will it improve performance?

hi blenderheads!

as i succesfully managed to run blender 2.45 on my (x)ubuntu 7.10 (originally ubuntu, but i installed xfce environment), i have a question about blender’s perfromance :slight_smile:

this setup is on my notebook. specs are: intel celeron M 440 (1.86GHz), 512MB RAM, intel gma 950 (others are not necessary).

when i run new instance of blender and run benchmark (draw), the result is 883 iterations. when working with 4 viewports, and in 3 of them is a blueprint of size mostly 1024x1024 or 1536x1536, the draw benchmark score is usually 120 iterations.

now i want to ask 2 things:

  1. how to improve performance without making any HW changes?
  2. if i buy another 512mb module of RAM and put it to dual-channel, will it result in noticeable performance boost?

thank you in advance :slight_smile:

There are a variety of ways to improve the performance. If you look on the net you’ll find a mixture of commands & configurations to improve performance.

For example: If you can go into the fstab, “sudo pico /etc/fstab”
You’ll see a line like so:


UUID=1726a8e1-9315-437a-a8d9-8949f4212187 / ext3 defaults,errors=remount-ro 0

at the end of the line put noatime.

To look like so:


UUID=1726a8e1-9315-437a-a8d9-8949f4212187 / ext3 defaults,noatime,errors=remount-ro 0

This can improve performance by approx 10%-20%. When reading from the harddrive.

There’s more ways to improve performace but I’ll leave that to you & google. :wink:

You can buy another 512MB stick of RAM, but you can’t just “put it in dual channel”. Dual channel uses matched pairs of “dual channel” ready RAM. But my recommendation would be buy another 512MB (or if you have enough 1GB) stick of RAM. Xfce is supposed to be the “lightweight” DE for Ubuntu, but with later releases it’s getting more and more resource hungry, so you will see an improvement (how much I don’t know, but it will be there).

maleficus: don’t worry, i know things about dual channel :wink: and thx for reply

kit89: thx, i’ll give it a try :wink: