- Posts: 704
- Joined: Wed Aug 22, 2012 4:23 am
- Location: Melbourne, Australia
Hi Everyone,
We shot our very first short film on the Blackmagic Cinema Camera over the weekend. Massive thanks to the amazingly supportive John Brawley for loaning us his baby, Blackmagic Design and to Cail from Inspiration Studios for all of his help and support!
We shot DNG - and yesterday I started transcoding all the rushes (1.33TB's worth) to ProRes LT for the offline on my iMac. Apart from a minor bug, everything worked perfectly - but it was really slow.
After reading on John's blog that he was getting "pretty close to real time" renders on his MacBook Pro, I decided to do some testing.
On my MacBook Pro (15-inch, Late 2011, Model 8,2, OS X 10.8, 2.5GHz Intel Core i7, 8GB RAM, AMD Radeon HD 6770M 1024MB, Apple SSD TS256C), I can get about 6-12fps. I don't have a Thunderbolt drive with me - so these files are just running off the laptop's internal SSD (about 160 MB/s write - 203.1 MB/s read), so I'd imagine maybe with an external Thunderbolt drive I may even get better results.
However on my iMac (27-inch, Mid 2011, Model 12,2, OS X 10.8.1, 3.4GHz i7, 32GB RAM, AMD Radeon HD 6970M 2048MB, Apple SSD TS256C), with all the media stored on a Promise Pegasus R6 Thunderbolt RAID (I get about 505.2 MB/s write and 481.5 MB/s read), I can only get about 1-6fps.
As you can imagine - this surprises me! Why does the MacBook Pro out perform the iMac?
I'm am going to do some tests on a MacPro (2 x 2.93 GHz 6-Core Intel Xeon, 26GB 1333 MHz DDR3 RAM, ATI Radeon HD 5770, NVIDIA Quadro 4000) running Resolve 8 (we’re waiting for the "shipping" release of Resolve 9 before upgrading) later today to see how it goes.
Is anyone else seeing the same results? Have Blackmagic optimised their code for the MacBook Pro, or am I missing something really simple? These results really intrigue me, as both the iMac and MacBook Pro are using AMD Radeon mobile graphics cards.
What surprises me even more is that as far as I know, Resolve does all the DNG processing on the CPU – it only uses the GPU if you do any colour correction - is that correct? All compressed files, including Apple ProRes and RED R3D, are decompressed by the CPU prior to grading by the GPU – so I’m assuming BMC DNGs are the same?
Can anyone shed any light on these findings?
Thanks in advance...
Best Regards, Chris!
We shot our very first short film on the Blackmagic Cinema Camera over the weekend. Massive thanks to the amazingly supportive John Brawley for loaning us his baby, Blackmagic Design and to Cail from Inspiration Studios for all of his help and support!
We shot DNG - and yesterday I started transcoding all the rushes (1.33TB's worth) to ProRes LT for the offline on my iMac. Apart from a minor bug, everything worked perfectly - but it was really slow.
After reading on John's blog that he was getting "pretty close to real time" renders on his MacBook Pro, I decided to do some testing.
On my MacBook Pro (15-inch, Late 2011, Model 8,2, OS X 10.8, 2.5GHz Intel Core i7, 8GB RAM, AMD Radeon HD 6770M 1024MB, Apple SSD TS256C), I can get about 6-12fps. I don't have a Thunderbolt drive with me - so these files are just running off the laptop's internal SSD (about 160 MB/s write - 203.1 MB/s read), so I'd imagine maybe with an external Thunderbolt drive I may even get better results.
However on my iMac (27-inch, Mid 2011, Model 12,2, OS X 10.8.1, 3.4GHz i7, 32GB RAM, AMD Radeon HD 6970M 2048MB, Apple SSD TS256C), with all the media stored on a Promise Pegasus R6 Thunderbolt RAID (I get about 505.2 MB/s write and 481.5 MB/s read), I can only get about 1-6fps.
As you can imagine - this surprises me! Why does the MacBook Pro out perform the iMac?
I'm am going to do some tests on a MacPro (2 x 2.93 GHz 6-Core Intel Xeon, 26GB 1333 MHz DDR3 RAM, ATI Radeon HD 5770, NVIDIA Quadro 4000) running Resolve 8 (we’re waiting for the "shipping" release of Resolve 9 before upgrading) later today to see how it goes.
Is anyone else seeing the same results? Have Blackmagic optimised their code for the MacBook Pro, or am I missing something really simple? These results really intrigue me, as both the iMac and MacBook Pro are using AMD Radeon mobile graphics cards.
What surprises me even more is that as far as I know, Resolve does all the DNG processing on the CPU – it only uses the GPU if you do any colour correction - is that correct? All compressed files, including Apple ProRes and RED R3D, are decompressed by the CPU prior to grading by the GPU – so I’m assuming BMC DNGs are the same?
Can anyone shed any light on these findings?
Thanks in advance...
Best Regards, Chris!