43

I'm a new graduate student in a computational lab, and the workstations provided by the lab I'm in right now are...slightly subpar. Long story short, a lot of it is nearing a decade old, and there doesn't seem to be a push to replace it with new hardware anytime soon. Since our lab tends to work with pretty demanding programs, this creates a major issue.

Over at Workplace SE, the standard advice seems to be to point out the economics of the situation (devs = expensive, time lost = money lost), but I'm not convinced that the same reasoning holds in academia. In addition, from talking to labmates and various administrative staff, it seems that the lab won't have the funding to replace all (or even most) of the machines anytime soon.

Which brings me to a potential solution: I have a decently powerful computer sitting at home. Would it be a good idea for me to spend some of my own money to upgrade it and bring it in to the lab? My advisor is okay with this plan, but I don't know if there's other things I should worry about (e.g. unclear line between personal/lab equipment, or people coming to rely on the machine).

This is somewhat related to this question, but it seemed to be asking specifically about money and a grant application, while I'm more concerned about potential consequences of bringing personal property to the lab and using it for an extended period of time.

Community
  • 1
chipbuster
  • 741
  • 1
  • 7
  • 16
  • 33
    Do follow the standard advice. Even the most underpaid graduate students are far more expensive than a $1K-$2K computer. Also, ask them today how they will pay for your academic travel in a year or two if they cannot afford a computer now. – Boris Bukh Jul 21 '16 at 06:11
  • 27
    @BorisBukh: you're making a lot of assumptions about how their budgeting works and, if travel and workstations can be paid from the same budget, how close that budget is to running out. Maybe not paying for workstations is what allows them to pay for your travel. Maybe the relevant funding for the two different things isn't fungible. But of course you're right that it's worth asking, because if you know the reasons (however stupid they might be) then the situation may become less frustrating. – Steve Jessop Jul 21 '16 at 09:17
  • 9
    @BorisBukh In a First World country, sure. But in Second/Third World countries, where the monthly "salary" for a PhD student may be around $200, buying even a single $2000 (or even $1000) computer may be nearly impossible. Academic travelling? Ha! – 101010111100 Jul 21 '16 at 11:07
  • 4
    @101010111100 but then you are less likely to own a powerful computer. – Davidmh Jul 21 '16 at 11:56
  • 1
    @Davidmh Not necessary true. My understanding is that in such countries, academia (along with many other professional areas) simply doesn't pay well. So you can make something like 2-5 times your PhD salary if you work outside of academia, say for one of those fancy technological companies which are popping everywhere, and you should be able to afford a reasonably priced computer, which in all likelihood is much better spec'd than anything your university can offer you. – 101010111100 Jul 21 '16 at 12:24
  • 1
    @SteveJessop's example is particularly true in the UK. For each PhD student (funded by EPSRC) There is an equipment budget, which would be used up on a fairly rubbish PC, and a travel budget. For a good PC the supervisor would have to find top-up money – Chris H Jul 21 '16 at 15:50
  • 2
    As an alternative to bringing in your own machine you could spin up a VM through a cloud service and SSH into it. The monthly cost might be more palatable to your adviser than buying a new machine plus it gives you a lot of flexibility and no worries over IP ownership – scurrie Jul 21 '16 at 16:05
  • 2
    Related: http://academia.stackexchange.com/questions/62589/is-offering-to-use-personal-funds-for-research-a-good-idea – StrongBad Jul 21 '16 at 16:18
  • 10
    So you can make something like 2-5 times your PhD salary if you work outside of academia — This is also true in the US. – JeffE Jul 21 '16 at 17:57
  • 1
    Don't bring your PC to the lab. Instead, consider doing your work from home. Also, some Universities have HPC systems available to researchers. You could try to get access to that if it exists and suits you. – Alexey B. Jul 22 '16 at 23:07
  • 1
    Do you intend to plug your computer into the university's network? If so, your network administrator may have an opinion about this. – Dawood ibn Kareem Jul 23 '16 at 00:50
  • Sadly its often such an administrative burden to get a computer quickly that using a BYO computer is often the most expedient solution to getting started on your actual work. But I suggest that approach as an interim solution while waiting for something to be supplied by work. – beldaz Jul 25 '16 at 02:52

9 Answers9

57

Another PhD student in my floor asked how to fix a slow system. Here is my supervisor's reply:

If you need to have a fast desktop, the best is to convince your boss to spend 3 days of your salary to buy a top of the line Linux machine (or a week of salary to buy a Mac).

