Downside of DIY

Posted by on Mar 27, 2017 in Hardware

In the long war between the PC Master Race and the Console Peasants (their words, not mine), PC aficionados point to their platforms modularity as being one of its overarching strongpoints. Each console generation, they say, limits users to whatever is in the box. When technology changes the only recourse is to buy a whole new console at full price — often while the current console generation is a long way from being obsolete. PC users, on the other hand, can upgrade piecemeal over time; while the initial outlay far outstrips the cost of a single console, it’s easier to upgrade over time where it counts: a new video card here, more storage capacity there.

Unlike building something like furniture, buying PC parts to build a custom platform has its own dangers that consoles will never (should never) see. To do this, individual parts need to be ordered and arrive in sealed packages. In all honesty, the act of building the PC — putting the parts together — is simple and is well documented both in writing and on the Internet (you do have another PC or tablet that you can use to watch YouTube videos in the event of an emergency, right?). What usually throws things off kilter is the quality of the parts.

Buying an “off the shelf” PC from IBuyPower or Falcon or Alienware is generally looked upon as a cardinal sin by the PC Master Race that views the act of putting parts together in the same way Jedi view building a custom lightsaber: it’s a rite of passage and a display of both mastery and an expression of individuality. What they don’t care to take into account, though, is that pre-built systems usually go through an intense “burn in” period where the assembler lets the system run intensive operations for an extended period of time to ensure that everything is working as intended. When building a custom PC in one’s own home from parts that arrive individually shrinkwrapped, it’s up to the end-user to perform that burn in and should anything arise from the process, it’s up to the end-user to deal with the issues.

This past week I had run into the dreaded (and increasingly rare) BSOD — Blue Screen of Death. It always seemed to happen after playing Mass Effect: Andromeda, which is probably the most taxing application I currently have. Often the BSOD would pop up after about an hour of gameplay, sometimes when just the game was running but sometimes when I was trying to work with the streaming platform Lightstream (which was something that occupied a lot of time this weekend).

BSODs are cryptic and despite the notion that it might be helpful, aren’t very user-parsable. I downloaded some memory dump readers from Microsoft that could help decipher these errors, and I think I had narrowed it down to an issue with the RAM — one of the parts that is always one of the most obvious culprits when things go belly-up. I downloaded and ran an app called memtest+ which boots into a Linux partition and puts the RAM through several tests and reports on error conditions, and when I returned, I found this:

The particulars aren’t important; know that “red is bad…very, very bad” and there’s a lot of red there. Having suspected that the RAM was at fault, I overnighted the same RAM from Amazon and had it sitting on the desk for its inevitable deployment. After the memtest results I took the system down and swapped the RAM, booting straight back into memtest and running it on the new memory which passed without any issues. I have yet to actually try putting the system through the wringer again (i.e. Mass Effect) but the fact that the original RAM showed errors and the new stuff did not gives me a high probability of success.

The problem is that while the RAM was obviously bad…what if the BSODs were a result of something else, and RAM just became the scapegoat? I’ve been monitoring CPU temperatures when playing The Elder Scrolls Online last night and they’ve all been around 40-60 degrees Celsius (for an i7-7700K that’s supposedly a good range). Voltage for the CPU and the RAM is also within normal operating limits as I’ve not gone anywhere near the overclocking abilities of the motherboard. I would hate to have the problem be my video card, although when I was in the case working with the RAM I realized that I had never secured the card to the rear plate with screws, so I fixed that oversight immediately to ensure that the card didn’t rock itself out of it’s PCI slot.

Last time I upgraded my PC I bought an off-the-shelf system because this kind of situation is exactly the kind of thing I didn’t want to have to deal with. Of course, that was almost 10 years ago, and we now live in the age of Amazon Prime and no-hassle returns. After posting this screenshot on Twitter, Belghast mentioned that he was sad that the age of local PC parts stores has faded into memory (no pun intended). It would have been a lot easier if I could jump in the car and drive down to a shop to swap parts, or at least to have had them test the RAM before I bought it. Amazon may be convenient, but it’s still a step back from where we used to be when CompUSA and Computer City were a thing in my area.

My PC will now destroy any console currently on the market (it remains to be seen with Scorpio) but at what cost? I do prefer the PC for gaming for many other reasons, but the potential for hardware issues straight out of the box is a major strike against it at least until the parts have gone through their proper burn in phase.