Dual Processor Sidebar in Emedia Mag

DataMeister wrote on 10/20/2003, 3:48 PM
I thought this was an interesting side bar in the October issue of Emedia Magazine. The "Processing DV Production" article has a sidebar on page 24 that covered some tests they performed between all the big name apps witha multiprocessor computer. I don't know what Premiere did different but it should give the Sony Programers something new to shoot for now.

Here's the web based article without all the cool graphs and pictures. Scroll to the bottom for the sidebar.
Page 3 of web article Processing DV Production

JBJones

Comments

Jsnkc wrote on 10/20/2003, 4:08 PM
The first thing I noticed when I went to that page was a big ad for Adobe right on the top. Possibly a Biased website?? maybe takin a little extra cash on the side to make premiere look a little better?

I never believe anything I read in these "render tests" The results are always completely diffrent and seem very biased.

johnmeyer wrote on 10/20/2003, 4:23 PM
I never believe anything I read in these "render tests"

Good point. However, I give Emedia top marks for actually doing scientific tests instead of just publishing uninformed opinions by someone that uses the software for a few hours and then writes a report.

While I am as suspicious as you about bias in these tests (because of Adobe's advertising clout), I am also suspicious about the conspicuous silence from Sony. If my product had been trashed, and the test was rigged or biased, I'd sure as heck post a response on my web site, and I'd write to the editor of the magazine and request (demand) a clarification in the next issue. I did this all the time when I ran software companies back in the 1980's and the editors of PC Mag, etc. were always very responsive. The fact that Sony hasn't done this makes me believe that the results correctly reflect reality.
TheHappyFriar wrote on 10/20/2003, 6:45 PM
A couple things I found suspecious:

1) No mention of what it was rendered to (DV, PTT, DVD mpeg-2, etc).

2) No metion of what they used in these rendered tests. Did they just take a 30 minute DV clip? Did they do cuts, added effects, etc? I've used Vegas to render a captured 30 minute mpeg-2 (all i frames) to a 5,000,000 mpeg-2 for use on a broadcast Mpeg-2 player in about real time. On top of that, I had 2 copies of Vegas 4 open, both doing 2 seperate renders, and the combined time was 45 minutes to finish.

I used an AMD-XP 2600 for that. Now, if it's a straight file render (no cuts, etc), that would say that an Intel Duel Xeon setup is slower then a single AMD-XP 2600 (which should be wrong).

Also, they mention that if you added RT hardware to Premiere, it would make things even faster. But... last I remember, hitting the CTRL+ALT+DEL while using a Matrox RT card had a good chance of crashing your computer (so no way to see if it's being efficient).

Of course what no review of Vegas (any version) ever mentioned is that you can run multiple copies if Vegas at once, while you can't with other NLE's. So... If it takes you 2:30 to render a 30 minute clip, you're spending that 2:30 working on your next project instead of sitting down wasting time for 40 minutes waiting for a render.


On a side note, it seems that the NLE world is behind the times in hardware usage. SGI has been using openGL for years (along with game companies and 3d companies): not much NLE support for OpenGL effect/renders. Also, id software started supporting duel processors on Quake 3 back in '99. So it only took the profesional NLE world 4 years to catch to the game companies! :)


edit: added a few minute later

One Page 1 they do say what you are working with and what you are rendering to. I'm just wondering one thing: Who was the poor sap that had to build each project in each NLE! You can't copy/past them after all! :)

since they say that they are rendering to DVD Mpeg-2, then the rendering would deal with the encoder NOT with the NLE. Since they say you are rendering for DVD (no mention if they used Vegas+DVD and premiere Pro+Encore in their budgets), we will assume they are not spending extra $$ on the DVD packages for these programs.

The base price of Vegas (as on Sony site) is $560 (will not cost more then this).
Base price for PPro (from Adobe site) is $700 (will not cost more then this).

If you bought Vegas and the lastest Mainconcept encoder (v1.4 @ $135), the total would be the same as just PPro. So, you would get the SAME render time as Premiere Pro if you spent the same amount of on Vegas + Encoder.

farss wrote on 10/20/2003, 9:32 PM
If you bought the standalone MC encoder you have to render to AVI first and then use the the MC encoder. You could get around this with frameserver but the issue will be how fast VV serves the frames.

The fastest thing I've come accross for rendering is MGI Videwave, sure it doesn't do much, about straight cuts and dissolves are all it can manage and forget about doing anything to the audio but for segments that are unaltered it goes into Smart DV Copy mode and is only limited by HD speed.

I only mention this to demonstrate that render speed is not only a function of the number of things you have applied on the time line but also the number of things that have to be checked to see that they HAVEN'T been applied.