Anything else is stupid on their part. Note that your cost is not only salary, but is also taxes, work insurance, university fees (where applicable), rent of the office...

Addition: Also, they got a grant to pay you, they need you to publish. The longer time you spend waiting for code to run, the less you will publish, and the less competitive they will be for future grants.

*Note: he was being moderately flippant, don't take the two days and top of the line too literally, but the point still stands.

The problem with bringing your own equipment is that:

  1. It is yours, and if you wanted to do something else with it, you wouldn't.
  2. If it breaks, and a lab is a moderately risky environment, it is your loss and no one else will pay for it. Also, you may not get support from the IT department to fix it.
  3. It sets a precedent for other students to do the same. The stipend you get as a PhD student is quite meagre, and buying a computer out of it is a significant investment.
  4. In some institutions research data is considered confidential, and they forbid for it to live in non-institution owned devices.

This said, I often bring my laptop to work when I need the extra mobility, or my own workstation (a top of the line Linux machine) is overworked, but I don't rely on it.

Davidmh
  • 21,215
  • 5
  • 60
  • 105
  • 4
    The economics of this are very different for postgrads than for employees. Postgrad time does not have a high economic cost and sometimes does not have a high value placed upon it. – jwg Jul 21 '16 at 09:33
  • @gerrit it is not only salary, it is also taxes, work insurance, university fees (where applicable)... It does amount to a lot of money. It is in Sweden, by the way, I think you are familiar with it. – Davidmh Jul 21 '16 at 10:13
  • 2
    Even if the machine only costs 1000 USD (which is nice machine but certainly not top-of-the-line), to earn that much in three days you'd need to make 6.5k/month (assuming 19 work days a month), which seems unlikely unless you're a full professor. But of course even 1-2 weeks of salary wouldn't be too expensive for something we use all the time. – CodesInChaos Jul 21 '16 at 10:13
  • @Davidmh True. I was thinking of the money I get in my pocket rather than the money my employer spends. Deleting my comment (although top of the line to me means something else, as I running code on a 48-core Xeon machine with 2 TB RAM, admittedly shared with a lot of other people, but certainly more than 3 days my salary costs). – gerrit Jul 21 '16 at 10:15
  • @jwg it was a PhD student, so still cheap labour. – Davidmh Jul 21 '16 at 10:20
  • @gerrit well, top of the line for a desktop computer. – Davidmh Jul 21 '16 at 10:25
  • 5
    @CodesInChaos You're missing a subtle difference (which is also explained in the answer): he's not asking to spend 3 days of his pay himself, he's asking his boss to spend the money it would cost to hire him for 3 days. This is a substantially larger amount since it includes the taxes, insurances, fees, office space rent, etc. – Cronax Jul 21 '16 at 10:44
  • @Cronax on fairness to CodesInChaos, I edited the clarification after he had written the comment. – Davidmh Jul 21 '16 at 10:56
  • @Cronax I think the rule of thumb is 20% on top of gross wage for insurances&co in Germany, so it shouldn't make that big a difference. And 6 m^2 of office space can't be that expensive either. – CodesInChaos Jul 21 '16 at 11:22
  • @CodesInChaos I could be mistaken, but as far as I know, every euro a Dutch or Belgian employee receives in their bank account has cost his/her employer two euros. You'd be surprised how expensive office space can get as well, although that cost is divided across all of the employees when factoring into this particular calculation. – Cronax Jul 21 '16 at 12:10
  • @Cronax I treated those 6.5k/month as "gross" wage from which the employee still has to pay income tax and part of the insurance. I'd expect a PhD to be below 4k/month. – CodesInChaos Jul 21 '16 at 12:45
  • Even if we generously assume that I cost $3 for ever $1 of stipend that I see, it'd still be a little more than 3 days of my cost...;) Your 3rd and 4th points are great though. I never even considered that. – chipbuster Jul 21 '16 at 18:03
  • 1
    @chipbuster even if we are talking about two weeks of salary, in three or four years pays off several times. – Davidmh Jul 21 '16 at 18:21
  • @Codes 20% is way too low. For 3000€ salary the employer already pays 3728€ which is 25% of the salary. But that ignores all the additional costs, particularly rental space and so on (vacation money!). When factoring all that in you'll be likely at 30-35% (the less you earn, the worse it is). – Voo Jul 22 '16 at 23:17
19

