I have been, as you may know, highly concerned with buying or building a new computer. If you've known me long enough, you know that my thought and decision process typically starts and ends with writing about it, be that on IRC or on the blog. So, here it goes!
As I mentioned last week, when I performed "Computer Upgrade 3" and made the purchase of superslab, I had a fairly well-defined set of parameters for the search. I knew I wanted a laptop, that the laptop would be at most a 14-inch system, I knew I wanted to favor business-class systems, because I wanted to use a docking station. I knew that I had a budget of "about $2,500" and that I wanted the system to do certain types of tasks. At the time, I needed it to be a good performer at the Adobe Creative Suite, as well as running virtualization software. (At the time, my goal was to run a Windows XP virtual machine.)
The project and the machine have been an unprecedented success, and it has been one of my favorite computers ever. At the time, it was much lighter than my previous laptops, and it was faster than either of those laptops and my desktops. Even when faced with significantly faster hardware, such as eisbrecher, I continue using superslab because I just like using it.
But, having finished with school, it's reasonably clear that I don't' need that particular type of horsepower on the go anymore, and my Microsoft Surface RT has been my on-the-go computer for writing and day to day activities since I got it in October, 2012. (More on that in a subsequent post or two.)
And so, it makes sense to get a desktop again. My biggest challenge has been determining what the desktop is for and how I'll use it, as well as a budget. In 2009, I had a really clear timeline and budget, and today, as of this writing, almost nothing is decided – at all. I haven't decided between prioritizing photo and media work, gaming, and virtualization, how important expandability is and the types of stuff that should be used, nor have I even had success in deciding between buying an OEM system configured the way I want and building my own system.
Just about the only thing that's decided is that it's going to run Windows.
The worst part is that because I can't decide that one of these things is obviously the lead use case, I imagine what will end up happening is that the build will prioritize all three of these very different tasks, and I'll need to decide exactly how far to go with each task. For example, if I decide that I'm really likely to bring up a lot of virtual machines, and that there's a good chance I'll want, say, a test SharePoint and Exchange server running at the same time, then I'm going to need something that will support 32 gigabytes or more of memory – I'm probably looking at X79 or X99 (or an AMD platform) at that point.
The problem is that a move up to X99 as the base platform for the build implies a much higher budget for the whole thing than using H97 or Z97, which are suitable for "up to 32 gigs of memory," and somewhat ironically may be better for virtualization boxes, as you don't need a discrete GPU with those platforms.
Unfortunately, as always, I'm not exactly in the position to start buying right now. However, it's never too early to start the planning process!
The challenge is that without something to really limit the build, because I have a working system that does "what I need" (an issue I discussed last week,) I keep piling things into the build. When I first started assembling ideas for systems after Haswell was dropped in 2013, I spent most of my time looking at things to attach to the Intel DZ87KLT-75K, which usually resulted in $2000 builds. This isn't necessarily a problem, but the question is whether or not it's practical.
I've since tried my hand at a few more practical starting points. A mental roadblock I've always had has been replacing perfectly good components in a system with different ones after the initial build has happened. I even felt guilty for replacing the one gigabyte of memory and 80 gigabyte hard disk in superslab with the components that that system used. In service of that, and because there's not likely to be a really compelling new CPU come out for Z97 or H97 motherboards, I'll probably buy really close to the top of the CPU stack.
Without virtualization and hard-core gaming as needs, there are actually a lot of really neat small computer options these days, such as the Intel NUC and the Gigabyte Brix series, which includes a model with the i5-4570R or i7-4770R, which have Intel's IRIS Pro 5200 graphics, and a model that has a laptop CPU paired with what's essentially a very powerful GeFORCE graphics chip. The challenge of those systems, however, is that they are very noisy and with a sixteen gigabyte RAM ceiling, they're not really suitable for heavy virtualization.
Mini PCs, small OEM systems, and all-in-ones are really interesting for their ability to be a stationary computer and "create a space" for computing, but also to introduce limits in how the computer should be configured and how much it should cost. The biggest problem with those systems is that they lend themselves to frequent replacement, which in and of itself isn't bad, but isn't necessarily how I want to build and use my main computer.
I have yet to decide an exact priority for the build, but since I have the Computer Upgrade 4 page created, I'll be able to keep track of updates on the process conveniently.