Why doesn't every company buy developers the best hardware?

Good questions to ask when being interviewed:<p>- What is your standard developer hardware configuration?<p>- How often is it refreshed?<p>You get bad answers to those questions and it's a definite warning sign (IMHO).<p>If you're working for a cash-strapped startup (possibly bootstrapped), you may want to negotiate a deal whereby you supply your own hardware (in exchange for higher salary and/or stock/option grants) if they're only in a position to provide substandard hardware.<p>One side note: what constitutes good hardware depends on what you're doing. If all you're doing is writing PHP/Python/Ruby in vim/emacs and running a local Apache/nginx and possibly a MySQL server then it almost doesn't matter what hardware you have... <i>apart from the monitor</i>.<p>But if you're compiling huge C++ (or even Java) projects then you probably want good I/O (possibly an SSD, preferably in RAID1 config for redundancy), lots of RAM and a good CPU.<p>As one data point: I have a 6 core machine with 12GB of RAM and 2 24" monitors (some opt for 1 30") plus a Macbook Pro 15" with SSD. You can refresh your hardware every 18 months if you want to (but most don't unless they have a pressing need; I know some people with 4-5 year old workstations because those are fine for what they do).

I empathize with you, the developer. I still want the best machine money can buy.<p>But I have also been on the other side. Let me give you a simple question to ponder: assuming an engineering team of 50, and the cost of 2% to upgrade machines, would you rather upgrade everyone, or hire one more developer? As the manager, I can tell you that everyone is asking for an extra headcount: "we need a full-time person to handle builds", "Tom could really use an extra hand with the XYZ module". Etc.<p>It happens to cost the same: hire an extra developer, or upgrade. So, as a manager, you handle the trade-off. Life is all about trade-offs, not absolute. I'd love to have a faster machine. I'd love to have more engineers. I can't have both.

To any development manager who wishes they could offer better hardware within their budget: If you haven't already done this, buy an SSD for every workstation. Nowadays, you can get 128GB for $200 or less.<p>Do that, and make sure the system has 4GB+ RAM, and 2 monitors at high resolution, and your hardware will feel brand new to the guys on your team.<p>I did this a couple years ago as one of my first acts after being promoted, and it worked great. I paid a bit more to get the drives that came with an upgrade kit (that is, an enclosure and imaging software). Gave them to each dev to take care of themselves.<p>I didn't have to wait on and work with the bureaucratic IT department, and a dozen SSDs was cheap enough I could put it on my card without needing CFO approval.<p>You don't need "The Best" hardware. In fact, it reminds me of the saying that "People buy horsepower, but they drive torque."<p>What matters to developers isn't how many FLOPs you can do, but how quickly can you load your VMs and start Eclipse and grep your local filesystem.

I have mixed feelings that the best hardware yields better productivity.<p>I worked with a Macbook Pro, top of the line, with 2 x 24 inch monitors attached, with an ergonomic keyboard and wireless mouse and all that crap.<p>Now I work with a $500 ASUS that doesn't even lit my keyboard, with Ubuntu Linux installed and I'm using it directly (no external monitors or keyboard) since I'm on the move a lot.<p>As far as my productivity is concerned, I still get things done at the same pace. Hardware is not my bottleneck.<p>Of course, this has more to do with my other preferences. I like keeping my toolchain as light as possible. I don't use bloated IDEs, even when working with languages that require an IDE, such as Java. I don't do heavy processing often and when I do, I prefer offloading that work on AWS. I am also proficient with manipulating virtual desktops.<p>As far as startups go, I think that spending money on expensive hardware is not frugal spending. If you're my boss, I would rather prefer a bonus than the latest and shiniest crap -- if I want shiny crap, I can buy it myself while respecting my own priorities.

My company proactively offered to upgrade hardware for whoever needs it. Even bought a huge monitor for everyone, which I didn't need and doesn't fit on my desk.<p>But a $99 software license I asked for 4 months ago that will make me more productive every day, I'm still waiting for it.<p>One time I was at a large company that gave all developers ridiculously fast and powerful machines. By far the highest specs I'd ever used. But the machine crawled because of all the million background processes IT put on there, and the crappy tools in our software chain.<p>Software is so much more important than hardware.