They really ought to provide adequate hardware, but realistically it may not happen. I have no problem with using my own computers for academic purposes (and had no problem during my PhD). However I wouldn't bring a personal desktop PC in to work. Or at least I wouldn't until I knew the security and IT support staff well, and then I'd be reluctant.

It's yours and you should have the first and main say on what happens on it, while they will rightly try to run policies to protect their network (which may conflict with your gaming/bittorrent-for-perfectly-legal-purposes). Software licences may also be an issue; they may imply or explicitly state that the software is to be installed on systems owned by the institution purchasing the license.

If you're confident to set it up securely (and there aren't any licensing issues), how about remoting in to it at home? You can do your development work on the machines provided (whether that's code, simulations or whatever), and then run them on your own fast machine.

Background:

An issue that your supervisor may face is that bog-standard desktops are a reasonable price but as soon as you want anything better the (only approved IT) supplier who got the deal on the basis of volume sales will charge a small fortune, meaning that there just isn't enough money in the budget (ability, not willingness, to pay for it). Combined with this, university IT departments can make it very difficult for staff/students to build a decent machine and get it on the network. These restrictions can often be circumvented if you have good local IT support and are on friendly terms with them (which is essential if you're doing anything remotely computational).

Chris H
  • 8,576
  • 22
  • 36
  • 4
    +1 Exactly this. If you have a faster PC at home, run the experiments there. There is no need to bring it to work, because it will be a hassle for IT administrators (security) or in terms of logistics (the PC does nor belong to the university). – Alexandros Jul 21 '16 at 09:20
  • 5
    Not to mention who pays if it gets stolen/damaged on uni presmises – Chris H Jul 21 '16 at 09:28
  • Regarding remote computing at home: 10 years old PC is still fast enough to perform daily computing. I use my laptop for CAD, typesetting etc. and it was "medium class" 10 years ago. And with high computing power the elevated power consumption comes - are you wiling to pay for your gear and it's consumption? A friend of mine rapidly downgraded his PC when he moved from dormitory to his own flat. – Crowley Jul 21 '16 at 15:29
  • Power consumption is a good point, especially for the sort of thing I'm assuming (computationally heavy). I'm ignoring the purchase price as the OP seems to be refrring to minor upgrades of existing kit. I'm only on my third desktop in 20 years (and second laptop in ~12) so for daily use up to casual games I'm with you. – Chris H Jul 21 '16 at 15:37
  • 3
    Power consumption is not necessarily that much of issue. Modern i7 processors (e.g. i7 4770) are very low on power consumption and support 32Gb - 128Gb of Ram. Most power consumption then depends on the graphics card. – Alexandros Jul 21 '16 at 19:05
  • 1
    "software is to be installed"? – user2943160 Jul 21 '16 at 19:05
  • 1
    @Crowley A 10 year old computer for typesetting and CAD? What? Replacing the desktop I had at university with my Ultrabook/Desktop at home I had a huge speedup compiling LaTeX (i.e. typeseting). I think you are severely underestimating the benefits of newer hardware. And as was pointed out correctly by Alexandros, power consumption of hardware has reduced dramatically. (My 2008 laptop has a 35W TDP processor, my more powerful Ultrabook - in terms of processing power - has a 15W TDP CPU.) – DetlevCM Jul 21 '16 at 21:25
  • @DetlevCM I've observed same speedup when I changed the newer "office" machine with my older laptop. My point was, that when computing power really matters and differences are significant (waiting seconds or waiting minutes to get results) then power consumption is not negligible. Yes, newer processors have low TDP, but as mentioned above, GPU eats a lot. OP is in computational lab, whatever it means, therefore I suppose that they will push the gear to its limits. – Crowley Jul 22 '16 at 07:31
  • @Crowley and even GPUs have become a lot more efficient. Nvidia GT 630 -> GTX 750 Ti, similar power draw, maybe 15W more on the 750 but 5 times the processing power and that's just a low/midrange card. So even if they do GPU computing, newer models will offer more computational capacity at lower power draw. What we have mainly seen in recent years is a slowdown in the processing capability growth while at the same time power consumption is continuosuly decreased. Another plus of newer hardware is new instruction sets that can give speedup as well as faster RAM. – DetlevCM Jul 22 '16 at 07:45
  • @ I agree that newer gear is faster and cheaper to run. Even though theeconomy increased significantly the power draw still cannot be neglected. I've just looked around and mobile processors have TDP from 2 W (Atom) to 60 W (i7 49x0 MX), desktop processors have TDP 35 W (i5 6x00 T) - 130 W (i7, Xeon). Doing computations 10 hours a day will cost you at least 0.3-1.3 kWh a day. You will save your time in the lab, but you will pay it by yourself. I do not say "Do not upgrade"; I say "Do not run your work at your home". – Crowley Jul 22 '16 at 08:23
  • 2
    @Crowley Here 1kWh is ~10-15p. Lets round up and say it costs £1/week or twice that if it's running 24/7. You can almost certainly save that elsewhere (if you're not working through the night because you model took a week to give nonsense data, you need less coffee). Plenty of people leave their machines fully on anyway, in which case you'd be comparing working vs. idle, reducing the cost some more. – Chris H Jul 22 '16 at 08:42
  • @ChrisH Fair enough. I don't know the exact prices, but if one sacrifise one beer in the pub he can pay himself a two weeks of computing! – Crowley Jul 22 '16 at 09:03
9

At least here in Germany there is this thing that (at least in public institutions) electrically powered machines (even phone chargers) need to be checked for safety by a specified person and then labelled that they are safe to use. This is because of liability. What happens when your PC causes a fire that burns down the building (obviously the worst case)? If the lab didn't officially approve, you might be in trouble.

In case this is handled differently in your workplace and you get the OK from your boss, you'll be fine I think. But as the first commenter said, you should not be forced to bring your own hardware. Keep bringing up the topic. Tell them how much time you're wasting with old hardware and maybe they'll budge.

Ian
  • 3,408
  • 1
  • 16
  • 33
  • 3
    That practice is not very widespread. I have seen it in the UK, but not in Spain nor in Sweden. Usually, the check is just looking at it, seeing that it doesn't have visible damage, and put a sticker on it. – Davidmh Jul 21 '16 at 07:20
  • 1
    Use of personal laptops is common here in the UK (e.g. the uni provides/you opted for a desktop, but you want to take your presentation on the road). The charger can be PAT tested when you first bring it in (I keep a spare in work). This aspect is (or should be) a non-issue) – Chris H Jul 21 '16 at 08:09
  • 3
    It should be a non-issue, I agree. Yet at the same time I am forbidden to buy a monitor from my private money and use it at my desk for the reason I described above. – Ian Jul 21 '16 at 08:17
  • 1
    Note that at least in Germany, these rules do not prevent anyone from plugging in all kinds of "unauthorized" devices nonetheless (and even if employees might go through the hassle of officially getting an authorization, all the students who bring their laptops to uni certainly do not). Might not be happening so much with stationary devices (such as desktop computers), though, which cannot be "hidden" as easily in the rare case of someone actually checking. – O. R. Mapper Jul 21 '16 at 09:19
  • 2
    Yeah, it's slightly odd, you can plug in your own phone charger but you can't plug in your own monitor. OK, the phone charger draws less current and is fused lower. But if you're allowed to just assume that the device draws the current it's supposed to and is fused correctly then there's no point certifying equipment, you've already assumed it's OK. I once had my own device certified for electrical safety at work by accident: I unthinkingly left in on my desk the night they were doing it, next morning it had a label on. No fuss, no problem ;-) – Steve Jessop Jul 21 '16 at 09:23
  • @SteveJessop: The way it's described in this answer, you cannot officially plug in your own phone charger. (Although, as I said, such rules are commonly disobeyed in my experience.) – O. R. Mapper Jul 21 '16 at 09:52
  • @O.R.Mapper: that's what I meant, I agree with you and was talking about places where it's done nonetheless. Doing it undermines the electrical testing, so there's no safety reason to be officious about monitors once you turn a blind eye to phone chargers. Well, there's a safety-from-getting-caught reason as you say, but not a good safety-from-fire reason. I wasn't clear that where I said "can" and "allowed", I meant in practice by your nearby colleagues, rather than by the rules. – Steve Jessop Jul 21 '16 at 10:04
  • If you take a monitor (or a coffee maker) to the office it will have the sticker that shows that it has been tested a year from now. – Carsten S Jul 21 '16 at 10:18
  • Do you not have national organizations [like in North America we have UL and CSA] to do this kind of certification? – Random832 Jul 21 '16 at 13:59
