Ozmandias
Posts: 88 |
FF Newbie. Reading the forums on how slow FF is in rendering and the various replies on how to help the little engine up the hill. I decided to conduct my own experiment. Using what I felt were the two slowest filters in the most recent 30-day trial of FF, I timed how long it took to render the included lifesaver image. I believe it is a 600x600 pixel image?
I have read the recent "Ways To Reduce Your Render Times" thread and will test the other suggestions, however, previous threads going back several years had posts from GMM and others implied that changing the RAM usage from 60% to 90% would help considerably. So that is what I focused on for this test. As you will see below, it made NO DIFFERENCE whatsoever on any of the 4 systems. The two filters were in the "Stone" category. They were "Layered Stone" and "Crysta" System 1 - 8yr old laptop. P4 3GHz with Hyper Threading, 1G RAM, 128MB ATI Mobility Radeon 9600, 32bit WinXP Home. Layered Stone at 60% = 3min 51sec Layered Stone at 90% = 3min 42sec Crysta at 60% = 4min 24sec Crysta at 90% = 4min 19sec System 2 - 3.5yr old desktop. C2D E8500 3.16GHz, 4G RAM, 512MB ATI Radeon 4870, 32bit WinXP Home Layered Stone at 60% = 58sec Layered Stone at 90% = 58sec Crysta at 60% = 1min 8sec Crysta at 90% = 1min 8sec System 3 - 2.5yr old desktop. C2D E7500 2.94GHz, 8G RAM, ATI Radeon 4350, 64bit Vista Layered Stone at 60% = 1min 4sec Layered Stone at 90% = 1min 5sec Crysta at 60% = 1min 17sec Crysta at 90% = 1min 19sec System 4 - 1week old laptop. Dual Core i5 3210M (w/HT I'm assuming because Task Manager shows 4 CPU graphs) 2.5GHz, 8G RAM, on board HD4000 graphics, 64bit Win7 Layered Stone at 60% = 40sec Layered Stone at 90% = 40sec Crysta at 60% = 44sec Crysta at 90% = 43sec As you can see, changing the RAM usage from 60% to 90% was of no help whatsoever. -Oz |
|||||
Posted: February 19, 2013 8:06 am | ||||||
Skybase
![]() |
This is the part where I'm not technical enough but I think RAM's supposed to contribute to additional space for cached nodes. So it's more of the ability to store more data, hence your next round of edits to the filter would run faster. Pretty sure I'm right on that. But maybe I'm wrong. Would be nice to have further explanation.
|
|||||
Posted: February 19, 2013 8:16 am | ||||||
Vladimir Golovin
Administrator |
RAM consumption is complicated. See my reply in this thread:
http://www.filterforge.com/forum/read...&TID=11020 |
|||||
Posted: February 19, 2013 8:36 am | ||||||
Skybase
![]() |
Vlad, I'm mildly confused.
![]() |
|||||
Posted: February 19, 2013 8:51 am | ||||||
SpaceRay
![]() |
Thanks for making this render time and speed tests, and also for showing it with different RAM percentage.
I have seen in others posts that had been said that using higher RAM percentage would be better for speed, but never tested it myself if it was true or not and I have put it at 85%, but now seeing this test and it seems that the RAM at 60% or 90% would be the same, is better to keep it at 60% and avoid much more the Bad allocation bug. |
|||||
Posted: March 1, 2013 3:26 pm |
Filter Forge has a thriving, vibrant, knowledgeable user community. Feel free to join us and have fun!
33,712 Registered Users
+19 new in 30 days!
153,533 Posts
+31 new in 30 days!
15,348 Topics
+73 new in year!
20 unregistered users.