PDA

View Full Version : Render me Senseless!



gosmond
05-08-2003, 08:03 PM
OK, so I did the boyscout thing and got me a very nice Intel board, dual processor P4 2.4GB Xeon, 1GB mem, 533 FSB, 73 GB System SCSI, 73 GB SCSI for sound/etc, 200GB Video SCSI array, SX-8, Dual 17" LCD screens and a couple of external IDE 120GB drives w/USB 2.0. Unfortunately, I'm using a Matrox G-550 right now while I apply the electric paddles to my wallet. I have noticed a few things.

Toaster does indeed rock! I'm still giggling...

I notice my CPU's while rendering are sitting at about 50% or less and my memory is under utilized if anything?

My hard drives are laughing waiting for more to do.

So.... will a better video card make my rendering blues go away? I was hoping TV[2] would suck the life out of my system in return for a blazing render? Any real nice cards out there that will make a difference in the render process, or is there some bottleneck I'm not aware of? Go ahead, I can take it....arrrrggh!

ScorpioProd
05-08-2003, 11:09 PM
You may just be seeing the limits of real-time.

For instance, on my dual-2.8GHz T[2] system, I also hit about 50% CPU on both as a max when rendering to DV or M-JPEG, BUT, the rendering is effectively happening in REAL-TIME. So, it can't get any faster. T[2] is limited to no faster than real-time.

OTOH, when I do my MPEG-2 encoding in TMPGEnc, if I'm using CBR I get 1.5 times FASTER than real-time, and my CPUs are both pegged at 100%.

It's like Toaster can only travel the speed of light, but TMPGEnc has a faster-than-lightspeed drive. :)

SBowie
05-09-2003, 06:25 AM
"Rendering? What's rendering? We don't need no steen-king rendering..." :-p

gosmond
05-11-2003, 09:16 AM
Originally posted by ScorpioProd
You may just be seeing the limits of real-time.
It's like Toaster can only travel the speed of light, but TMPGEnc has a faster-than-lightspeed drive. :)

Ah! Realtime... There does seem to be a dependency there. No doubt, the programmers have already decided this is where is stops for now? But in the future, no reason we can't see a render as fast as the system will allow.

Interesting note about the MPGE. Simpler algorythm for compression? At any rate, I'm able to render in the background and type this message realtime. Good enough for me!


Thanks! :cool:

Gordon
05-15-2003, 09:56 PM
If the CPU's are pegged at 50% it is because the application is not multi-threaded to take full advantage of both CPUs. We are not speaking of VT[x] here but of the DV, or M-JPeg software codecs themselves.

The reason that TMPGEnc peggs both CPUs at 100% is because it's codecs are written to take advantage of both CPUs and they mention that fact in the features list.

Gordon
05-15-2003, 09:59 PM
Originally posted by gosmond
... while I apply the electric paddles to my wallet.
:p

Both my wife and I love that line. That should be animated in Lightwave!!

gosmond
05-16-2003, 11:58 AM
Gordon,

Yep. I tend to agree with you. The hardware technology is often ahead of the compiled software. Since I was a programmer for about 18 years professionally in my past life, I know that one of the tricks that we always did was to work on multiple 'screens' or graphics in different pages of memory and to also 'recursively' call sub-routines that modified graphics until the system cried.

I'm surprised at certain vendors like Matrox that actually brag about rendering realtime. (Not to be confused with seeing things happen in realtime). I was at B&H in Manhattan last week, and the sales person was trying to sell someone the new Matrox product. His exact words were '...and this will render to MPEG-2 in realtime since you can't go any faster than that..." I had to laugh.

I wish Steve was right - 'We don't need no stinkin' rendering" but unfortunately, I need it a lot!" Thanks for the sanity check and insight...


'CLEAR!'

ScorpioProd
05-16-2003, 08:40 PM
Yeah, anything I want via analog is real-time... But I do most things digital now... ;)