I am about to begin a project which is being shot at 1080p29.97 XDCAM HD 50, and am becoming concerned about our systems ability to handle editing the format.

I was wonderinig if anyone had any recommendations as to desirable system specs and setup.

It will be edited on a Mac Pro Dual Core 2.66 Xeon with 4Gb of ram and a composite Mojo. A Quad PPC 2.5 is also a possibility.

In terms of SW I had in mind MC 3.0.5, OSX 10.5.4 and QT 7.4.5 (as recommended in the MC readme)

Finally, although is suspect it is of no consequence, the footage will be stored on Editshare shared storage, and distributed via a 10Gbe router but using the gigabit ethernet port on the mac.

With the test footage I have been given, I havent been overwhelmed by the performance of this machine. Short of buying a new 8core MP, what options do I have to optimise performance. More RAM?  Should I be using 10.4.11? Does cutting in 30i (so that the comp mojo displays) have a performance hit? How much does the GPU impact on performance? Also re. RAM, I'm of the understanding that having 2x2GB is preferable to having 4x1GB. Seeing as the latter is the case at present, how susbstantial is the upgrade.

The other thing that is a possibililty is to transcode via AMA in MC3.5 to DNxHD 145, the rationale being that DNxHD felt as though it were less demanding on the machine, although the increase in data rate is an unfortunate sacrifice to have to make, along with greater ingest times.

One final query I have is regarding advice in the XDCAM pathway guide on the avid site. It suggests that if heavy graphics work is to be done on the footage, transcoding to DNxHD is advisable as it stands up better. At what point do people think this really applies, as there will be a reasonable amount of GFX work done on this.

Any help would be appreciated, and apologies for the rather broad and vast number of questions.

Thanks!

Murray

I'm only speculating, but with XDCAM HD I'd suspect that CPU would be the bottleneck, generally speaking lower-bitrate more complex compression methods are more processor intensive.

Converting to DNxHD would probably provide for a better editing experience, so could be worth doing. You can test that now by simply transcoding the XD50 material to DNxHD and seeing if that performs better. If it does, then yeah, the AMA procress will probably be a faster ingest.