|The hard drives take up relatively little space in the very bottom of the case, after re-arranging some cables.|
- The first thing I did with my new shipment of parts was to swap out the right-angled SATA cables with straight ones. I had to remove the lower front fan from in front of the hard drive bay so I could put the two drives back in the configuration that I wanted them in, but the advantage was clear - absolutely no obstruction to the two intake fans.
- Next, it was time to install the fan controller. More on this story in a minute.
- I installed two cards, which I appropriated from my old desktop - a wireless network adapter PCI card and an nVidia 9600 GSO graphics card. The latter is definitely a place-holder - I had to check the benchmarks to verify that this card is actually superior to the Intel HD 3000 integrated graphics on my CPU - but I decided to go with this approach for two reasons. First, I'd like to see how the machine runs with the current hardware, so I can tell how much additional firepower I need. Second, Nvidia's new chips are expected shortly, and I see no reason to rush into a purchase when there will likely be new models or bargains on old ones in the near future.
- Now it was time to start routing cables. Because my build is relatively minimalist, it only took me a dozen cable ties or so to get everything secured to my liking.
|The finished interior of the machine.|
Now it was time for stress testing and basic overclocking. PC builders seem to have universally rallied around a program called Prime95 to test whether machines will crash when pushed to do hard computations for extended periods of time. After surviving an initial round of testing, I went into the ASUS bios and enabled their pre-configured automatic overclocking feature.
With little more than a single click of a button, this ramped the stock speed on my CPU from 3.3 GHz up to 4.2 GHz, and an extended round of stress testing indicated no problems. A real overclocker would push the envelope further, but this spot on the effort-reward curve is good enough for me, at least for the moment. The GPU, whatever I end up with, is already going to be the bottleneck for this system, and there's no reason increase power consumption, wear and tear, etc to go beyond that until/unless I need to.
From here, it was on to testing all the clients I copied over from my laptop. Here, the results were universally positive. I'm not able to max out all of the Anti-Aliasing and Shadow settings, but I can have max draw distance and texture quality, which are the two settings I care the most about. I can sort of notice the improvement in quality when you start throwing in the fancier effects, but it's not worth gutting frame rates. I'm able to run WoW, EQ2, LOTRO, DDO, DCUO, Runes of Magic and Rift on the new machine at my primary monitor's 1920x1080 native resolution, while leaving a second monitor (an old 1280x1024 monitor) up with a web browser and other utilities.
Overall, the project is starting to look like a win, though it will definitely be interesting to see how SWTOR, with its higher requirements, runs.
Unfortunately, the fan controller turned around to give me my first troubleshooting experience with the machine. I selected a Lamptron FC4 because it sounded functional and matched my aesthetics - no fancy LCD screens, and it was black with blue LED's, matching the case fans - and because it seemed priced reasonably at $30. I was otherwise going to be out probably $15-20 for molex adapters for my four case fans, so this seemed like a small additional expense to add some additional functionality.
Reviews on Newegg were mixed - many on Newegg commented that the build quality looked a bit inconsistent - but I figured that this was the issue you always have with any sort of online review - the one person who got a dead unit has a stronger incentive to complain than the dozen who didn't. Once the device arrived, though, I saw a bit of what they meant. The metal brace used to mount the device in the 5.25" bay is unnecessarily short - just barely long enough to even reach the first screws - and it took a lot of repositioning to get the thing in correctly.
Then, about half a week in to the life of the machine, I noticed that one of the LCD's on the front panel was no longer fully lit. After a quick check, I determined that the connected fan was also no longer running. My first guess was that maybe I did something wrong while installing it, so I removed both side panels and the front cover so I could unplug and replug the offending fan. No luck. Then I swapped the fan with one of the others, and it was immediately obvious that one of the four fan channels was no longer working properly, as the previously non-working fan fired up and the previously working fan sat, inactive. (The big advantage to seemingly cosmetic LED's on fans is you can immediately tell when one isn't running.)
Eventually, I discovered that pressing the knob down reasonably hard (not something that you'd otherwise have reason to do) caused the fan connected to it to fire back up, only to stop again the moment the knob is released. I'm not sure if something is bent or so poorly connected that it stopped functioning after days, but I'm concerned about trusting this device with the operation of my fans if it's this shoddy. It's possible that a pair of pliers may be able to fix the problem, but I'm not sure if I want to risk doing something that could affect my ability to return the thing if it's really that poorly made.
Meanwhile, I didn't spend money on a backup solution to the problem of how to hook up the fans, so I would need to partially dismantle the machine and leave it out of commission until a replacement arrives. Given that I cannibalized my old desktop for parts to set this thing up, this is a bit inconvenient. Ah well, a relatively minor first issue as far as DIY computers go, and at least I know enough to know how to identify the problem, rather than having to go to a shop with no clue what it's not working.
Final thoughts to follow once I figure out how to deal with the fan controller issue and have a bit more time to test the machine in action.