Doesn't it seem that everyone is avoiding the "elephant in the room" here? Virus scanners are a severe drain on productivity. Perhaps I'm in the minority by running a Windows box, but when I open up Process Monitor (<a href="http://technet.microsoft.com/en-us/sysinternals/bb896645" rel="nofollow">http://technet.microsoft.com/en-us/sysinternals/bb896645</a>) it sickens me to see how much I/O is going toward stupid virus scanning operations. Not to mention the 500MB of real memory that the various scanner binaries are consuming. My virus scanner consumes more memory and I/O than my IDE (Visual Studio 2010).<p>Maybe I'm in the minority here, but I would expect that a developer should know enough about computing to not get a virus. This, of course is assuming the organization has a competent IT department and is "securing the perimeter" with firewalls/proxies, and only allows approved software loaded on machines.<p>Sure, this would be a radical approach to IT, but I think it's a policy that needs to be implemented.

I worked for a huge company in their IT dept. We tried to convince our management to upgrade all the development machines using some of the arguments already mentioned in this thread. They did not want to believe that a hardware refresh would actually improve productivity ("show us the ROI"). A couple of us decided to build a tool to gather some metrics on how long it took to build and deploy our app on localhost (if you're interested see <a href="http://lopb.org" rel="nofollow">http://lopb.org</a>). We then compared results before and after a simple RAM upgrade on a few machines. We were able to show hard numbers to support the claim that better, faster hardware would save time waiting for build &#38; deploy to complete.<p>Although building in less time did lead to happier developers, it did not lead to more features getting built in the given time frame. We did eventually all get RAM upgrades at least, but the development process and technology stack we were using were the real time sucks that we could not fix with better hardware.

I've made this comment before on here. I purposely develop on older hardware because of performance concerns. Pretty much any code you write performs well on an i7 but will it perform well when it gets to your clients who are all stuck running Celerons? I always keep my development machines back a few levels so that I know if it runs well for me, it will for the end user.

In a lot of organizations, software development is secondary, so the machine you get is going to come out of some system that does a mediocre job of procuring machines for salespeople and secretaries. If you're lucky you get a new mediocre machine, if you're not you get a hand-me-down laptop from a salesperson who couldn't sell.<p>It doesn't make any sense, but 70% or so of people who hire developers don't seem to look at this rationally. On the other hand, they usually don't provide you good specs either.

What constitutes 'best hardware'?<p>I have a friend in high finance whose desktop is a (dual?) quad core with 12GB ram which he regularly maxes out. I have a colleague that writes firmware on windows XP, which with his various drivers and Eclipse problems causes a BSOD about once a fortnight. I keep offering him to get a new computer, cards, whatever, and he refuses because of the cost to him in terms of getting his environment set up 'just right' again - he's quite picky (at my previous workplace, a digital circuit designer on six figures didn't want to move from his PIII to a Core 2 for the same reason).<p>Another colleague runs a quad core on an SSD and complains that his CAD program runs slow... but when we run it through its paces, it doesn't seem to hit any bottlenecks bar the initial load (which isn't his complaint); it just 'feels' slow.<p>as with all things computery, the answer is: "it depends"

I don't care so much about the hardware as long as it does its job and there are no crazy wait times. Having good and recent software on the other hand is very important to me.<p>I simply hate having to put up every day with bugs that have been solved five years ago. And especially if there is no reason beyond simply the slowness and conservativeness of the IT dept that causes this.

My philosophy.<p>Buy a nice monitor (2 if you can afford it).<p>Buy whatever hardware is reasonable for ~1k or less. Repeat yearly.<p>This will always keep the developer with a good machine without spending too much money. With the way HW is nowadays, it's possible to go 2 years on each machine.<p>The story changes if you need laptops. After using LOTS of laptops over the years, if given a choice I'll only use MBPs now.