  • @Random832 part of the issue is preventing problems arising from the users, and not necessarily from the equipment. Imagine people using an extension cord and plugging in 6 or 7 heavy-current-draw equipment into one outlet, or people using old/second hand equipment with faulty electrical wiring due to old age or normal wear-and-tear. // Even in the United States frequently when outfitting a lab you will need to double check with facilities to make sure the equipment you intend to install in the lab will not overtax the electrical supply. – Willie Wong Jul 21 '16 at 14:30
  • @WillieWong A sticker on each piece of equipment doesn't seem like an attempt to address that either though. – Random832 Jul 21 '16 at 15:01
  • @Random832: I quote from Davidmh's comment "... seeing that it doesn't have visible damange ...". Also, extension cords are equipment too. – Willie Wong Jul 21 '16 at 15:38
  • @Random832 devices have to be CE marked, which is not quite the same but a similar idea. This is a type test though, and isn't performed on every unit, so the manufacturer is essentially slef-certifying. I believe this is the same in the US given the cost of testing a single item. Hardware gets damaged, or may even be built wrong, and detection of dangerous states resulting from this is up to the employer/premises owner. That said, the only hardware I've come across causing real problems had current tests - and caught fire due to poor design which couldn't be detected.electrically – Chris H Jul 21 '16 at 15:40
  • They recently checked the electrical devices im my office. I now have a checked sticker on my old power supply whose wiring I re-soldered myself years ago (to match another laptop, and to eliminate a galvanic-coupling problem). Much to my surprise it turns out I did that in a VDE-conformant manner... – leftaroundabout Jul 22 '16 at 19:00
  • @Davidmh There is actually a full recommended procedure for PAT testing in the UK, though it's not compulsory. Most academic institutions and a majority of commercial enterprises with an in house IT department use it. It involves checking for damage, yes, but also the condition of the wiring and its resistivity (a change since the last test can indicate an electrical issue that has dangerous potential). – ArtOfCode Jul 23 '16 at 23:50
4

I personally use my own laptop in lab, though mostly because it's convenient to have my data on hand whether I'm at work, at home or wherever, not because they couldn't afford a computer for me. As I understand it, there is no need for it to pass any sort of certification or be put on the institute's balance (at least here in Russia). Although it may be important to note that neither me nor my computer get in the same room with actual laboratory work too often.

The only possible issue is that some inspection may get wondering how exactly this particular bioinformatician is supposed to work without any device to his name. It sounds contrived, but I've heard that FASO inspectors sometimes are, let's say, overly meticulous. And, if you are planning to ever bring your machine out, better avoid getting it anywhere near the institute's bureaucracy. If it has a label on it, you are always at risk of being accused of theft and having to spend lots of time proving that you aren't a camel.

Synedraacus
  • 141
  • 2
4

This is possibly more of a comment, but a bit of personal experience/observations from Europe:

What is the ideal situation: You are provided with the equipment you need. What is the reality? You generally find you lack equipment - be it a good PC or a second monitor, etc.

