Dismiss Notice
Join Physics Forums Today!
The friendliest, high quality science and math community on the planet! Everyone who loves science is here!

What is the computer hardware needed by a physicist?

  1. May 6, 2017 #1
    I am looking at desktop and laptop options, but I do not know which will be beneficial to me as a Physics student.
    I am a sophomore and am looking for something that can last up to 5 years of performance without having to upgrade and also it should fit my needs as a physicist.

    Since I know so little about how physicists actually use computers, I am asking here.

    And also if anyone has arguments to support or oppose desktops, please let me know that is another choice I am making.
     
  2. jcsd
  3. May 6, 2017 #2

    jedishrfu

    Staff: Mentor

    Id look for computers that can run Matlab and Mathematica well. Matlab is an engineering standard at many schools and works in numerically solving differential equations, doing matrix manipulations...

    My preference are Macs because the MacOS is unix based. That makes it easy to work in a linux environment too.

    If you want it to last 5 years then you probably need to go with something based on an Intel i7. In five years it will probably be low end by then with the newest machines running 4 to 8 times faster computationally.
     
  4. May 6, 2017 #3
    I'm a sophomore too so what I'm telling you is just based on my experience as of now and what I've been suggested by my professors.
    I believe laptop is the way to go for a student, since you'll likely be moving around a lot and you'll want to carry your pc around with you(or at least thats what I've been doing in the last 2 years; my laptop is always with me).
    I think I would pick a desktop only when I'll find myself in a stable working situation. Performances will probably not really be a big issue until you are a student, so pretty much any mid/low-end pc with some 4/8gb ram and an intel i5/i7 processor will do it for you.
    I see a whole lot of researchers and professors in my university for some reason use macs, probably, as noted by jedishrfu, because they give you a solid working environment while still allowing a linux-like experience; i feel like they are a bit overpriced though.
     
  5. May 6, 2017 #4

    jedishrfu

    Staff: Mentor

    @mastrofoffi what is your experience with software for the sciences at your school?
     
  6. May 7, 2017 #5
    Can you please tell me if a laptop or a desktop is better? and thanks for the advice.
     
  7. May 7, 2017 #6
    I'm studying at La sapienza in Rome and this is what we did:
    first year, first semester: basic introduction to computer science and scientific programming in C
    first year, second semester: data analysis in R (relatively to statistics course) and primer on latex
    second year, first semester: computational physics 1(integration of ordinary and partial differential equations, pseudo-random numbers and statistical tests, random walks, percolation, cellular automatas) in C

    We haven't had any direct experience with Matlab or Mathematica or anything else; the approach is probably somewhat old-fashioned but I think the instruments we were given are enough to face the kind of problems we have to deal with. Also I guess they did not want us to be solely dependent on the use of proprietary software, in fact so far I've only been using emacs, texmaker, R, gnuplot.
    Matlab, Mathematica and Simulink courses are held at the engineering faculty, where they have more of a pragmatic approach, so we were suggested to attend those courses as externals if we were interested, but having learned C i believe that moving to one of these scientific environments, if I'll need, won't be a big deal.
     
  8. May 7, 2017 #7

    jedishrfu

    Staff: Mentor

    Of course, you'll need the mobility that a laptop provides. A desktop is great but you can't bring it to the lab when doing lab work or to the library when doing research. You'll need access to the internet from time to time wherever you are and the laptop fits the bill better than a desktop.

    However be aware that you'll need to keep a close eye on it as they are popular items of theft. Don't leave it on a table to get a coffee or in the library,to get a book. You must keep it close by wherever you are.

    Also with respect to Matlab, many schools in the US use it. Matlab has a full function student edition for about $95. They want to get students hooked on the software so that later in industry they can charge the full price for the professional edition with its specialized toolboxes.

    To be fair, It's quite good at what it can do and while there are open source alternatives Matlab is still the king for usability, speed and accuracy.
     
    Last edited: May 7, 2017
  9. May 7, 2017 #8

    phyzguy

    User Avatar
    Science Advisor

    I work in a university physics research lab, and almost nobody uses desktops anymore. A laptop probably has all the power you will need and you can carry it around with you. When you need more compute power than the laptop can provide, you connect over the network to a larger computing system. As jedishrfu said, you need to have a Unix based machine. This gives you two choices. The easiest is to buy a Mac laptop, which is already Unix based. This is what most of the scientists in my lab do. A second alternative, probably cheaper, is to buy a Windows machine and load a free Linux OS like Ubuntu on it.
     
  10. May 7, 2017 #9
    Obviously there is no one answer. We all have different needs, and for some their choices are constrained by their study or work conditions.

    Having said that, I prefer a desktop. I do not need my computer to be mobile. I like a big keyboard, a big screen, and the ability to easily upgrade the graphics card or the CPU. I like being able to quickly get in and swap out components when I need to. I like being able to upgrade the motherboard. I like being able to maintain the computer in one spot and move the screen and keyboard around if I want to. These things are important to me, but perhaps not to others.

    One point you mention is having the computer for five years. That's a rather long time for a computer to be useful without any upgrades. I would always prefer a computer that is easy to upgrade. I'm not familiar with the ease of upgrade for mobile computers. Others may wish to comment on that point.
     
    Last edited: May 7, 2017
  11. May 7, 2017 #10

    Mark44

    Staff: Mentor

    Ditto on all of these. My desktop has a real keyboard and a 28" LCD monitor (that's larger than a lot of regular TVs I've had).
     
