Maya 2020 fundamentals - modelling the real world
Get halfway through a model and find it's an unworkable mess? Can't add edge loops where you need them? Can't subdivide a mesh properly? If any of this sounds familiar check this course out.
# 1 26-08-2010 , 07:38 PM
murambi's Avatar
Head of 3d Admedia
Join Date: May 2007
Location: nairobi , ken-ya
Posts: 1,340

CPU cores



Last edited by murambi; 26-08-2010 at 07:57 PM.
# 2 26-08-2010 , 07:57 PM
gster123's Avatar
Moderator
Join Date: May 2005
Location: Manchester Uk
Posts: 6,300
Some softwares can, I'm sure maya can.

There is a bit of talk about the more cores, as I remember when Intel brought out the 6 cores there was a test somewhere when it actually was slower rendering than a 4 core due the the cache in the memory making a bottleneck.

I'm not as up with hardware as I one was, but I'm getting a vibe that theres goingto be a big discussion in the future over CPU and GPGPU's and how they fit.

Intel et al are going for a more core route as a CPU, but the overall performance of haing a GPU (or GPGPU) doing a lot of the calcuations that a CPU would do is very strong. I've done some simple programming using CUDA and and its very good, well above a CPU.

Granted it's easier to programme with the CPU and thats it's advantage, but GPU's are very very powerful, and the tide is moving with realtime rendering etc.

Thing is at the end of the day the closer you go to higher cores the closer your going to parrallel processing and the problems your going to get.

Overall -

Intel and AMD dont want to let moores law down as thats the defining progress with a processor and they are at the top of the tree. If the GPU comes into full throttle and is taken up with all PC's then the CPU would just be milling arround doing the menial stuff, OS, HDD, RAM etc and a card would be doing the mass processing.

Intel are trying to go both ways and make a CPU that will be scalable into a GPU type architecture.

I may be very wrong as I'm about 1-1/2 year out of date on hardware but thats how I think its going.


"No pressure, no diamonds" Thomas Carlyle
# 3 26-08-2010 , 09:35 PM
EduSciVis-er
Join Date: Dec 2005
Location: Toronto
Posts: 3,374
This is a bit off topic, sorry, but if rendering was done by the GPU, which class of card becomes better, still the Quadros, or the GeForce?

# 4 26-08-2010 , 09:48 PM
elephantinc's Avatar
Level 32 pachyderm
Join Date: Dec 2007
Location: England
Posts: 1,859

Originally posted by stwert
This is a bit off topic, sorry, but if rendering was done by the GPU, which class of card becomes better, still the Quadros, or the GeForce?

Quadros I think, aren't GeForces practically identically but with a few things disabled?

# 5 26-08-2010 , 10:03 PM
EduSciVis-er
Join Date: Dec 2005
Location: Toronto
Posts: 3,374
They have different drivers I know... but for the same price, the specs are different. That's sort of what I was wondering. For example
GeForce 480:
$450
CUDA Cores 480
Graphics Clock (MHz) 700 MHz
Processor Clock (MHz) 1401 MHz
Memory Clock (MHz) 1848
Standard Memory Config 1536 MB GDDR5
Memory Interface Width 384-bit
Memory Bandwidth (GB/sec) 177.4

Quadro FX 1800:
$450
CUDA Cores 64
Core Clock (MHz) 550 MHz
Memory Clock (MHz) 800 MHz
Memory Size Total 768 MB GDDR3
Memory Interface 192-bit
Memory Bandwidth (GB/sec) 38.4

So.... yeah.

# 6 26-08-2010 , 10:19 PM
gster123's Avatar
Moderator
Join Date: May 2005
Location: Manchester Uk
Posts: 6,300
If your processing on the GPU then the more cores and RAM availalbe would be the key, so £ per power would go to the Gforce, same as rendering in Maya, just on a different platform.

Quadros are the same (or used to be) as Gforces but with different firmware, you used to be able to soft mod a gforce to a quado, I thnk they are more "top end" now as in that the hardware will fall down into Gforces.

Quadros are more "3D package" ready for open GL for viewport performance etc in the firmware, not in their actual make up, or, at least they used to be.

As for processing I know of a place where they replaced a small cluster with a Tesla and got a far superior performance with genetrc algorithms in matlab.


"No pressure, no diamonds" Thomas Carlyle
Posting Rules Forum Rules
You may not post new threads | You may not post replies | You may not post attachments | You may not edit your posts | BB code is On | Smilies are On | [IMG] code is On | HTML code is Off

Similar Threads