When I did my PhD, I got an old computer - a Dell desktop that at the time was about 4 years old and technically weaker than my old laptop at the time... well, I used it - got a desktop at home a year later which was a lot more powerful (which I specced considering my PhD work) and then another 1.3 years later got so fed up with the old Dell that I replaced the entire thing with my own Ultrabook (which despite weighing 1.4kg is or was more powerful). Incidentally, I did bring up the idea of buying my own computer (small desktop) when they PAT tested the Dell (weird British habit these "PAT tests"...) and the suggestion was that it would not be allowed. Hence my solution was to get the Ultrabook instead (which cost 3 times as much but which I still have - and from which I write this post actually) The chance that I would have gotten a new PC towards the end of my PhD was also pretty much zero - why should they pay for it? (EPSRC only funded my tuition fee and maintenance... - heck, I had to be happy that the university paid for some experimental work...) I could also still use the Linux boxes and the Cluster just fine vie the University wifi using SSH so it wasn't a problem in any way. Incidentally, other students were better off - the "CFD lot" got new computers at the start and one guy funded by industry even got a laptop to use (not sure if he got to keep it, it was an option, not sure what the end result was though).

Now as PostDoc at a different institution I would love a second monitor - and instead have a Dell workstation that is weirdly specced... It has an E5 Xenon in there which I don't need and an NVidia Quadro (not a low one) but I only get a single monitor... I'm not the only one who'd like another monitor but effectively everybody in the institute has to make do with a single monitor. Why? I don't know. IT decided what we get to use.

A PhD student in the UK started and the mood was "he has a powerful laptop he does not need a PC". (I hope they did get him one eventually...) Incidentally, the same VM setup that took 1 hour to run on my desktop took 2 hours on his laptop - despite it having about 70% of the processing power of my desktop in benchmarks...

So what is the reality? Invariably in academic research you generally end up using your own equipment - because the equipment supplied is lacking in some respect or you want to work at home, etc. etc.

However, there are a some points to consider:

  • If you do commercially or otherwise sensitive work for industry, using your own equipment may be very much frowned upon.
  • Different institutions have different policies - some object to you taking your work home. Obviously, if you start writing a literature review at home there is little they can do, however you would be better off following company policy and leaving work in the office. (Unless they furnish you with a work laptop that you can take home.)

Then there is the topic of licenses:

  • If you can use free software or can afford the licenses for the software you use, great, no problem.
  • However I have noticed that it appears not to be unusual for some students to obtain software from illicit sources, this may have consequences if any questions are asked, for both the student and potentially the university (if they condoned such practices).
DetlevCM
  • 1,508
  • 8
  • 12
  • Is there really no source of spare monitors in the department? I've got a 1920x1080 that came with my PC (advantages of being in a well-funded, growing group) and a 1280x1024 with almost exactly the same vertical height. Even some of our PhD students have similar setups. It's a matter of knowing who to ask nicely. As for mystery software sources - if you're into that, it's a good reason to run personal hardware. – Chris H Jul 22 '16 at 07:55
  • 1
    No, it's institute policy. Some people found a low res second monitor but it isn't officially approved just accepted... Also it is only a 1 year PostDoc, so I don't have any leverage, if it were a permanent position I would insist including offering to buy the thing myself and leave it to the institute when I leave if it must be. But not for another 6 months. Back during my PhD I managed to get a second monitor before moving to my Ultrabook. When I was using a desktop at home I also had a dual monitor setup for about 8 months (and then had to leave that as I couldn't move with my desktop...) – DetlevCM Jul 22 '16 at 08:08
  • It looks the policies are often made by people with absolutely no idea of what they are deciding about and no will to get even slight insight. – Crowley Jul 22 '16 at 08:42
3

So much depends on the specifics of your situation. I used my personal laptop throughout grad school [in the US] and it was totally fine; in fact, most of my department did the same. I was the only one who used the machine, but no one cared that it wasn't university property.

In contrast, one of my jobs since graduation had very strict departmental guidelines on what computers could be used when and where. We had a few very old project laptops that were struggling to meet our data collection demands. Our IT department came back with a quote for a laptop that was insanely overpowered for our current and any future needs and about twice what we would pay for a comparable laptop on the open market. When we went ahead and bought one on our own anyway, it turned into a Big Incident that (really unnecessarily, IMO) went all the way up to the dean, because our IT department had a fit. For the same reason, I wasn't allowed to donate an old unused machine of mine for project use.

TL;DR: You already know your adviser is okay with it. Check that your department (admin, or IT person/department if you have one) doesn't have an issue with it either. If it's all good, agree on ground rules with your adviser before actually going ahead with it - who can use it, what happens if someone other than you breaks it, what happens when you graduate, etc.

