[Steve Modica]

I’m testing over here as well.
I have proven that I can replace the Movie directory with a symlink and put it elsewhere. (I do this at home with iMovie!) I ran out of time today but I plan on proving that multiple systems can edit from the same events.

Ideally, I think I’d prefer that only “Events” were shared rather than projects. I was able to put a project on a stick and move it around. So long as the events are consistent between machines, I think it should just work.

Steve

Steve Modica
CTO, Small Tree Communications

[Steve Modica]

To “share” project files, you would either move or duplicate the project (you can copy everything, just used clips, or the project file alone) and then move it to the second system.

When you import from shared media, it creates symbolic links, so I expect you could even copies those and import the whole thing.

Steve

Steve Modica
CTO, Small Tree Communications

[Steve Modica]

Gosh, I’m late to the party.
I like Small Tree, but that’s just me 🙂

Steve Modica
CTO, Small Tree Communications

[Steve Modica]

I did a bunch of testing yesterday.
It will work with shared storage. In fact, it imports very quickly because it creates symbolic links to the files. If you don’t set the special “don’t move everything to events folders” flag, it starts copying everything local in the background.

All in all, it was really good. Our FCP guy was testing and editing a project (H.264 no less) right away. Performance was good. I was able to have 4 or 5 streams going with 8-10 audio and work with no trouble. It renders automagically in the background, but pauses that during playback. I could force it on and it would play at a lower frame rate. I could enable dropped frames and force some drops.

I was doing all my tests on a non-jumboframe imac. So that was worst case.

A *huge* bright spot is that this forces “good” sharing behavior. You can share your media files, but projects will be local.

Steve Modica
CTO, Small Tree Communications

[Steve Modica]

[Walter Soyka] “Like watching a puppy chase his own tail?”

Yes… but it wasn’t cute, and I was standing there with a bunch of geeks in a cold computer room watching a machine that looked hung. So I guess it was nothing like watching a puppy 🙂

Steve Modica
CTO, Small Tree Communications

[Steve Modica]

[Alexander Papadopoulos] “I dont think that it has more work to do. When the same work is done on a pc the time remains constant! Its after fx on the mac causing the trouble!. It seam to accumulate each render time. Even if I stop the render and start over it will pick of from the last render time. The only way I found so far is to close the application and restart it. But that doesn’t work for me if I have to do that 60 times 🙁

Posts Index”

Have you tried bringing up activity monitor and looking at a memory map of the process? You might find that it’s growing in memory and perhaps it’s running up against some limits. You could compare before and after and see what’s growing. (the purge cache idea sounds good to me)

Steve Modica
CTO, Small Tree Communications

[Steve Modica]

[Alex Gerulaitis] “The hybrid drive is caching – and it may work wonders for non-streaming apps – and maybe even for some streaming ones. Hard to say until I test one.”

agreed

Steve Modica
CTO, Small Tree Communications

[Steve Modica]

[Michael Locke] “I’m realizing most “faster, better” features with computer hardware is aimed at “Tour of Duty” players than swapping 200G of prores around. Love to see an AJA test on one, say in a VR Mini”

I’d like to see a latency graph over a long sequential read. That would be the telling thing. What happens when it crosses over? How does it handle migration?

I have a story:
Many years ago when SGI designed NUMA, they wanted to be smart and “migrate” memory around the machine to be near the executing processes. They also wanted to “migrate” threads around the machine to be near resources. So imagine our joy when we got to watch our processes and memory race around after each other. That feature got turned off forever.
How is this relevant? It’s not… but it’s funny 🙂
Steve

Steve Modica
CTO, Small Tree Communications

[Steve Modica]

I have not tried these at all.
For raid based shared editing, my take is that the streaming reads are large enough that the SSD space won’t be helpful because we need to read too much data in realtime. When we hit the block storage, we’ll drop unless it’s very fast.

My personal choice was that having no moving parts far outweighed the additional cost and loss of storage. It’s lighter and quieter and uses less power, so I went full SSD for my laptop.

Any time there’s a new technology like this, that tries to mung together the best of both worlds, I get leery.

Steve Modica
CTO, Small Tree Communications

[Steve Modica]

When you fire it up and it starts running, look at system activity reporter and see what kind of throughput you’re getting. Check again the next time and compare. If it’s not slower, than there must be more work to do for some reason.

Steve Modica
CTO, Small Tree Communications