PC Magazine Review - Performance Conclusion Right?

WillFastie wrote on 4/28/2003, 9:49 AM

I'm a bit confused by the performance results described in PC Magazine's review (http://www.pcmag.com/article2/0,4149,1017029,00.asp). I thought Vegas and Premiere had comparable performance, yet the review has Vegas doing rendering tasks at half the speed of Premiere. I also thought both products used the MainConcept encoder, which should level the MPEG-2 playing field.

It looks like the test case the reviewer used was pretty simple. I've seen some posts here that suggest that complicated timelines require more rendering times. Was the reviewer's use of a chroma key a factor?

Can anyone shed any light on this? I just switched to Vegas from Premiere and hope I didn't dig myself into a hole.

Back to post

Comments

DDogg wrote on 4/30/2003, 9:54 AM

Your post is what got me thinking. As you said, encoding to the MC engine internally should be (seems like it oughta be) faster than encoding via a frameserver. That just makes the question of internal pre-filtering or other "pre" actions stand up like Johnny boy at school waving his hand. :-)

/edit: Removed stuff I think was off-topic

elCutty wrote on 4/30/2003, 10:08 AM

I'm not sure if this happened to be the same with V4.0 - in V4.0b the MPEG luminance range is reduced and this is not caused by the MC encoder. May this cause the performance impact?

WillFastie wrote on 4/30/2003, 2:01 PM

CPU: Pentium 4 2.26GHz, 1GB DDR SDRAM, 533MHz FSB
HDD: 2x120GB, 7200rpm, 8MB cache hard drives, straight IDE (no RAID)
Both hard drives on IDE Channel 0; both optical drives on IDE Channel 1
All programs on drive C:
All video (captures and output) and video projects on drive D:
ODD: Sony DRU120A (DVD+R/RW & CD-R/RW combo), Creative PC-DVD 12x (DVD-ROM)
OS: Windows XP Home Edition

This system is dedicated to digital photography and video. No games and no funny software. A very clean system. There are multiple NLEs installed, and several have been installed and then removed. I may have codec creep, but I haven't noticed any problems. Thre isn't even an anti-virus program, although that is coming.

WillFastie wrote on 4/30/2003, 2:20 PM

Yes, I have suggested to the author that he come up with a simple baseline test that can be applied to any NLE regardless of sophistication. Then he can use his "torture" test on NLEs that support the required features.

I doubt he can afford the time to regress the projects as you suggest. He's a freelancer, not a PC Mag lab guy with time to burn.

I've started to do the regression but haven't been able to scare up enough time to do everything. However, Sonic should know very well what kind of timeline stuff eats up rendering time.

DDogg wrote on 4/30/2003, 2:44 PM

"However, Sonic should know very well what kind of timeline stuff eats up rendering time." heh, ... well don't hold your breath ... :-/

d1editor wrote on 4/30/2003, 3:36 PM

"HDD: 2x120GB, 7200rpm, 8MB cache hard drives, straight IDE (no RAID)"

Your system would benefit with a raid and some additional RAM. The stripped raid will outperform the current configuration....

AS IDE HARD DRIVE manufacturers squeeze more and more storage capacity onto new drives, they're hacking the warranty coverage for standard drives down to one year. You get more data to use, but manufacturers seem less and less willing to guarantee the integrity and safety of all those extra bytes. RAID can help you take back some of that reliability, but that's not all. A RAID array can also dramatically increase your overall hard disk performance. In some cases, IDE RAID can even offer you the best of both worlds: redundancy to protect against drive failure and better overall storage performance to pry open the bottleneck.

You might want to look into the raid to increase performance....

WillFastie wrote on 4/30/2003, 5:27 PM

Thanks. I'm well versed in RAID and do plan to consider it going forward. I also don't think the change in warranty is based on a plan to drop reliability. Instead, I think the manufacturers want to prevent what I did last year. I returned a 4GB drive that had died one month short of the end of the warranty at a time when typical drive sizes were 60+. If I were a manufacturer, I would not want to have a stock of 4GB drives for three years just to service the small number of failures that will show up. By the way, that drive is the only one that failed me since the 1984 30MB IBM AT hard drive fiasco.

But that's wandering off topic. The issue with configuration can only be whether I have something ridiculously obvious. Other than that, I'm looking at this as a software issue. Premiere vs. Vegas on the same hardware. Which renders faster? What does one have to do, in software, to get the best rendering performance? What, if anything, did the PC Mag reviewer get wrong?

I want my rendering performance to be good within the constraints of my hardware. If I can learn what was wrong, if anything, I can make sure I don't fall into the same trap.

elCutty wrote on 5/1/2003, 1:29 AM

Once again the question to Sonic, will we see an improvement on the internal MPEG encoding with the 4.0c? V4 must do a lot of internal 'work' to double the encoding time.