Choosing the Right Processor for Fluid Flow Calculations at Low Reynold Numbers

Click For Summary

Discussion Overview

The discussion revolves around selecting an appropriate processor for fluid flow calculations at low Reynolds numbers, with a focus on hardware specifications, performance considerations, and software compatibility. Participants explore various processor options, operating systems, and the feasibility of computational fluid dynamics (CFD) analysis.

Discussion Character

  • Technical explanation
  • Debate/contested
  • Exploratory

Main Points Raised

  • One participant inquires about upgrading from a Pentium III to a Pentium 4 or AMD processor for fluid flow calculations, questioning the significance of processor speed and hyper-threading.
  • Another participant humorously suggests building a quantum computer to expedite calculations, indicating the complexity of the task.
  • Some participants note that the computational requirements depend heavily on the specific problem being modeled, emphasizing the importance of memory and operating system choice for performance.
  • A participant mentions running software under Linux, suggesting that it may enhance computational performance compared to Windows.
  • Questions arise regarding the necessity of CFD analysis and whether simpler methods could suffice for the calculations needed.
  • One participant expresses doubt about the feasibility of using a computer for the task and considers reverting to experimental methods like wind tunnel testing.
  • Another participant proposes the idea of breaking the task into a distributed-computing project to gather volunteers for assistance.
  • Concerns are raised about the participant's lack of experience with CFD, which may complicate the process.
  • There is mention of new 64-bit processors and their compatibility with operating systems, with a recommendation for AMD's 64-bit processors due to cost and upgrade potential.
  • A participant references a guide on choosing a CPU, suggesting that there are resources available to assist in the decision-making process.

Areas of Agreement / Disagreement

Participants express a range of opinions on the best processor choice, with no clear consensus on whether to choose Intel or AMD, or the necessity of advanced computational methods. The discussion remains unresolved regarding the optimal approach for the fluid flow calculations.

Contextual Notes

Participants acknowledge the complexity of the calculations and the potential for different hardware configurations to yield varying performance outcomes. There is uncertainty regarding the specific requirements for the analysis and the implications of using different operating systems.

sid_galt
Messages
503
Reaction score
1
I need to do some fluid flow calculations at low Reynold numbers. My current PIII won't suffice. Which processor should I go in for?

I was looking towards the likes for P4 3GHz with HT. I know with this speed, the calculations will take hours but I am willing to tolerate that. Will a 3.2 GHz with HT make a significant difference? Or should I dump Intel and go in for AMD?
 
Computer science news on Phys.org
From what you mentioned, I think you have to build yourself a quantum computer, it should take you only a few msec to calculate the whole stuff.
 
... of course this depends very much on your problem ... your problem can be modeled such that hardware wise the computations can be performed, whether it'll converge to anything meaningful would require a bit more detailed evaluation concerning the computational requirements of the problem. However, if you're working with low Reynold's numbers my gut feeling would be that by upgrading to recent intel / amd you would be able to increase the numerical performance to a level where you'd be able to do the analyses with typical PCs (just make sure memory & HDD specs are adequate, memory is pretty much as important as CPU). Also, running the system in any other OS than windows is a big bonus in terms of computational performance, e.g. the performance of same hardware configs running windows and linux is typically quite different (at least in all appls & benchsmarks I've run with FEA, CFD etc. software windows is a total resource hog / waster).
 
Thanks, I'll be running the softwares under Linux.

The Reynold numbers are very low - around 50000-80000.

So AMD and Intel both will do?
And how much advantage will 2GB Ram offer over 1 GB? Should I go in for DDR or DDR2?

Will an ordinary graphics card do as the ones specifically designed for this task like are very expensive (the softwares are compatible with most graphic cards.)
 
Sid, From your explanation, it sounds like you're trying to do a CFD analysis. Is this for school? If not, do you need to do a CFD analysis- do you need to do that rigerous of an analysis? What you're trying to calculate may be simple enough to do on your existing computer using a different method.
 
I had to go in for a new CPU anyway, just thought I might go in for a bit higher side. After doing some more research I have realized that I was extremely naive in thinking a computer can do the job. I think I'll stick with the wind tunnel
 
LOL -- you're probably right, would be easier to stick with experimental stuff ... not as much fun though.
 
sid_galt said:
After doing some more research I have realized that I was extremely naive in thinking a computer can do the job.
Can you break it down into a distributed-computing job? Advertise your distributed-computing project and volunteers will come.
 
Well, I could I guess but then I have very little experience with CFD so there is quite a chance that I'll mess up the thing.
 
  • #10
sid_galt said:
I need to do some fluid flow calculations at low Reynold numbers. My current PIII won't suffice. Which processor should I go in for?

I was looking towards the likes for P4 3GHz with HT. I know with this speed, the calculations will take hours but I am willing to tolerate that. Will a 3.2 GHz with HT make a significant difference? Or should I dump Intel and go in for AMD?

you know they have these new 64 bit processors coming out. I think they are already out, just not in common use. If you hae the money I would invest in one of those, although I am not sure which Operating Systems (if any) knows how to handle these types of processors. :confused:
 
  • #11
eNathan said:
...although I am not sure which Operating Systems (if any) knows how to handle these types of processors

Linux already supports 64-bit processors. Microsoft's Windows XP Professional x64 edition has just been released as well. I recommend using AMD's 64-bit processors over Intel's solutions. Not only are they relatively more inexpensive and support extended WinXP instructions, but with the socket 939 version, you can upgrade to dual core X2 64-bit processors later on without having to change your motherboard. AMD seems to have better implementation in regards to dual core, see http://www.pcstats.com/articleview.cfm?articleID=1797 .
 
Last edited:
  • #12
Have you read the sticky guide i posted here in regards to choosing a CPU? 0_o
 

Similar threads

Transitional Reynolds Number When Analysing Flow Through Orifices
  • · Replies 4 ·
Replies
4
Views
2K
Undergrad How is the Reynolds number derived (is my derivation wrong)?
  • · Replies 6 ·
Replies
6
Views
4K
Fluid flow around a wind turbine blade -- help please
  • · Replies 16 ·
Replies
16
Views
2K
Undergrad Reynolds Number in Organism-Fluid Interactions
  • · Replies 4 ·
Replies
4
Views
3K
Undergrad Laminar-turbulent transition and Reynolds number
  • · Replies 5 ·
Replies
5
Views
2K
Reynolds number & Mach number for flow regime
  • · Replies 6 ·
Replies
6
Views
2K
Undergrad Fully developed flow, Reynolds number
  • · Replies 12 ·
Replies
12
Views
4K
Can Reynolds Number < 70 Cause Turbulence in Elbow Fittings?
  • · Replies 20 ·
Replies
20
Views
5K
Graduate Confused about Reynolds Number- Fluid Mechanics
  • · Replies 11 ·
Replies
11
Views
5K
Controlling (liquid) flow and pressure at very low flow rates
  • · Replies 7 ·
Replies
7
Views
2K