  12. May 7, 2017 #11

    phyzguy

    User Avatar
    Science Advisor

    To each his own. I have a keyboard, mouse, and a large monitor at work. I plug my laptop into them each morning, and have the advantage of working with a large display, but then unplug it and take it home with me at night, where I have seamless access to everything I've done during the day.
     
  13. May 8, 2017 #12

    fluidistic

    User Avatar
    Gold Member

    My take on this seems quite different from most people who answered so far. Here's my undergraduate experience :
    I started with a desktop pc, a dual core core2duo E6300 which was fine to run Mathematica/Maxima/octave, compiling latex, etc. Its major challenge was solving the eq. of motions of 1000-2000 molecules using GROMACS. This took it about 10 days 24h/24 of computations. Note that the computer was still responsive despite the 2 cores being 100% busy, so I could still use it to check emails and browse the web.

    Then I went for a core i3 3217u laptop (dual core) which again was fine for all the purposes I had to face in the undergrad. Could run any software and compile and execute all codes I had to. Windows made it unbootable about 2 weeks before I graduated, so I had to switch to another laptop, a celeron n2830 which was a downgrade regarding the 2 previous hardwares. A really huge downgrade. What took about 10 s to compile latex on the core i3 would take between 1 and 2 minutes. I still finished writing my thesis in Latex on it though, and even followed an Internet course in programming in science (MIT's 6.00.2x) afterwards. The celeron was slow but did the job. It was very unpleasant though (due to slowness and 11 inches screen), I used it for about 5 months if I remember well.

    I am now using a desktop pc -a core i7 7700-. Even though it's easy to see its limits (for example using FEniCS to numerically solve partial differential equations using finite differences), I'm sure it's an overkill for an undergraduate degree in Physics even if you don't want to upgrade in the next 5 years.

    Overall I'd say a modern core i3 (7300 for instance) will be enough. Maybe even a pentium G4560 but I'm not 100% sure about the latter.

    About desktop vs laptop, it really depends on what you prefer. A laptop is handy if you have to make presentations or if you want to take notes with a computer when you're in the classroom. If you don't have a laptop and absolutely need one for a presentation, someone's going to lend you one so it's not a must, IMO.

    To sum up in a sentence : you don't need a core i5/i7 to solve and plot the solutions to the equations you'll see in your undergrad degree and maybe even in your graduate studies.

    Edit : About the OS my experience is that Linux was enough for most of the things I had to deal with. An exception was the use of OriginLab. I dual booted with Windows and ran it from there but I had a bad experience with Windows so if I had to use it again, this would be from a VM (virtual machine environment). This is what I have set up for now, Windows installed in a VM just in case I'm forced to use it.
     
    Last edited: May 8, 2017
  14. May 9, 2017 #13
    Thank you for the advice. And don't worry about Matlab, my University has a license for it and I got it for free.
     
  15. May 9, 2017 #14
    Thank you. I was just going to consider a Xeon v6 core which is apparently totally overkill as I see from your advice. It think that I am going to settle for a core i5, since I already have a core i3 and it doesn't do the job that well.
     
  16. May 9, 2017 #15
    I think that my final solution is the Dell Precision 5520 with a core i5 or i7 and an Nvidia Quadro. It has good performance and good battery and is made for such work (comes with Ubuntu) . If you have feedback on my choice I am open to it. And thank you all for the advice.
     
    Last edited: May 9, 2017
  17. May 9, 2017 #16

    phyzguy

    User Avatar
    Science Advisor

    Sounds like a good plan to me.
     
  18. May 12, 2017 #17
    Since it sounds like this computer would be for mixed personal and work/school use, you may want to go with Mac or Windows for the wider choices of commercial software for your personal interests. Modern Mac has the advantage of having inherited some Unix, which has advantages as noted above. But then you're restricted to Apple hardware unless you deal with the effort to maintain a Hackintosh. Windows 10 has WSL (Windows Subsystem for Linux) that allows a more efficient way to have Ubuntu on your machine than a VM. There are limitations but if you go that route, I'll be curious about your impressions (see my newly posted thread https://www.physicsforums.com/threads/what-do-you-think-of-windows-subsystem-for-linux-wsl.914487). In either Mac or Windows case, you can dual boot or VM a full Linux distro if your software needs require it.
     
  19. May 12, 2017 #18
    I just graduated with an applied physics degree. Any laptop will be sufficient. I used a dual-core 2.0 Ghz, 1gb of RAM laptop from 2008 running Manjaro Linux. SageMath (http://www.sagemath.org/) could be painful at times, but that was it. The runtime was slower than my partner's who was using an i5 processor but never more than a minute or two)
    As for desktop vs laptop, the desktop is far more upgradeable. If the i5 CPU is inadequate in five years, all you have to buy is a new CPU, not a whole new rig. Additionally, GPUs have many cores and are from my understanding started to be used to massively parallelize simulations. https://developer.nvidia.com/cuda-gpus
     
  20. May 12, 2017 #19
    Difficult if not impossible. Some upgrades like HDD to SSD and increasing RAM are possible, but usually these aren't points of conflict. CPU upgrades may be possible if the MOBO allows, but is rare. MOBO swaps in laptops is even rarer.
    Upgrading GPU in a gaming laptop maybe be possible, but is beyond my knowledge how likely.
     
Know someone interested in this topic? Share this thread via Reddit, Google+, Twitter, or Facebook

Have something to add?
Draft saved Draft deleted



Similar Discussions: What is the computer hardware needed by a physicist?
Loading...