has the best support becuase they make tons of cases for a lot of the big PC companies. I think they were the primary case supplier for AlienWare before the Dell acquisition.
Last edited:
