Render time?

Comments

PeterWright wrote on 7/10/2005, 1:57 AM
Bringing up the old Render Test brought some interesting findings - the default settings of the rendertest.veg are NTSC/Best, and I found rendering to NTSC was considerably faster than PAL, and, with a Dual CPU machine, V6 around twice the speed of V5.

Results for Dual AMD 2000, 2 Gb RAM:

V5: NTSC 3.07 PAL 4.43
V6: NTSC 1.38 PAL 2.26
Bertie28 wrote on 7/10/2005, 4:55 AM
Yes,

Since Vegas 3 i was playing with everyday PC's like today's commercial ones with 1.8, 2.0, 3.0, dual3.0 with these ones we can only make home videos and jokes, but when it comes to deal with important clients and they are sitting just behind you watching the work is done for them in realtime and the main thing is a lots of $$$ they put in, man you need to be prepared for everything because one glitch or frozen screen or else and your clients will disappear faster than you can say "Holy Bunkers" :-)

I was not happy with commercial PCs and i lost a lots of important clients because of slow PC. For 1 project of 1.3hrs length they had to wait for two days.

So i asked my brother to build a special PC just for me and my Vegas...i actually said to him "look, i want you to make me a Beast PC and the money is no problem" and he just smiled and said OK.

After a week or so he came on to me with a bigwooden box and said "your beast is ready".

When he put it all together it was all hand made PC's in racks actually modular ones four in all and he stacked them together one by one and when he finished he started telling me what he actually did and what are these racks for.

Every rack has a dual 3.0Pentium, 2GB RAM, two 120GB HD and a big cooling fan it has more but i couldn't understand :-(

And a results are out of this World.

So, when i look back and see all these commercial PCs i say "Man, i have a machine"

I invested a lot of money but it was worth it, but then again it depends on what you do with Vegas and how much money you earn doing video editing.

Later,

Bertie,
Guy Bruner wrote on 7/10/2005, 5:56 AM
What operating system are you using that will tie that beast together? Are you using it basically for network rendering? I don't understand how your system would make Vegas faster than, say a single Pentium D or dual core Opteron.
GlennChan wrote on 7/10/2005, 11:18 AM
My machine did test render is 1:48.

Sherman, if you optimize your system you'd be getting close to 1:29. This is for a Prescott-core Pentium (faster than Canterwood at Vegas), 800FSB, RAM running in dual channel, no other programs taking up significant CPU cycles.
At best, you might be able to make your system 21% faster. Which won't really help you, when it's taking a day to render.

Things to try:
A- Check your hard drives are in DMA mode.
In device manager, get properties on each IDE/ATA controller
click advanced tab- it should not say PIO anywhere
B- Reboot your computer in safe mode, try rendertest.veg. If times speed up, go start --> run "msconfig" --> startup (tab)
google the names of things, and disable the ones that don't matter. If you're lazy, you're probably OK just unchecking everything- if you're missing anything, just re-enable it. This is not exactly safe though- you may accidentally deactivate something good like antivirus or firewall (and some firewalls, when deactivated, will stop your internet from working).
If you have a spyware/adware infestation (i.e. you get random popups and webpage redirections), those programs that startup will be harder to remove.
C- Check RAM: Reboot, and read your BIOS screen. On some BIOSes, you have to enable the POST messages that say whether you're single or dual channel.
Try F1-F12 and the DEL keys to enter BIOS.
To get RAM running dual channel, you need PAIRS of the EXACT same RAM in the right slots on your motherboard (they should be color-coded).

Those are the top 3 things to try I believe.

You can also upgrade your hardware, but that also only provides mild speed increases. You should hit like :78s in rendertest.veg with new hardware.

**Warning: The instructions above are really condensed because I'm too lazy to type. There's a small chance you can mess up your computer if you do the wrong thing. If you would like fuller instructions, please ask.
Bertie28 wrote on 7/13/2005, 7:54 AM
well, they are not networked so i don't do network render and regarding OS i run WinXpSP2 and it is installe only in one rack PC so three others have no OS at all.
But all together it works like a one PC a huge power plant :-)

Later,

Bertie,
gbugarin wrote on 7/20/2005, 5:52 PM

We just built a P4 2.8 Dual machine. 1 gig (512mbx2) Geil DDR2 667 memory, abit As8 motherboard, 2 sata HDD's, with an ATI FireGL V3100 video card.

Rendered out using the rendertest to NTSC AVI and it finished in 1:22.

I'll do the test on the old computer in a few days! The old computer is a P4 3ghz machine with 1 gig (245mbx4) mixed Corsair/Buffalo ram, abit motherboard, 2 pata HDD's.... forgot what kind of video card it has.

gbugarin wrote on 7/20/2005, 5:56 PM
We are still using Vegas 5d. BTW, I used the old rendertest.

I ran the render test a second time and got 1:21.

Ran it a third time and got 1:21.

------------------

Just ran the old rendertest on the P4 3ghz machine and it took 1:21 to render.

No difference.
ForumAdmin wrote on 7/21/2005, 6:19 AM
While it is a valid .veg file, the Vasst render test is not in my opinion THE definitive benchmarking test for Vegas even though we see it being treated as such in many threads. What would be interesting, if you have the time, is to get some results rendering with Vegas 6 (and 5 if you have it) on your various machines using the sample projects found here:

http://www.sonymediasoftware.com/download/step2.asp?DID=496

These projects were created in Vegas 5 in February of '04 for distribution on the install CD, so hopefully nobody will think they were contrived to show off any of the rendering changes in Vegas 6.

Note: The biggest difference in V5/V6 render times will show up on true dual proc systems; you probably won't see any or much difference with single procs, HTs maybe a little.
JJKizak wrote on 7/21/2005, 6:50 AM
Using the Vasst render test in V6.0b, P4 3.4, 800 fsb, Hyperthreading on, 2 gig ram, 4 threads, 16 meg ram setting, mpeg2, DVD-NTSC, quality 31, all others at default. Time was 1:41 with processor temp at 144F, 100%.

JJK
GlennChan wrote on 7/21/2005, 10:50 AM
While there could probably be a more definitive render test that gets closer at approximating real world render times (for your projects), I think rendertest.veg is quite useful because it's the only Vegas benchmark for which there is a large pool of results. It's useful if you want to check if your system is performing as it should.
jlafferty wrote on 7/21/2005, 11:37 AM
"Note: The biggest difference in V5/V6 render times will show up on true dual proc systems; you probably won't see any or much difference with single procs, HTs maybe a little."

What can be said of optimizations for the new dual core processors? Are these reliant on what XP64 can allow, or can and will Vegas take more full advantage of them in the near future?

- jim
PeterWright wrote on 7/21/2005, 10:34 PM
Okay - did some renders of 4 of the sample projects. Must say it's not an easy exercise - if you don't happen to be watching the screen when a render finishes, the information you're waiting for is gone! Especially annoying with V5 when it told me there were 4 minutes to go - I glanced away and it finished!!

Anyway, with my Dual AMD 2000 machine, V6 took generally about 54% of the time that V5 did.

Results:

Project_______________V5 time__________V6 time

24p widescreen_______1 min 23 sec_________0.55sec
cube with shadow_____8 min 03 sec______4min 22 sec
curtis photo_________11 min 25 sec______6 min 07 sec
rack focus___________7 min 29 sec______3 min 55 sec