Something I always wanted to do as a manager was to offer not just great hardware, but essentially an allowance. That is, the developer is expected to buy and maintain their own hardware, but they get a budget. Say, $3000 at signing, $1500 annual refresh, and you have to pay it back if you quit or get fired for cause within the first 12 months. Coder gets to use whatever hardware they please, and keep it. I have predictable support costs, and in the grand scheme of things, it's a pretty cheap perk.<p>Never could convince my superiors. But maybe when I start my own...

If you're working in an in-house IT dept. then giving the staff of a cost-centre better equipment than the "people who make all the money" is a big issue.<p>If you're working in a big company then central IT probably have a standard PC supply agreement and standard image and will oppose allowing anything non-standard on the network until someone signs off on 2 or 3 extra support engineers to "support" this non standard stuff.<p>If your company isn't making much money, any capital expenditure like this is hard to justify.<p>So, you really want to work for a small software company that's making lots of money.

Another important 'hardware' that doesn't get enough attention is the chair &#38; desk. I believe that having an ergonomically designed and adjustable chair and desk is as important as the PC specs to programmer productivity. In my current job I have a bad chair/table and it is breaking my back. Probably this issue also gets worse with age. Investing in good furniture might also save the company on health care costs. From personal experience I have spent quite a bit on fixing my back.

From the other side, as founders, when did you switch everybody over from using their personal laptops to company laptops? A funding event? Certain size of the team?

Am i the only one who thinks developers asking for good hardware is an extremely bad idea? All the greatest software ever written has been developed on extremely slow hardware (especially by today's standards). Fast compile times is an oft-cited reason, but you can always work on the program while it compiles. Are 'developers' these days really that perfect and busy that they need to build their 10GB codebase in 10 seconds? And they have /nothing/ else to do while it builds ?<p>According to me, all programmers should be given absolutely the bare minimum hardware to program on. This way we can eat our own dogfood and hopefully reduce program bloat. The primary reasons programs suck these days is because 'developers' have terabytes of RAM on their development boxes and consuming 1G of memory for an applet isnt a big deal for them.<p>I say give all these developers asking for more hardware a 386!

In the past I have been content with hardware which is several years old. One advantage with slower machines is that benchmarking becomes easier, such that any slow operations are really obvious at the development stage. The other advantage is that customers are also often using similarly aged hardware. Older hardware forces you to make your algorithms efficient, rather than relying upon faster machines to hide the bloat.<p>In my case I was also often developing for embedded target hardware which was a good deal slower than a typical PC, so older PCs were more realistic for testing.<p>The advantage of always being on the latest hardware is if you're developing large software systems which take a long time to compile, or if you're doing something which fundamentally requires significant number crunching - such as games or computational chemistry.

I work with 135,000 of my co-workers and we all get rather slow, outdated, Windows 7 boxes with 2GB of ram. It's the standard config regardless of what you do. Supposedly it's much cheaper to manage for the IT dept than if we got our own stuff individually and I imagine it's much cheaper to buy 50,000 of the same computer from Dell than order a whole bunch of variety.<p>Seeing as developers are a small percentage of the total workforce, even if all of them complained, it would be drowned out by the mass of people who have computers good enough for their jobs. The cost of having to deal with ordering and supporting different computers (beyond just laptop versus desktop) is not 0. The quantifiable gain from having some people have better computers is very difficult to calculate. Thus, it's easy to just give developers the slow boxes and listen to them complain.<p>Related, I've previously emailed to ask why my company has a policy that every PC must be shutdown every night even though it can take up to 8 minutes to start up in the morning (we have a lot of required anti-viri/spam/malware and disk scanners that run). I was told that the company expects us to not be fully efficient all the time, go get some coffee while your computer boots. Also, different budgets cover PC cost versus payroll.

Real question: I know these are huge companies with diverse teams and projects but can anyone share what type of hardware Google, Apple, and Microsoft provide for their devs?

