Howdy, folks. I've been working on benchmarking some new games with the latest video cards, a task that requires me to learn how to play a host of different games, usually from different genres. Tough, I know. But it does present productivity challenges. If I get into a game, it's easy to lose track and burn through more time than intended while getting a feel for how to test it. If the game is unusually complicated and I'm not terribly interested in the genre—these two often go together—testing can be like pulling teeth.
Also, every once in a while, you just open up a package of rotten goods like Medal of Honor: Warfighter, a massive stinker that I endured on your behalf, folks. I played much deeper than necessary into this armpit of a shooter trying to find a "good" section to test. Finally, I realized that any section that feels like you're running through a tunnel with incredibly limited options and rote, scripted action is as representative as any other.
This morning's would-be productivity was burned in a different fashion, though: confirming for your information that the XCOM: Enemy Unknown demo runs at about 60-90 FPS, with virtually zero perceptible slowdowns, at its highest image quality settings at 2560x1440 on current enthusiast-class graphics cards. Frustratingly, it appears the demo won't allow you to enable higher levels of ambient occlusion, which might present more of a challenge. I'm debating whether I'm too cheap to buy the full game to see if higher AO levels slow things down much, which tells you all you need to know about how cheap I am. Still, as dated as the graphics look, I doubt many XCOM players are running into performance issues with recent discrete GPUs, even with high-quality AO enabled.
That's good news in a way, of course, for PC gamers. But it doesn't make for very good GPU testing. Neither does the nature of the action in a turn-based game like this one, where you tend to sit at the same view for longer periods than most. Results that show a long line of frames pumped out at 12 milliseconds, followed by another line of frames at seven milliseconds, followed by another at 17 milliseconds—not very exciting. Also, games played in this fashion are fraught with repeatability problems for testing. If you happen to linger in the seven-millisecond area too long in one run and at the 17-millisecond area too long in the next run, the overall results from run to run can look very different.
I know some folks ask why we test the types of games that we do in our GPU reviews, and part of the answer is that certain genres don't lend themselves to the task of comparing graphics cards against one another. That doesn't mean we aren't trying to include RTS games and the like in our test suites. Oftentimes, though, we sink some time into determining that a game isn't a good fit, and we come away without much to show for it.
|Gigabyte SA-SBCAP3350 puts formidable power on a single board||7|
|Alphacool Eisblock HDX-2 and HDX-3 help M.2 SSDs beat the heat||6|
|Corsair Lighting Pro Expansion Kit lets builders turn up the lights||8|
|Adata D16750 power bank is tougher than the average juice pack||14|
|Deals of the week: fast memory, an AM4 motherboard, and more||14|
|Corsair RMx White Series PSUs take a walk on the snowy side||24|
|Intel crams 100 GFLOPS of neural-net inferencing onto a USB stick||40|
|Toshiba's XG5 1TB NVMe SSD reviewed||9|
|Microsoft and Johnson Controls put Cortana in a thermostat||25|