How to Optimize Rendering from FCP?

C

craigom

Member
We are using FCP6 for editing our predominately HD footage (DVCPRO HD) from our HVX200 camera. We edit on a Mac Pro 8 core machine (2.8 GHz processors with 2GB of DDR2 RAM). Our installed graphics card was a ATI Radeon HD 2600 XT but it has recently failed so i am replacing it with a ATI Radeon HD 4870 1GB.

So my question is, how can we significantly improve our render times? Our work involves projects that require quick turn arounds. So i am unclear what is the main factor in determining Render times. Is it purely processor based i.e. no of cores (but even using the multi-processor function in AE and in Compressor we havent noticed huge gains). Will a faster graphics card with more graphics memory help? Also we are assessing upgrading to Snow Leopard and FCS3. Has anyone noticed any big gains with the new software?

We were also considering purchasing a Matrox accelerator card. Do these make a big boost for render times? Does any one have experience with something like Matrox CompressHD? Does it only help with H264 codec? What about if we are rendering to FLV? Do any of the other Matrox I/O solutions help with render times?

Sorry if they seem like stupid questions but i cant seem to find any clear answers on this.

Thanks

Craig

 
Render times for what renders?

What sequence settings?

What output?

What Matrox accelerator? CompressHD is the only one I am aware of and it deals with h.264 encoding, et al.
 
We have tired various settings using Compressor. I guess i am more interested in general terms what are the main factors affecting render times: No of Processors, Graphics Card, Software's ability to utilize multi-core, OS? Or is this too complicated an issue to boil it down to any one factor? We have an 8 core Mac pro but i dont feel we are getting much advantage from that in rendering. Render times between our dual core and 8 core, using same settings, are not huge.

If we were to upgrade our editing suite, where is it best to invest in? Better Graphics Card, a dedicated hardware accelerator, upgrade to Snow Leopard and FCS3?
 
sorry, failed to clarify. We typically render 5-6 minute HD clips (predominantly interviews) with some minimal on screen graphics and simple cross fade transitions. Outputted to high quality mov file using h264. We also at times import mov file into AE for some additional on screen fx and output to FLV. We are also looking for a better workflow for outputting to FLV from FCP, as the double render times kill us. We are looking at Automatic Duck and XML scripting solutions....
 
craigom said:
We have tired various settings using Compressor. I guess i am more interested in general terms what are the main factors affecting render times: No of Processors, Graphics Card, Software's ability to utilize multi-core, OS? Or is this too complicated an issue to boil it down to any one factor? We have an 8 core Mac pro but i dont feel we are getting much advantage from that in rendering. Render times between our dual core and 8 core, using same settings, are not huge.

If we were to upgrade our editing suite, where is it best to invest in? Better Graphics Card, a dedicated hardware accelerator, upgrade to Snow Leopard and FCS3?
Click to expand...
I like your post, as we have the exact same set-up as you. Only you have not specified a couple of details.
1) You're talking about the final rendering not graphics rendering (onto the screen). They're different things you see. I made the same card upgrade and experienced faster REDRAW (screen rendering).
2) You're right, FILE rendering efficiency is a function of all things taken together. The file type, render settings, and how you are using those 8 cores.
It's unfortunate that in all the amazing information on this, and many other forums related to HD, not one actually shows people's workflow diagramatically. Sad.
Lots of people experienced unimpressive performance with 8Core MacPros, untill they mastered the art of the virtual cluster. Indeed, Snow Leopard is supposed to speed these machines up natively. Have a read...
http://www.barefeats.com/octopro5.html
 
Thanks. I have spent quite a lot of time on these forums and i am struggling to find some clear information about how to get an optimized setup with our 8 core Mac Pro. And yes i am referring more to final renders than in-edit graphics rendering. I sense we are wasting all that processing power but im not so clear on how to tap into it. I will definitely research some more virtual clustering. So if anyone as a workflow which they believe OPTIMIZES their multicore editing machine, would be happy to see it.
 
you might be rigth and or u might just want blood out of a stone... how fast does it render now? I think if your system works, trying to gane 10 % speed for the chance of lets say messing up other settings..might not be worth it.. ..i'm just saying. I dont know
 
Even with FCP7 on SL with a Nehalem 8-core, you won't get renders to improve very much.

Background encoding, however, will significantly speed up with a quickcluster
 
You must log in or register to reply here.
Back
Top