Mac Pro for Final Cut Pro X

Nigel Barker · December 10th, 2012, 12:32 PM

Virtual cores are not the equivalent of real cores they are just a way of better utilising the available CPU time by increasing the opportunities for parallelisation. Eight real cores will always outperform 4 real cores each with two virtual threads of execution all things being equal - which they probably aren't:-) The only way to really tell is to take the same operations & test them on both machines.

Craig Seeman · December 10th, 2012, 01:38 PM

The all things being equal but probably aren't is key.
Lots of programs haven't optimally used the cores. Virtual cores can help It depends on the programing.
You might want to compare your 2008 8 Core MacPro (which I also have) to the same functions using the same programs on a Ivy Bridge Quad Core i7. I'm certainly seeing a speed advantage on the i7 and my 2011 MBP is only Sandy Bridge I believe.

I don't think one can flatly claim more cores is better as chip architecture has changed over time. 8 Core Xeon 2008 Harpertown Xeon may not have a big advantage over 2012 Ivy Bridge Quad Core i7.

These test are now over a year old but still informative.
2011 Macs - running Pro Apps

Think about how well the 2012 15" MBP Retina performed in my previous posted link.

Look at these FCPX tests from still earlier.
Final Cut Pro X on three different Macs

Just Cinebench and Geekbench test of the various MacPros from 2008 to 2010.
Cinebench and Geekbench results for 2010 Mac Pros

It's hard to make direct comparisons but it seems the 2012 15" MBP Retina might be closer to the 2010 Westmere 6 core than 2008 Harpertown 8 core. Obviously there are factors relative to the program and GPU but I think Dave's perception that FCPX seems faster on than his MacPro really fits with all the descriptions on hearing.

I believe Barefeats will be running tests on the BTO 2012 27" iMacs when they arrive ... in January 2013 (maybe I shouldn't be calling it the 2012). I suspect he'll test it against a 2010 or 2012 MacPro Westmere 6 core at least.

Buck Bito · January 7th, 2013, 12:58 PM

Quote:

Originally Posted by Nigel Barker

... I just got a IBM LTO-5 drive cheap on eBay & put it into a HP Microserver ... have been using LTFS to dump files to tape. I copy files to a 1TB DISK on the Microserver then when that is full I write to tape...

Apologies for continuing to pull this thread away from the original topic, but I have been hoping to implement a very similar system with an HP Microserver and hope to pick some brains about OS and configuration. I'm not very handy linux-wise, but have been running a Lubuntu server on an ancient pentium3 box for a couple of years serving as a LAMP for our small company's intranet. I would like some help configuring a Microserver to act in the same capacity Nigel describes for our HD and 2K video files to be sent from Mac Pros running FCP and Windows7 workstations running CS6 and Resolve via our gigabit wired network to the Microserver for backup and layout to LTFS LTO5 or 6 tape. I've been stuck at choosing an OS for some time and could use some info from people that have it working. I currently have Scientific Linux on a boot SSD in the optical bay on an N36L with 4 x 1TB in a RAID 10 in the main bays, but have not yet done the work to try and get the workstations writing to it nor have I obtained the LTO or the HBA to feed it.

Nigel Barker · January 7th, 2013, 02:18 PM

I tried both Linux (CentOS) & Windows Server 2008 before settling on Windows 7 Pro as the most reliable & simplest OS & application to manage. Which version of LTFS you use depends on which tape drive you have as the IBM implementation won't work with HP drives & vice versa. It's all open source so I guess you could edit the code that checks which brand tape drive you have but it's easier just to take the pre-built binaries. The IBM Windows implementation is OK & you can just use Explorer & drag'n'drop but there is also a faster command line utility. The way I use it is to use rsync or similar from the Mac to the Windows share staging area on the Microserver (4 drives hardware RAID-1 for system drive & hardware RAID-0 for the staging drive). When the staging area is full I copy to tape. It's not very streamlined & it's not as fast as it could be but for archiving projects that probably will never be looked at again it's OK. If you wanted a better system you would use BRU or similar robust backup application but this works for me.

