UPDATE: Well, we just met with Monolith, developers of the original F.E.A.R. and we don't have much new info to provide. We were told that Monolith heard from Logitech about the issue and Vista, but that Monolith never heard back when it asked Logitech whether the glitch happened in Windows XP. Even if it had, it would still be up to the hardware vendors to dig up the problem and point to a specific place in the F.E.A.R. code. The developer theorized that it could be due to something going with DirectInput, the Direct X component that deals with mice and keyboards, but it didn't have anything concrete. Dizzy yet?
In better news, Monolith showed us a playable demo of the as-yet-unnamed sequel to F.E.A.R., which looked absolutely awesome, and seemed to really push current-generation graphics to the limit. They didn't allow cameras at the screening, so we have no screenshots (they're waiting until they have a name before they release any art), and wouldn't even take any real questions about the game. What we found interesting is that in the part we saw, at least, rather than ceding to critics of the original's claustrophobia by blowing the game up into a larger environment, Monolith ramped up the intensity of the close-quarters shooting to highly creepy, dramatic effect. It comes to PC, PS3, and Xbox 360 in 2008.
ORIGINAL POST: As bug discoveries go, we can't technically claim to have discovered this one. Still, it seems that by stumbling upon a major slowdown in F.E.A.R. during a recent Maingear desktop review, we've come across what might be a more universal issue in PC gaming performance. We're not trying to cause a panic. F.E.A.R is the only game we've seen suffer a significant frame rate loss. But an intriguing post on the official F.E.A.R. forum indicates that the issue might be more widespread. The potential culprit? USB mice and keyboards.
From what we've seen, if you have either a USB mouse or keyboard connected to your PC while playing F.E.A.R., you'll get a major frame rate drop on even the mightiest of gaming PCs. We're talking from 150 frames per second down to 13 fps or so. Using old-school devices appears to fix the issue. We found big performance hits with Logitech and Saitek hardware, although there's a world of USB devices that we haven't tested. We found that the Creative-made mouse and keyboard that come with Velocity Micro PCs don't seem to cause as much trouble, but they still show a measurable decline in performance, by roughly 10 fps.
The less severe drop-off with certain hardware might explain why we never noticed the problem until the Maingear review. If all USB input devices cause some kind of performance loss in F.E.A.R., the scores for all of the systems we've reviewed with that test will have suffered, giving us no "normal" baseline to compare to. We only noticed the problem because, for some reason, the Maingear system caused a significant-enough slowdown. We were also able to replicate the issue by connecting the same Logitech G5 mouse and Saitek Eclipse keyboard to a to-be-reviewed Falcon Northwest PC we have in the lab, so the issue isn't unique to Maingear (and that's why we didn't slam that system in the review). Still, we've tested F.E.A.R. on other PCs in months previous with one or both of those devices connected to them, and saw no frame rate drop until now.
As we said, we are not the first people to find this problem. Posters over at the official F.E.A.R. forum reported this issue last year. A forum thread on Nvnews.com also eventually finds its way toward talking about input devices. Interestingly, a poster by the name of "Noe" put a new thread on the F.E.A.R. forum this past July 3 (a few days after our Maingear review posted), asking readers for information to provide feedback on the forum, which will then be forwarded to "a group of qualified people looking into the technical aspects of the 'USB bandwidth issue'." This lends credence to Noe's claim later in the thread that the issue isn't unique to F.E.A.R., since (assuming Noe is telling the truth) it's an external group conducting this troubleshooting research, rather than Vivendi, Sierra, or any other companies behind F.E.A.R.
Before we had any idea it was an input issue, we simply saw F.E.A.R. scores tank during testing on an otherwise powerful PC. Alongside our trusty lab technician Joseph Kaminski, the guys from Maingear and Nvidia technical marketing director Nick Stam narrowed it down to the input devices, largely based on the suggestions of the Nvnews forum thread. We were afraid that Nvidia would bow out at this point, since the slowdown didn't appear tied to a 3D driver like we initially thought. We're glad we were wrong. As Nick wrote me two weeks ago: "I'm still very interested in understanding the pure technical reasons, as this is definitely one of the strangest issues I've seen in a long time."
With Nvidia working on the problem independently, we then moved on to Logitech and Saitek, neither of whom had heard of the issue. We're still waiting to hear back from Saitek's technical team (which has been traveling), but Logitech's was available and got back to us quickly. It doesn't appear that Logitech's SetPoint mouse and keyboard software caused the issue (it wasn't installed on the Maingear PC during testing), but from what we've heard, a forthcoming SetPoint update will fix the problem.
We're glad that Logitech hardware owners, at least, may have a solution, but we'd still like to know what exactly causes the slowdown, and whether it affects other games. We have a meeting with Sierra (one of F.E.A.R.'s developers) at E3 this week, so we'll definitely be asking them in person.
Of the vendors we've talked to so far, all of them remain interested. That gives us faith that the problem will get the attention it needs. We were frustrated, though, when we saw people had posted about this issue in official and unofficial forums as long as 10 months ago. Apparently it takes a while for the wisdom of crowds to sink in.