I just purchased an Nvidia GTX 260.

After benchmarks, I’d say it’s a minor-to-moderate upgrade from my 8800 GT, depending on the games played.

A few months ago one of my hard drives started vibrating excessively, causing a reverberation with my fans. I put off replacing parts for awhile, which was a bad mistake: By the time I’d noticed my PC’s heat rise to excessive levels it had already done the damage. My videocard has been acting flaky since.

The 2xx series has come down in price recently (I paid $220 Canadian + tax). It’ll probably come down even more after the release of DX11 later this month, but once I was seeing graphical smears / glitches on my desktop, I knew it was time to replace my beloved 8800 GT.

Benchmarking the games that matter most to me right now:

As you can see, UT3 didn’t change much. Supreme Commander (Forged Alliance) went up the most and Left 4 Dead actually went down. All three are DX9 games. My guess is the 2xx series is probably meant mainly as a DX10 upgrade. I have noticed LOTRO in DX10 looks prettier on the GTX 260.

I also went with the base version of the card from EVGA, rather than the overclocked / superclocked variations which would have shown a greater edge. I was more concerned with heat and noise than the extra mileage. However, I did make sure to get the core 216 (more streams are better right?) version built on the 55nm (smaller = cooler) process.

For benchmarks, Michelle conveniently has the same rig:

E8400 Core 2 Duo CPU.
2 gigs of RAM.
ASUS P5K SE motherboard.
Windows 7 (RC1)

This is what I’d consider a moderate-to-high end machine, self assembled and running smooth. The 2 gigs of RAM doesn’t bottleneck as much on Windows 7 as Vista, but probably more than XP (but then I wouldn’t have DX10 available). Michelle’s hard drive is different / slower, but Supreme Commander’s simulation score was exactly the same (the difference was in the graphical score: 7313 vs 8048, whatever that means) so I don’t think the tests were degraded much by any caching.

The specifics of the benchmarks:

• All tests at 1280×1024, with vertical sync and frame smoothing off.
• Unreal Tournament 3: “Vivid” post processing. No AA. UT3Bench: Ran 60 second bot (12 bots) matches on 10 maps, then averaged the FPS.
• Supreme Commander Forged Alliance: “High” Fidelity settings. No AA. Benchmarked using /perf which simulates an intense 4-way battle.
• Left 4 Dead: Max settings, 4xAA. Benchmarked via my own timedemo of the farmhouse finale of Blood Harvest on advanced difficulty (which I suck at). Ran the demo 3 times and averaged the FPS.

One thing I have noticed is good AA performance on the GTX 260. I went back and ran the tests again but cranked up the AA and saw no noticeable difference, which I know would have slowed down the 8800 GT. In Left 4 Dead I can comfortably use 16xQ CSAA with only a frame or two change. To me, that’s the biggest difference with this card. I’m running at a moderate resolution mind you, this might not be as dramatic at something like full 1080p (1920×1080).

Another note is I like the card design. Here’s a quick snapshot of it next to my 8800 GT:

It looks pretty monstrous out of the box and it’s quite heavy. It takes up two slots, with an intake vent directly from outside of the case, designed so the fan pulls air over the massive heatsink that covers most of the card. I feel secure that this will dissipate heat better.

Overall I’m mostly happy, but I can’t help but feel I’ve spent a couple hundred bucks to get back where I was. D’oh!