Jason Van patten
Forum Replies Created
-
[Dennis Radeke] ” I’ve been meaning to write up a quick post on this for many a month and this thread prompted me to finish it up in my spare moments.”
It basically re-iterates what I wrote in my first reply in this thread. Just in, perhaps, an easier-to-swallow format. 😉
-
[Samuel Frazier] “SO, was hoping to spend around $1,200 if GPU won’t matter for me and $1,600 or a bit more if the GPU is crucial.”
Are you trying to stick with the Mac platform, or do you see yourself moving to Windows? Given your budget, I’m assuming the latter. You’re not going to get much of a Mac for that kind of money.
Assuming Windows, do you expect to assemble the system yourself, or are you looking for something pre-built? The latter is going to be challenging with that kind of budget, but not completely out of the question. I know there’s a well-loved and highly-regarded system builder that frequents this board and the Adobe forums that might be able to help you out. I have no experience with him since I’m a Mac guy (mostly). But if the PC platform is your next step, consider talking to EricBowen.
-
[Dale Roberts] ” I do note tho that PP doesn’t really like long clips… scrubbing was VERY difficult. It froze most of the time… mind you that could be my Mac… it is a little old and slow!”
It’s not your necessarily your Mac, but a well-known CS6 bug with spanned AVCHD clips. Adobe has announced that there will be a fix for it in the coming weeks. (FWIW, neither CS5.5 nor CC suffer the same issue). Shorter, unspanned AVCHD clips don’t appear to cause a problem with CS6. Just the longer ones.
-
[Ronny Courtens] “No Jason, Oliver has used different GPUs that are good for CUDA as well as for OpenCL acceleration. Your card is perfect for CUDA, not good for OpenCL.”
*head slap* Never mind.
I’ll take “Missing the Point” for $600, Alex.
-
[Ronny Courtens] “Nope. We have had this discussion on another forum. The most important difference is that you did your tests using an upclocked nVidia GTX570 GPU that is great for CUDA but lousy for OpenCL.”
Read the whole thread Ronny. Specifically the OP’s machine specs. He’s using a single 5870 card in a MUCH SLOWER SYSTEM than mine. Do you think that ancient AMD card is enough to make that much of a difference? (Answer: it isn’t).
His results are so good because he didn’t transcode. Period.
[Ronny Courtens] “As another poster pointed out you are also running FCPX on a fast but older MacPro:”
So is the OP. Your point?
-
[Gary Huff] ” Exactly how?”
If I had to guess: they overhauled how the software uses OpenCL so that it can take advantage of multiple GPUs.
-
[Santiago MartÃ] ” Premiere Pro CC does support dual GPUs right now, it would be cool to see those benchmarks too. I couldn’t find any on the web.”
Actually, it supports n GPUs, not just duals. And it seems to work with and without SLI (nVidia, obv) but appears to scale linearly when using SLI. That testing was done w/Windows and 2 Titans, if I remember correctly; there’s a thread on the Adobe’s forums somewhere. Since OS X doesn’t support SLI right now, it’d have to be done sans it in a Mac Pro. But then again, it’d take a heavily modified Mac Pro with extra power supplies to power more than one card…
-
[Oliver Peters] “This might be affected by staying in ProRes, which is clearly best for X, but for which CC has also been tweaked.”
That’s a significant understatement, to be sure. Final Cut definitely prefers and works well with ProRes. The strength of the Adobe software is that it works fairly well with most anything. My own tests with the software have Premiere (CS5.5 and CC) handling my tasks in literally 1/2 the length of the clip. For FCPX? Twice the length of the clip. The main difference between my tests and yours: I’m not starting with, nor transcoding to/from ProRes. I’m starting with AVCHD and ending with h.264 MP4.
-
[Abraham Chaffin] “The forums have two features which are similar to what you are asking, but not exactly”
Thanks Abraham, you confirmed what I thought. I appreciate the reply.
The sorting and organization seems very Usenet-ish, which is interesting in a 20-year-old sorta way. 😉
-
[Dan Stewart] “Not trying to be a downer but the more I think about the tube the more I realise I’ll never use one – a second hand cheesegrater will make sense at launch, let alone a year or two in when those firepros are antiques”
If you’re editing in FCPX, it may make sense to consider another alternative, and that’s a Hack. The specific reason why I’m suggesting it is the availability of Intel’s AVX, as I’d mentioned previously. You’ll never have access to those on the ‘old’ Mac Pros. But any Intel chip Sandy Bridge and newer will have the extensions available.
FCPX can and does use those.
As I wrote on another FCPX forum: I did a homemade benchmark the other day with FCPX on 2 different Macs. One is a heavily modified Mac Pro 5,1 with 2 6-core x5690 (3.46GHz) Westmere chips, a flashed GTX570, 48G of RAM, and a bunch of media RAID’d together. The second: a stock 2012 Macbook Pro Retina with a 2.6GHz chip and 16G of RAM. The test: export my 20-minute AVCHD video out to 720p h.264 MP4. If t=20 minutes, the Mac Pro took 2t to complete the tasks. The laptop? It did it in time t. Literally half the time of the Mac Pro. And I’d accidentally disabled the nVidia GPU in the laptop during the test, as well.
From a pure CPU-brute perspective, the Mac Pro will crush the laptop. But clearly something was at play here that made the export in FCPX literally twice as fast as the big monster. I’m betting it was AVX.
Now, whether you do a Xeon version akin to the Mac Pros, or stick with the desktop chips is entirely up to you and your budget. Picking the right motherboard and other parts isn’t difficult any longer; most of that homework has already been done by other folks. The biggest factor at that point is: which video card? Thankfully, with a Hack you’re no longer locked into needing an EFI-capable one and can use any PC-based card for which drivers exist.