Raspberry Pi Forums

Before i spend a bunch on money and time on Raspberry Pi3's i have a question that i can't seem to find an answer for.

I have been seeking to build a faster computer by clustering and i know you can cluster the Pi3's in order to have a fast CPU. So im researching some info to see if the Pi3 can help me.

I currently have an HP All in one with only 2 Cores and 2 Logical Processors. But i only use this CPU to run 1 program and nothing else. The program is Strategy Quant v.3.8 which is a x64 bit windows program (which can be found at strategyquant.com).

So the question is can i add at least 1 Raspberry Pi3 to my HP (which runs windows) in order to get more cores/processors for faster processing and if so then i can assume that i can cluster them for even faster service?

If i cannot add the Raspberry Pi3 to my original HP, then can i just buy a few of the Pi3's, cluster them together then download my Strategy Quant program and run this program indipentenly only on the Pi3's? I will of course have to buy a separate monitor and keyboard and such. I just need to be able to run my Strategy Quant program at faster speeds than are allowed on my CPU.

Also how many Cores and Logical Processors will each Raspberry Pi3 be?

Thanks in advance, Ryan Brignac

Short answer: No.

Longer answer: (still no)

BrignacCapital wrote:I have been seeking to build a faster computer by clustering and i know you can cluster the Pi3's in order to have a fast CPU.

I don't know where you read that, but it's not really true. You need special software to distribute the load and even then the Pi3 isn't anywhere as powerful as a modernish PC.

The program is Strategy Quant v.3.8 which is a x64 bit windows program

Which will not run on a Pi3 without emulator software which will make it really, really, really slow.

If i cannot add the Raspberry Pi3 to my original HP, then can i just buy a few of the Pi3's, cluster them together then download my Strategy Quant program and run this program indipentenly only on the Pi3's? I will of course have to buy a separate monitor and keyboard and such. I just need to be able to run my Strategy Quant program at faster speeds than are allowed on my CPU.

See above. That software will not run on a Pi.

Also how many Cores and Logical Processors will each Raspberry Pi3 be?

A Pi3 has 4 Physical ARM cores. Logical Processors does not apply.

Clustering Pis is a way to learn about...clustering. Clusters are useful in problems that can be partitioned in such a way that parts of the work can be run in parallel with little or no interaction between them most of the time. It takes special programming techniques and, if you're interested in learning how it is done (and what kinds of problems are amenable to this sort of solution) a cluster of Pis (known as a "Bramble") is a relatively inexpensive way to get started.

As others have said, "adding" Pis to an x86 PC to run x86 programs isn't going to work, from either the hardware or the software perspective. If your existing PC runs too slowly for what you want to do, get a faster PC. Having more cores or more threads may or may not help you. It depends on whether or not the program in question is written to take advantage of additional cores/threads. Most are not. Check the documentation on your program to see if it is "multi-threaded."

BrignacCapital wrote:Before i spend a bunch on money and time on Raspberry Pi3's i have a question that i can't seem to find an answer for.

I have been seeking to build a faster computer by clustering and i know you can cluster the Pi3's in order to have a fast CPU. So im researching some info to see if the Pi3 can help me.

I currently have an HP All in one with only 2 Cores and 2 Logical Processors. But i only use this CPU to run 1 program and nothing else. The program is Strategy Quant v.3.8 which is a x64 bit windows program (which can be found at strategyquant.com).

So the question is can i add at least 1 Raspberry Pi3 to my HP (which runs windows) in order to get more cores/processors for faster processing and if so then i can assume that i can cluster them for even faster service?

If i cannot add the Raspberry Pi3 to my original HP, then can i just buy a few of the Pi3's, cluster them together then download my Strategy Quant program and run this program indipentenly only on the Pi3's? I will of course have to buy a separate monitor and keyboard and such. I just need to be able to run my Strategy Quant program at faster speeds than are allowed on my CPU.

Also how many Cores and Logical Processors will each Raspberry Pi3 be?

Thanks in advance, Ryan Brignac

You appear to be grossly mis-informed, the Drupal Cluster is a good example of what is actually feasible:

https://www.pidramble.com/wiki

Well i guess you guys just crushed my dreams of a inexpensive super computer, lol. Just kidding. I appreciate the feedback before i started spending the money. Just thought clustering the Pi3's might be a feasible way for my program to process algorithms faster.

Thanks

Ryan Brignac

Something to keep in mind... The overall speed of a computer and the utility of multiple cores is inextricably linked to the physical *size* of the system (and chips) involved. An electrical signal in a wire travels at about 0.5c, so to get a signal between two points in less than on nanosecond, the two points have to be within 6 inches of each other. Even if you could link a Pi to an x86 system, the communications lag between the Pi and the x86 chip would slow everything down.

W. H. Heydt,

The overall speed of a computer and the utility of multiple cores is inextricably linked to the physical *size* of the system (and chips) involved. An electrical signal in a wire travels at about 0.5c, so to get a signal between two points in less than on nanosecond, the two points have to be within 6 inches of each other.

Quite so.

Whether than is a problem or not rather depends on the work being done. If you can split your job into 64 separate parts that each run for an hour with no interaction between them then the inter-processor communication speed may be a negligible slow down.

In general the speed gains to be had by adding processors are limited by the interaction between processes.

As summed up in Amdahl's law: https://en.wikipedia.org/wiki/Amdahl%27s_law