Use Raspberry Pi to enhance computer performance?

Toggle sidebar Toggle sidebar
B

Benjamin Thayer

Honorable
May 19, 2013
10
0
10,510
Is it possible to connect many Raspberry Pis together and have a normal computer send it information to process? I have a laptop that I want to upgrade for gaming, but since it's a laptop, I only have a selection of a few processors to upgrade to and I could upgrade the ram, but instead of that I thought there might be a way to use the power of the R Pi to upgrade it instead. Is this even possible, If so, what would I have to do to get it working, and is it worth doing, or should I just build a desktop?

I found this guide on how to connect them for a supercomputer if that makes a difference. http://www.southampton.ac.uk/~sjc/raspberrypi/pi_supercomputer_southampton.htm
 
Solution
USAFRet
Can you build a Pi cluster? Yes.
Can you build a Pi cluster based on Windows? I've not seen one.
If possible, would it increase the gaming speed of a laptop? Doubtful. A cluster is built for specialized tasks. Games aren't built to take advantage of that.
Sort by date Sort by votes
USAFRet

USAFRet

Titan
Moderator
Mar 16, 2013
171,791
18,137
184,590
That Pi array is running Linux. You want a better gaming machine.
Those two things do not go well together.

Adding a bunch of Pi's to a laptop removes the only advantage of having a laptop - portability.

Sell the laptop and build a good desktop instead.
 
Upvote 0 Downvote
B

Benjamin Thayer

Honorable
May 19, 2013
10
0
10,510


Is it possible to do though? I would like to know how to theoretically do it, if it's possible, and how much it would increase the performance.
 
Upvote 0 Downvote
USAFRet

USAFRet

Titan
Moderator
Mar 16, 2013
171,791
18,137
184,590
Can you build a Pi cluster? Yes.
Can you build a Pi cluster based on Windows? I've not seen one.
If possible, would it increase the gaming speed of a laptop? Doubtful. A cluster is built for specialized tasks. Games aren't built to take advantage of that.
 
Upvote 0 Downvote
Solution
V

VeloSteve

Distinguished
Dec 2, 2013
1
0
18,510
The biggest single hurdle is that you would need source code for the game. Then you need to identify separable units of work which could be delegated to the Pi array. For example in a scene with many moving parts (such as projectiles) you might be able to delegate trajectory and lighting calculations for each object to a different Pi. Perhaps a scene could be broken into tiles, with each delegated to a processor. These are just examples. Another way to look at is that tasks which could be separate threads on a single processor are good candidates for moving to other processors.

There are lots of reasons this is hard, and potentially slow:

  • ■If the code isn't organized for parallel processing, modifying it could be almost as hard as writing a game from scratch.
    ■You would be communicating with each Pi on its 100 mbit ethernet port. No matter how fast the processors are, you still have to move all the data out to them, and move the results back.
    ■The separate tasks are only valid as long as they don't have to interact. Back to the projectile example - what if the projectile on processor 7 hits the target on processor 42? First of all, how do you know they even made contact? How long does it take to update the physics information for each object when they are on different processors? Does 7 know that it must send information to 42, or do they both have to check in with the laptop? How coarsely can you chop time slices before the results look jerky?
    ■Now, what about specialized hardware? A Raspberry Pi can do any calculation your CPU or graphics card can. Can the Pi do a lighting calculation within a factor of 10 (or even 100) of the speed of the specialized hardware on a graphics card? Probably not.
Short answer: no. Long answer: do some speed and bandwidth calculations and tests before you even think about an actual application.
 
Upvote 0 Downvote
You must log in or register to reply here.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom