My video card is pushing up daisies

[Ragicka]

Just a quick follow-up:

Received the new card today (GTX 960 FTW). Adjusted video and enb settings to reflect the new card and Skyrim runs smooth as silk.

Did a 30 min runabout which included bandits and a dragon attack on Riverwood. Using official high res textures and Realistic Water 2, Realistic Lighting and Realvision ENB (SSAO, DoF off). GPUZ recorded Max mem usage of 1968MB, Max temp 70 C. FPS stayed mostly around 60 (vsynch on) with rare and barely noticable drops to 50.

All in all, a dramatic improvement over my old 560 and a good fit for skyrim. Thanks again. :thumbsup:

[obobski]

Just a quick follow-up:

 

Received the new card today (GTX 960 FTW). Adjusted video and enb settings to reflect the new card and Skyrim runs smooth as silk.

 

Did a 30 min runabout which included bandits and a dragon attack on Riverwood. Using official high res textures and Realistic Water 2, Realistic Lighting and Realvision ENB (SSAO, DoF off). GPUZ recorded Max mem usage of 1968MB, Max temp 70 C. FPS stayed mostly around 60 (vsynch on) with rare and barely noticable drops to 50.

 

All in all, a dramatic improvement over my old 560 and a good fit for skyrim. Thanks again. :thumbsup:

 

Glad to hear it's working so well. I had figured it would. I'm guessing it also runs cooler than the GTX 560 doing all that too. One thing you may try if you're not hitting 100% at your vsync frame-rate is using "Adaptive Vsync" (technically speaking it should not be dropping from 60 to 50 FPS with vsync - it should immediately drop to 30 FPS if it can't hit full-speed and "double flash" each frame to maintain sync) - it would disable vsync when the frame-rate starts to dip. It's also supposed to decrease power consumption (and therefore heat). I've observed it working on both counts on my GTX 660, but if the game will run at full-speed it seems to add some stutter (I have no idea why this is so, but I have observed it in Skyrim). It's entirely possible that stuttering issue has been resolved with Maxwell.

 

One final thing to note: that "memory usage" is not dead-to-rights accurate and should not be trusted (for any card). The driver, DirectX/Windows, and the application will allocate memory beyond what they actually need, as a means of improving performance, stability, or just because that's how the developer felt the planets should align today - I forget which thread it was in, but I posted a number of links explaining this and the limitations of measuring "VRAM usage" on graphics cards, especially for DirectX 9 games. My whole point in raising this is that you shouldn't need to be worried that you're "up against the wall" if you have a 2GB card. The anecdotal example I tend to use is this:

 

Some years ago I had a Radeon HD 4890 (I still do but that's besides the point), it has 1GB of onboard RAM. Running Fallout New Vegas and Skyrim it would show 100% memory use in Catalyst (the AMD drivers) and GPU-Z. I replaced it with a GTX 660, which has 2GB of onboard RAM. Running the same games at the same settings, memory usage was reported at something like 25-30% (or around 600MB). I had expected to see 50% (or potentially higher, as I have seen all of the posts about "games these days" require gobs and gobs of VRAM). Performance was somewhat higher on the 660 as well. Neither card or implementation was or is "wrong" - there's just differences in how they're allocating and reporting their resources. It was mostly due to this observation that I did more research as to what exactly is going on.