ElleKay
  • 31
  • 1
1

I am also a grad student in a computational group and would agree with many of the answers above saying that you should be hesitant to buy/bring your own computer if that is going to be the only machine doing all of the calculations. In that case your PI should supply you with appropriate tools to do your research.

However, if (as is true of my group) you are using your computer as a portal to set up calculations to be run on a super computer cluster, the age and speed of the desktops in your lab is not going to affect your research in any meaningful way. In that case, bringing in a laptop to access articles and for music (my own guilty pleasure denied to those poor saps in wet labs) would be appropriate to supplement the slower desktop machines that you use to set up your calculations.

Rampallian
  • 111
  • 2
  • 3
    Ahh, I had left this detail out of the question for fear of complicating it, but about half our work is done on supercomputers as well. The issue is that the other half isn't related to the grants we have on the supercomputers, so we're stuck running it on our antiquated hardware. That probably reduces the incentive to upgrade (since we're not bottlenecked by our lab hardware for many projects) – chipbuster Jul 21 '16 at 17:55
  • @chipbuster How is your other work paid? It is contracted research? – Crowley Jul 22 '16 at 08:45
  • Unfortunately, I'm not 100% sure on what contracted research means :( Most often, an external organization (hospital, private company) will approach us with an interesting problem and we end up forming a collaboration. This sometimes comes with money from that organization (is that what contracted research means?) – chipbuster Jul 22 '16 at 15:38
1

I'm going to try to not answer this whole question,
"Would it be a good idea for me to spend some of my own money to upgrade it and bring it in to the lab?"

Rather I'll answer this part of the question,
"Would it be a good idea for me to spend some of my own money to upgrade it and bring it in to the lab?"

Since you already have approval to bring in your own, you likely can get approval for something which might be lower cost or be faster - public cloud computing.
If you are looking for long-term, then buying is cheaper, but for short-term, using cloud is great. AWS has a lot of compute and even GPGPU compute. They are a bit pricey, but there are plenty of other cloud providers out there. At a university I worked at last year, they often would use cloud computing for projects, especially those lasting less than a year.

In addition, you will eliminate any possible issues of:
- others using it, or turning it off while running
- power consumption
- risk of fire or other hazard
- storage
- location
- damage by others - and more.

MikeP
  • 1,505
  • 8
  • 8
1

I discuss about this problem for years now in my university (did my B.Sc. and M.Sc. here and now starting my PhD).

We also got a room with computers for students where lab sessions are held and some do their homework. However, several problems arise:

  • Software is hardly up to date, often more than one year behind the current stable release
  • Hardware was bought from the cheapest deal the university could get (of course, no money in academia) - resulting in a crappy keyboard, mouse and display and (most important!) most of the time equipped with a HDD instead of a SSD (should be getting the defa
  • Users do not have the rights to install or even update any software. Often important software is missing to do proper data analysis (e.g. notepad++ instead of the default windows editor)
  • If you come to the lab, somebody else might sit on your PC and you might have to switch to another PC at which you may not have set all your personal settings yet
  • Students pretend that they were not able to do certain analyses because the software was not installed/too old/crashing on one particular PC in the lab. Seriously?

I started early buying myself a good working machine which serves as my personal laptop at home and at work. I have everything installed I need and do not have to worry about others taking it. I can take an external display and work efficiently. Downside: Carrying the laptop with me every day or leave it in the lab (but miss it at home then).

I understand the point that universities have to offer capacities of what they want to teach. However, a lot of money is invested which could be used more efficiently:

  • Get the students a proper chair
  • Get the students a proper monitor, mouse and keyboard
  • Buy only 1/5 as many pcs as monitors and other stuff (for those who really have no PC) and let them bring their own laptop
  • Teach them using open-source software and avoid paying for licenses!

Nearly everybody nowaydays has their own laptop (even the most social sciene students) which is able to do the same as the (old) desktops PCs standing around in university labs.

However, as mentioned above, certain problems arise regarding softwares for which the university has only a certain number of licenses for specific PCs, sharing data in internal networks etc. (but this may all go beyond the question here).

In short: Get yourself your personal work setup and work efficiently! Do not wait for others to do something good to you. Your result counts in the end and nobody asks why you took so long or were not able to get XY running.

pat-s
  • 111
  • 3