Buck Bito · January 7th, 2013, 04:36 PM

Quote:

Originally Posted by Nigel Barker

I tried both Linux (CentOS) & Windows Server 2008 before settling on Windows 7 Pro as the most reliable & simplest OS & application to manage.

Thanks for the reply Nigel, I didn't even think Win7 would run on the N36. I was hoping you were running a Linux distro and that I could benefit from your experience with that. Can you tell me what put you off CentOS - was it just being uncomfortable with the OS, or did you have specific problems working with the tape drive?
-Thanks again!

December 10th, 2012, 12:32 PM	#16
Nigel Barker Inner Circle Join Date: Feb 2009 Location: Norwich, Norfolk, UK Posts: 3,531	Re: Mac Pro for Final Cut Pro X Virtual cores are not the equivalent of real cores they are just a way of better utilising the available CPU time by increasing the opportunities for parallelisation. Eight real cores will always outperform 4 real cores each with two virtual threads of execution all things being equal - which they probably aren't:-) The only way to really tell is to take the same operations & test them on both machines.

December 10th, 2012, 01:38 PM	#17
Craig Seeman Inner Circle Join Date: Feb 2004 Location: Brooklyn, NY, USA Posts: 3,841	Re: Mac Pro for Final Cut Pro X The all things being equal but probably aren't is key. Lots of programs haven't optimally used the cores. Virtual cores can help It depends on the programing. You might want to compare your 2008 8 Core MacPro (which I also have) to the same functions using the same programs on a Ivy Bridge Quad Core i7. I'm certainly seeing a speed advantage on the i7 and my 2011 MBP is only Sandy Bridge I believe. I don't think one can flatly claim more cores is better as chip architecture has changed over time. 8 Core Xeon 2008 Harpertown Xeon may not have a big advantage over 2012 Ivy Bridge Quad Core i7. These test are now over a year old but still informative. 2011 Macs - running Pro Apps Think about how well the 2012 15" MBP Retina performed in my previous posted link. Look at these FCPX tests from still earlier. Final Cut Pro X on three different Macs Just Cinebench and Geekbench test of the various MacPros from 2008 to 2010. Cinebench and Geekbench results for 2010 Mac Pros It's hard to make direct comparisons but it seems the 2012 15" MBP Retina might be closer to the 2010 Westmere 6 core than 2008 Harpertown 8 core. Obviously there are factors relative to the program and GPU but I think Dave's perception that FCPX seems faster on than his MacPro really fits with all the descriptions on hearing. I believe Barefeats will be running tests on the BTO 2012 27" iMacs when they arrive ... in January 2013 (maybe I shouldn't be calling it the 2012). I suspect he'll test it against a 2010 or 2012 MacPro Westmere 6 core at least. Last edited by Craig Seeman; December 10th, 2012 at 04:00 PM.

January 7th, 2013, 02:18 PM	#19
Nigel Barker Inner Circle Join Date: Feb 2009 Location: Norwich, Norfolk, UK Posts: 3,531	Re: Mac Pro for Final Cut Pro X I tried both Linux (CentOS) & Windows Server 2008 before settling on Windows 7 Pro as the most reliable & simplest OS & application to manage. Which version of LTFS you use depends on which tape drive you have as the IBM implementation won't work with HP drives & vice versa. It's all open source so I guess you could edit the code that checks which brand tape drive you have but it's easier just to take the pre-built binaries. The IBM Windows implementation is OK & you can just use Explorer & drag'n'drop but there is also a faster command line utility. The way I use it is to use rsync or similar from the Mac to the Windows share staging area on the Microserver (4 drives hardware RAID-1 for system drive & hardware RAID-0 for the staging drive). When the staging area is full I copy to tape. It's not very streamlined & it's not as fast as it could be but for archiving projects that probably will never be looked at again it's OK. If you wanted a better system you would use BRU or similar robust backup application but this works for me.