View Full Version : Faster renders: Our MacPro comparison


Scott Shuster
November 23rd, 2006, 04:08 PM
How much faster can you render FCP in a MacPro?

Of course it depends completely on what you put in that MacPro (processor(s), memory, speed-of-drive connection), and what you had in (and connected to) your former render environment.

But anyway here is a direct comparison for you to mull over. Has anyone seen other comparisons directly specific to renders in Final Cut Pro?

Here is ours:

Old environment:
G5 - 2x2.5ghz (single) processors
2GB of memory

New environment:
MacPro Tower - Intel Core Duo 3ghz
8GB of memory

Both machines are running FCP v5.1.2 ("universal") and both are getting the video from external drives via Firewire 400. Yeah, I know, we should be using Firewire 800 or eSata, both of which we have, but for reasons I won’t bore you with, we are using Firewire 400 today.

:::(Drum Roll):::

A certain specific 720/30p render in the G5 takes: FOUR HOURS

That same render in the MacPro?
:::(Cymbal Crash!):::
TWO HOURS!!

So in our case anyway, our investment in the MacPro is reducing our render times by half - ! WOW!!

Terence Murphy
November 24th, 2006, 10:50 AM
What was the CPU utilization on both systems during render? Are the MacPro's CPU's saturated (running at 100%), or is the performance constrained because of a file transfer bottleneck with the Firewire 400 drive?

-Terence

Scott Shuster
November 24th, 2006, 12:03 PM
That's a really good question Terence - I know I could look at that in Activity Monitor, but it's hard for me to muscle my way onto the desktops of my artistic staff and undertake that kind of measurement. They will hate opening any other app while a render is going on. The only way I got the data above was due to the enthusiasm of an editor who basically pointed it out to me kinda as a surprise. So I don't think I can get that info for you, at least not right away.

And Firewire 400 will be but a memory here by the end of next week. It's just a really bizarre situation that has us briefly using FW400 - I won't bore you with it - it was just a totally ridiculous random thing.

We're all over eSata here, not without problems, but that's for another post. There is, of course, no question that eSata makes an enormous difference. The copying speed is incredible. But then we had some pixelization-on-playback issues...not sure where the blame lies yet, maybe the card, maybe the cable. Playback with FW400 was fine, while playback of the same file with eSata was pixelized. I don't know enough to make any definitive statements on that one, but there you have a little sneak preview of another problem I may be discussing here soon. I can tell you this: It was VERY disappointing to run into that problem right after making the big eSata investment...

Another interesting issue will be to what degree the 8GB of memory are making a difference -- and to what degree maxing out the machine at 16GB of memory would make a further difference. I have seen ONE report about number crunching with the not-yet-released multiple-core chips that said all 8 bars of 2GB memory were merrily working, albeit briefly. The reporter noted that he found it difficult to come up with a task big enough to place a load on all 16GB of memory, but when he did, they all got into the act: Good news. Kinda funny that he had a tough time even coming up with a job big enough! It's true: You can do ICBM trajectories and pin-point your ten-year flights of interplanetary probes with these MacPros!

The answers to all of this may be interesting to know in detail, but the biggest excitement and joy around here is when the editors say "wow, this is SO FAST!" That's music to my ears. As far as EXACTLY how fast it is, while I would love to know I'm not gonna interrupt these high cost people to find out.