CGTalk - 500 pound burning hole .

View Full Version : 500 pound burning hole .

wodagone

07-04-2009, 10:14 AM

Hi,
I am currently working on P4 3ghz HP laptop ( 750mb ram on board sis GPU ... yes i know ,please dont kill me,,, but it handles AM's Brussells rig pretty well ) that is about to die on me. And i need to handle scenes with 2-3 bruseels res rigs if possible.
I am non cg pro
so as hobbyist only have @ 500 GB pounds to spend on new machine.

I have searched forums for last 2 days armed with a basic knowledge of hardware requirements..now im really confused. Every one seems concerened with render speeds.

all i am working on is multi character animation.
All that concerns me is having better performance with untextured rigs + meshes in viewports. Also to allow for playblasting to be as fast as poss.

1.viewport speed ..i understand that the Graphics card deals with viewport geometry / textures.

BUT
When the geometry is deforming ( as on a skinned character )
is this calculated by the GPU or CPU

2. playblast calculation speed.

again ..is this driven by GPU or CPU
Also i have read that dual/ quad core allow me to be constantly playblasting in the background ( am i misunderstanding something here surely this is not possible ??)

. Render speed is no priority.
So on to the subject of processors. I have been given
a machine with dual xeon 3 ghz + 4 gb Ram. Would i be better of investing in 250 pound graphics card card. I realise only 1 of the xeon processors are active in viewport performance but is this not the same situation for eg. quad core only uses one " core "
for Maya leaving the rest able to handle lots of potential other apps/processes. ( which i would not be using ). So im thinking ..more powerfull single core ( 3ghz xeon aint too bad !!) + mid range quadro card.

next question.. This would leave me just enough money to upgrade OS to xP 64bit and ram to 8-12 gb. Would this speed up mesh deformation / viewport / playblasts.

oh yarp,, i am using Maya 8.5 but also considering upgrading to 2009 ONLY if it would be advantageous with quad core and or 64 bit OS.

Sorry for rambling on hope im making sense as im aware my technical knowlegde is a bit patchy. Any advice greatly aprreciated
Thanks

imashination

07-04-2009, 01:46 PM

1) cpu
2) cpu
3) no, more ram probably wont really help you in this case, unless you have lots of other software running.

wodagone

07-05-2009, 08:12 AM

thanks Matthew that clears things up. I realise there are a thousand threads on this subject but everyone seems obsessed with render speed. Here is why i am after optimum performance inside viewports.
when blocking out poses and up to point of refining my animation i run time line at 24fps. Then playblast to see reults for timing etc.
When it comes to polishing pass i run timeline to play every frame. This is extremely usefull for tracking arcs , spotting any pops/ bad spacing etc.
Obviously i can step through playblast frame by frame in quicktime but to be constantly playblasting is very time consuming. (But i suppose thats animating for ya ).

I am unsure on some points.
1. When a model is static on screen but viewport camera is moving around to view object this is calculated by graphics card. But when the vertices and/or the mesh are moving this is processed by the CPU.

2. No matter how many cores your CPU has, Maya ( up to and including v2009 )
will only use 1 core for deforming meshes. AND is this likely to change in future releases of Maya.

I cant believe this. I have been waiting for the day when 100ghz processors meant we could work with a billion polys in real time in viewport. But now we are limited to @ 3.6 ghz
or whatever the largest single core available is.

imashination

07-05-2009, 10:20 AM

1) Although the graphics card does of course have to process and display your view window, the reality of most slowdown in the editor is rarely the polygon count. Its almost always some other cpu-bound calculation working out what it should send the gpu in the firstplace. So yes youre pretty spot on, tumbling the view = gpu, deforming meshes, rigs, particles, physics, all cpu.

2) we'll see when it happens. The difficulty is that lots of things can only be calculated when something else is finished, they they end up very linear rather than parallel. Take a modern computer game, these are using 3-4 cores sometimes, but these are split into individual tasks, 1 core for audio processing, 1 for running the AI, another for the physics.

wodagone

07-05-2009, 12:29 PM

Great, thanks a lot buddy. I really wanted to nail down those questions before i wasted my limited cash on unnecessary over top hardware. I will pimp up the xeon that i have and wait to see what potentially might happen within future releases of Maya.
Thanks a lot
Dan

CGTalk Moderation

07-05-2009, 12:29 PM

This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.