Posted Friday, July 8th, 2011 by rob-ART morgan, mad scientist
Updated July 12th, 2011, with 2011 (13") MacBook Pro 2.7GHz results
In this first pass at evaluating Final Cut Pro X (FCPX), we decided to compare three different popular Macs. We chose three initial tests. Our test project was a 32 second ProRes 422 (HQ) video, 1888x1062, 23.98 FPS, 172.22Mb/s data rate (with Surround audio 48KHz).
DIRECTIONAL BLUR EFFECT
With Background Render disabled, we applied the effect to all eight clips. Then with a stopwatch, we timed how long it took to render the effect in all eight clips. (Time is in Seconds. Shorter bar means faster.)
SHARPEN BLUR EFFECT
With Background Render disabled, we applied the effect to all eight clips. Then with a stopwatch, we timed how long it took to render the effect in all clips. (Time is in Seconds. Shorter bar means faster.)
EXPORT ProRes 422 (HQ) to H.264
Selecting all eight clips, we exported (transcoded) to H.264 1920x1080, Millions, 23.98 FPS, 12.06 Mb/s data rate. (Time is in Seconds. Shorter bar means faster.)
BONUS! COMPRESSOR 4 - CREATE BLU-RAY STREAM
Using our ProRes 422 (HQ) sample, we used Compressor 4 to encode it as a H.264 elementary stream for Blu-ray and AVCHD authoring. (Time is in Seconds. Shorter bar means faster.)
LEGEND of GRAPHS
MP333 = 2010 Mac Pro 3.33GHz 6-core Westmere with 24G of ECC DDR3 1333MHz RAM and Radeon HD 5870 GPU (1G GDDR5)
iMac340 = 2011 iMac 3.4GHz Quad-Core i7 with 16G of DDR3 1333MHz RAM and Radeon HD 6970M GPU (2G GDDR5)
MBP230 = 2011 (17") MacBook Pro 2.3GHz Quad-Core i7 with 8G of DDR3 1333MHz RAM and Radeon HD 6750M GPU (1G GDDR5)
MBP270 = 2011 (13") MacBook Pro 2.7GHz Dual-Core i7 with 8G of DDR3 1333MHz RAM and Intel HD 3000 Integrated GPU (384M DDR3 shared with main memory).
MBA180 = 2011 (13") MacBook Air 1.8GHz Dual-Core i7 with 4G of DDR3 1333Mhz RAM and Intel HD 3000 Integrated GPU (384M DDR3 shared with main memory).
mini250 = 2011 Mac mini 2.5GHz Dual-Core i5 with 4G of DDR3 1333MHz RAM and AMD Radeon HD 6630M GPU (256MB of GDDR5 memory)
The big surprise is how close the iMac and top MacBook Pro came to the Mac Pro in Final Cut Pro X performance. Of course the tests we chose and the sample video footage may have a role in that.
On the other hand, the 13" MacBook Pro, 13" MacBook Air, and Mac mini -- each with only dual-core CPUs, took at least twice as long to complete the tasks. If you are going to do serious work, you need a serious Mac.
SYSTEM RESOURCES USED
1. MEMORY - The highest amount of real memory used by Final Cut Pro X during our rendering and exporting was 2G. Contrast that with the real memory use of over 9GB we saw with a 600 frame Motion 5 playback/render test. However, when we opened a second project with similar assets, the real memory use by FCPX doubled. That implies that importing HD video clips of longer duration than our 30 second sample would impact real memory as well.
2. CPU - We saw as high as 800% cpu usage on the Mac Pro (out of 1200% possible with Hyper-Threading). The quad-core iMac and quad-core MacBook Pro topped out at 650% (out of 800% possible with Hyper-Threading).
3. GPU - We observed 300+MB of VRAM in use while rendering blur effects. And according to OpenGL Driver Monitor, the CPU had to constantly wait for the GPU. To better illustrate the GPU effect, we plan to render the Blur Effects using various Mac Pro GPUs like we did with our Motion 5 testing. Stay tuned for those results.
4. STORAGE - The kind of storage used had little effect on the times. For example, we Exported to both HDD and SSD on the Mac Pro. The times were within a 1/4 second of each other.
I recommend Silverado's White Paper on Final Cut Pro X explaining how it utilizes memory, CPU cores, and OpenCL GPUs.
See our playback and RAM preview testing with Motion 5 and seven different GPUs.
1. Since we have confirmed that the GPU is active and VRAM is in use during the rendering of Blur Effects, we plan to retest the Mac Pro using at least five different GPUs.
2. We will attempt to compare Final Cut Pro 7 to Final Cut Pro X.