The economics in the article are faulty because they do not include the considerable amounts of staff time entailed in swapping coumputers out. Those include not just the obvious time spent acquiring a computer such as determining what is available, comparing specs and pricing, ordering, receiving and physical installation, but also all the productivity losses which configuring a new computer entails - e.g. installing and configuring all the various pieces of software for the new OS installation which invariably accompanies those Macbooks which the author advocates (the same would hold true for Windows machines as well and even the drivers on Linux would have to be tweeked for new hardware). And lest we forget, there's handling and disposing of the old computer which also takes time.<p>It's the sort of thing which can easily consume 40 person hours - even without the considering the inevitable time lost playing around with the new toy.<p>Finally, there's dealing with the inevitable pissing and moaning which accompanies any change - some people just want their damn computer left alone because it works fine, thank you very much. Other's wanted the 15" MBP not the 17", while the OSS fanbois cannot believe that they were once again thwarted in favor of commercial software.

As well as just the hardware speed, I think for some it would also improve their motivation, and you may gain more from an employee in that.<p>It also depends what your doing, if you are after someone to produce great pixel perfect designs, get them some decent screens. If you have an app where the latest i7 and SSD can cut compilation time in half you can probably make a good gain there by not having the programmer get distracted each time they compile.

This is a reason why I'm interested in "Bring Your Own Computer" idea. I mean I still bring my laptop to work anyway. Normally my personal computer is more powerful than what I have at work anyway. And if the company would pay or give me some bonus for it I'd be really happy.<p><a href="http://www.zdnet.com/blog/sybase/the-year-of-bring-your-own-computer-to-work/113" rel="nofollow">http://www.zdnet.com/blog/sybase/the-year-of-bring-your-own-...</a>

If you work at a big company, it's because most people are bad at programming, and the assumption is that you are too. Therefore, doing anything other than the bare minimum to prevent employees from burning down the building is a waste of money: a new computer is never going to increase productivity if the person using it doesn't know how to program.<p>The reality is that organizations don't change, and if programmers are considered code monkeys at yours, you need to GTFO if you aren't one. The reason your coworkers don't do more to change the status quo is because it's great for them: no real obligations and a nice bump in titles every five years. They don't need a better computer because they don't do any work. If you actually want to program computers, though, then you need to look for other opportunities.<p>&#60;jedi hand wave&#62; This isn't the employment opportunity you're looking for.<p>If you work for a small company and have this problem, it's simply because they're cheap.

Another point to make: if you're a developer, don't be afraid to <i>ask</i> for the best hardware.<p>I've seen many developers sit around staring at laptop screens while I'm working on my 2x24" monitors.<p>Do not feel like you're being greedy. Ask your boss. The most he can do is say no, but the likely thing he'll do is 1) ask why, 2) say yes.

I once had to watch a startup skimp on dev hardware, only to splurge on 'launch' parties and (no kidding) billboards...<p>The founder, a marketing major, had a hard time explaining things to their investor once the inevitable end was clear.<p>I'd ask the question: why wouldn't a company trust the developers' specs for adequate hardware?

Those few seconds of delay between performing actions can mean the difference between keeping focus on your work, and getting annoyed and distracted. I simply can't understand why anyone wouldn't want to give their developers at <i>least</i> a fast CPU, plenty of RAM, and an SSD.

I've been reading a lot of Paul Graham's works and one of the big things he points out is being cheap is good.<p>"8. Spend little.<p>I can't emphasize enough how important it is for a startup to be cheap. Most startups fail before they make something people want, and the most common form of failure is running out of money. So being cheap is (almost) interchangeable with iterating rapidly. [4] But it's more than that. A culture of cheapness keeps companies young in something like the way exercise keeps people young."<p><a href="http://paulgraham.com/13sentences.html" rel="nofollow">http://paulgraham.com/13sentences.html</a><p>Take those few moments of lag or downtime or whatever and enjoy your day--do something else that's productive or have a drink or something; make the best out of life.

Due to all the costs of adding extra developers (management overhead, communications complexity, inertia), spending $20k per person (every year or two) on hardware (desk/chair, home and office setups, laptops, phones, etc.) is still a win, if it lets you have smaller teams for the same overall productivity.<p>The bigger hassle for me is that upgrading machines causes some downtime, so it's better to buy loaded boxes and replace them slightly less frequently (every 18-24mo) vs. a new machine of lower spec every year.<p>Tools also are a great place to spend money; having a great build/provisioning/tinderbox/etc. system saves developer time, and doesn't add communications complexity.

