Computing performance

Hi All,

Just a quick question I hope you guys can help me with, thanks in advance!

The thing is that I have a system (2,7 ghz-i7, 8 gb RAM, 512Mb ATI) that sometimes breaks down due to " out of memory" reasons when applying algorithms to work for hours to compute many iterations..

Is there any way I could rent "computing performance" or "cloud service" out of my system to boost both memory and computational speed up??

Replies to This Discussion

Permalink Reply by Danny Boyes on April 16, 2014 at 4:00am

One thing you failed to mention, are you running 64 bit or 32 bit OS?

I'm guessing 64 bit as you have 8GB of RAM but you have to remember that you need 64 bit software to utilize more RAM, are any of the elements 32 bit? If its all 64 bit you should think long and hard about doubling your RAM c. £80

Permalink Reply by jozze on April 16, 2014 at 4:13am

Hi Danny,

I´m running Rhino on Windows7 on Parallel desktop (virtual OS) for Mac. I know it sounds a bit tricky but is the only way I could get Rhino+GH installed in my machine. Windows here is a 32bit OS (about 3gb dedicated RAM), same processor speed. Mother OS is Mavericks 10.9.

Permalink Reply by Danny Boyes on April 16, 2014 at 4:33am

No wonder you are running out.

I believe by operating Rhino in Parallels you are halving your Computer performance. Therefore you only actually have access to 4GB of RAM. but this is not the whole story as Windows 32bit will reserve 2GB for itself and let programs see 2 GB unless you use the 3GB switch. This will allow software to see 3GB and only reserve 1GB for Windows.

It is highly recommended that you use Bootcamp to run Rhino on as this would utilize the full potential of your hardware, but unless you use a 64 bit OS with Rhino 5 64 bit you will never see more than 3GB of RAM

Permalink Reply by jozze on April 16, 2014 at 9:56am

ok..will try then to swap to bootCamp and see what happens..thanks for the tip!

regarding the first question just for curiosity now..any clue about "renting" computing performance somehow?

Permalink Reply by Danny Boyes on April 16, 2014 at 10:02am

I am afraid at the moment (that is until RV6 and GH2.0 come along) Rhino and Grasshopper cannot even make use of the multiple cores on your own computer let alone harvest the untapped potential of "The Cloud"

Permalink Reply by Danny Boyes on April 16, 2014 at 10:09am

A quick search (using Google of all things... go figure) has turned up this:

http://www.bigv.io/

But you are looking at £10 per month per 1GiB of RAM. So if you are looking for more RAM then it could cost you £160 a month for 16GiB of RAM

Permalink Reply by jozze on April 21, 2014 at 8:57am

Yeah, it seems that for now it is worth spending some more cash in some RAM modules ..

Permalink Reply by David Rutten on April 21, 2014 at 1:27pm

Note that memory amount and access speed isn't always the bottleneck in a system. Adding more RAM will not speed things up if the processor is the limiting factor.

David Rutten

david@mcneel.com

Permalink Reply by dvdrbls on April 21, 2014 at 1:57pm

i wonder if you are willing to share your system specs

Permalink Reply by jozze on April 22, 2014 at 1:13am

System specs:

Processor DualCore i7 at 2.8 GHz,

10 GB RAM

ATI Radeon HD 4850 graphics (512 MB)

Following Danny´s tip I proceeded to install RH+GH in Bootcamp instead of Parallels; although the optimum performance is not reached yet, it runs faster and smoother.

Permalink Reply by Jonathan Sheridan on November 16, 2014 at 11:08am

Would a faster graphics card help a slower mother board chipset?

Permalink Reply by David Rutten on November 16, 2014 at 11:53am

Help with what? Traditionally the graphics card is employed to draw stuff on the screen while the CPU is employed to perform calculations. If the calculations take a long time getting a faster GPU won't help. If the display is slow, getting a faster CPU won't help.

It's not quite so true any more these days, the GPU gets used by more and more software these days, especially games, simulations and rendering software.