I don't know if it's anyone else's motivation, but developers should have machines that approximate the machine of their software's customer base. Otherwise what runs OK on a developer machine can be painful to use out in the field.

Definitely a no-brainer. Always buy the best machine you can get.<p>I'm surprised that most developers don't take the same approach to tools. If you're using Eclipse or (god forbid) a <i>text editor</i> to write code, spend a minute and tally up all those 5-second chunks of your life you've spent this year looking up the names of variables, objects, whatever, and running into runtime errors from typos. Multiply by $$$/hour and see what you could have spent on a decent IDE.<p>JetBrains makes IDEs for pretty much every language out there by now, and any one of them will pay for itself in about four days.

My work dev laptop is bitchin fast, but runs slower than my 3 yr old personal.<p>Because of useless software (cya-ware) installed by corporate IT.<p>Useless anti-virus crap (ever hear of sudo?), ridiculous hard drive encryption, remote monitoring/management stuff.<p>Just working in Eclipse, I often wait every single keystroke. Yes, Eclipse is mostly a pig, and I've disabled/closed everything I could. But I have zero hassles working on my personal laptop, even when I have video (or audio) running too.

I had to buy myself a couple of SSDs. When building your project takes 2-3 hours of time, every improvement really matters.<p>Our management just tells us to go context switch onto something else, there's always lots of thrilling email answering and documenting to do. At other times you can switch to fixing bugs or working on another feature, even though I personally cannot stand continuous context switching as it decreases the quality of my work.

Minimum two large monitors, say, 24in 1920 x 1200. Preferably three. This, in my opinion, enhances productivity far more than the difference between a 3GHz and 4GHz machine.<p>That said...fast is good...faster is better...ridiculously fast is just fine!<p>All of our workstations have a minimum of two monitors, some three. A few are overclocked and use fluid cooling to keep them from going up in flames.

You're a programmer making $100K/year. If you're confident that a new whatever will demonstrably increase your value to the company, spend $1K/year on the new whatever and make the case later that you are more valuable and should be paid more.<p>And, yes, this doesn't take into account that the company might not want to support your idiosyncratic hardware choices.

Not everyone can do it, but for smaller purchases, it's much easier to just get it on your own and file an expense for it. 8GB memory SODIMM kits are about $75 and probably have the biggest impact on performance. I used to upgrade company-issued hardware all the time -- IT never knew and my manager didn't care.

It's probably easier for smaller companies to make hardware refreshes but in big corporations there's the whole issue of "standardised desktops" and so on to get around.<p>Getting a new machine isn't just a case of plonking a new one on your desk, I'd imagine a lot of red tape goes on in the background

As long as your machine provides reasonable performance, I doubt the marginal return of a better machine amounts to much additional productivity. I'd rather spend that money on 2 or 3 monitors, better software tools, or a quiet work environment.

I'd rather have developers use slow machines. This prevents a lot of foolish performance problems. The developer's laptop should not be faster and have 5x faster storage than the servers that will host the app.

I believe that most developers will gain 1 hour per day of productivity by having at least 2 monitors. Any company that isn't paying for multiple monitors is making extremely foolish management decisions.

Do you think John Resig needs the latest hardware to run VIM and Firefox?

I believe real cause that managers don't know math, bureaucracy or disability to see the big picture. Simple rule in our company we buy the best tools (software / hardware) that money can buy, because mathematically it make sense (there are a few exceptional cases)

I got the best machine they have at Alienware because I got fed up with managers and customers sending me 200mb pdf containing 8 raw bitmap files. I don't regret it although I still have my indy from silicongraphics for nostalgic reasons.<p>one of the best benefits? I can look something up in no time.

A rather silly post - there are a lot of reasons why companies don't buy the best hardware - the comments on the thread do a great job of listing them.