The Elder Scrolls V: Skyrim Tested, Benchmarked

Prosercunus said:
Using both a GTX 580 and a 6970 (currently RMAing the GTX 580) I can say they both will run the game wonderful at 1920x1080/1200 with max quality.

The GTX 580 runs it a bit better than the 6970, but is also 150 dollars more and the difference seems negligible.

Just what I wanted to here on a 920 @ 3.8ghz + 6950 @ 900/1325 with unlocked shaders.

I should be able to play at max at 1920x1200 based on these numbers.

I was also quite surprised at the speed difference in SB-E would love to see a Westmere chip thrown.
 
Skyrim is not an ideal performance benchmark. Unless your intent is to measure a PC's performance against that of an XBOX 360 or PS3.

The lack of any post-DX9 features such as soft shadows, and its overall poor texture resolution make Skyrim unsuitable as a performance benchmark for PC hardware. That's not to say it isn't a great game, because it is. Just don't waste your time using it as a benchmark.
 
I would like to point out the lack of optimization in the game engine which I blame for a lot of AMD's poor performing. For instance my build is as follows:

AMD Phenom II X6 1100T @ 3.7Ghz (OC'd with a 2Ghz northbridge/HT Link)
8 GB Kingston Hyper X T1 Ram @1860 Mhz
2 x Radeon 5770 HD XFired
No real need to mention the rest.

Out of the box performance was poor, however, looking a little under the hood and around the web you find no AMD CrossfireX optimization in the game engine at all. I added a modded d3d9.dll which helped with utilizing both GPUs and also installed an AMD post processing mod which enables some post processing effects which made the game look better and surprisingly increased frame rate.

The most peculiar part of it all is that if you ctl+alt+del in the game then just hit cancel and go back your frames increase dramaticaly.

Moral of the story is this: most games come out optimizaed for Intel and Nvidia products and AMD gets to wait in the back seat. To get the performance out of the AMDs you have to know how to do a little tweaking. My Xfired 5770s run maxxed out and post processed at 60 FPS for spurts, then I have to CTL ALT DEL to speed it back up from whatever behind the scenes memory leak.

It;s unfortunate that the hardware always takes the blame for shoddy optimization and lazy coding.
 
Your article really should detail exactly what constitutes "Max" settings.

If AA and AF can be set independently of the preset, how is it valid to test with AF on with High and off with Ultra?

If you read the review you would know that we did detail exactly what constitutes "Max" settings and even provided an image to avoid confusion. As for the high and ultra presets, they are just that presets. That said our Ultra did not seem to set correctly and that must be a bug in the game.
 
According to the chart my 550ti should have less than 30 fps with maximum...now im wondering if it's on maximum? I did click custom and put everything the highest it could go.
 
According to the chart my 550ti should have less than 30 fps with maximum...now im wondering if it's on maximum? I did click custom and put everything the highest it could go.

Not sure what to tell you maxing out the game reduced the 550 Ti to about 30fps.
 
If you have a beefy enough gfx card, try using transparency multisampling aa from the nvidia control panel instead of using FXAA. I have ran some personal tests and found it neglible in terms of performance hit. I am using a single GTX 580 and i7-930 overclocked to 3.8 GHZ

The blurring which FXAA introduces makes the whole game look slightly washed out.
 
You should definitely get a new graphics card or go SLI. I'm running on an E8400 OC'd to 4ghz with an OC'd GTX 295.. I get about 30 frames per second outside during vista views and action, and 60 avg inside with max settings @ 1080p w/ FXAA.
 
wow i definitely didn't expect my 5870 to get a solid 60 fps on max settings. In Oblivion, with everything maxed, i'm lucky to get 40 fps(This is mainly because of the grass, but even without it, it hovers around 50 fps).
 
For anyone running AMD card(s) with Skyrim, AMD now have Catalyst 11.11a performance driver available for download (HD5000/HD6000 series)

Seems to be a bit of a mixed bag :

The Good
Improves performance 2-7% on single GPU configurations

The WTF
Disables CrossFireX (to resolve negative scaling and image quality issues seen when CrossFireX is enabled)
Not sure if this represents progress tbh. :rolleyes:

Bug-fixes for Batman, Rage and BF3 included
 
So reading that, it sounds like my stock clocked i7 920@266 will get 57fps with a 580 on Max settings at 1680x1050.

I have a 460gtx instead and that shows 45fps with an i7 2600k at max and 76 on ultra. So I can still play it on Ultra stock clocked and get 60fps. I hope.

Still, xmas sale for me. Looks like a lovely game but will I get the time away from Battlefield 3?
 
Good review, I'm personally running an Ultra mix and having no issues on my system (6850). I am a bit disappointed with some of the textures and 'overall' look as some things downright look bad. Fortunately modders have released several great additions from tweaked files to high-rez textures that really take Skyrim to a whole new level.

yukka said:
Looks like a lovely game but will I get the time away from Battlefield 3?

Got me away or at least to take a break from it :) .
 
So no comment at all about how this game only uses 2 threads which even further adds to the poor AMD CPU performance?
 
Definitely excellent review.

I have been playing at 1080p on MAX settings with 2 GTX 280 in SLI (using the beta driver just released from NVIDIA, else SLI is not supported) and an Core 2 Duo OC to 3.8GHz.

Indoor scenes i got smooth 60FPS all the tyme (limtied by vsync, otherwsie probably would get more).

Outdoor i get about 40-50fps.

However, both GPUs are only 40-50% utilized each, while both CPU cores are used 97%,

Definitely your CPU is primary bottleneck for this this game.

I will try reducing shadows. I was wondering of a core 2 quad will help, it was interesting that the review said only one core was used at max, with the core 2 duo both cores were used at max.
 
Skyrim works on exactly the same 5 x 5 exterior cell grid as the Oblivion version of the Gamebryo engine so "draw distance" is the same as Oblivion. Type TCL in the console fly high in the sky, type TB in the console then look down and see an identical 5 x 5 grid as in Oblivion. The grid outline is even the same colour, do an identical procedure in Oblivion to compare. This is not meant as a criticism of the interesting article nor of the Creation Engine. I wish this engine had been available for Oblivion, it renders much more efficiently.
 
This is why a lot of folks are modifying the number of grids to load. I'm currently (luckily) running with uGrids set to 11 and a modded exe to address more RAM. I've heard horror stories about stability after doing this but I haven't had a single crash yet so I'm thankful. I'm also running with very high draw distances for trees and ground shrubbery and I've settled on a shadow value of 8000. Now if I could just get the menus reworked I think I'd be done tweaking. I was hoping for some better water tweaks also, but it looks like I may be out of luck for now. The ability to interact with (non-combat) things from my horse would be nice also.
 
If the game doesn't scale well beyond 4 cores, why is the 3960X so much faster than the 2600k?
 
hi, how exactly do you set it to maximum as there is no max in the options/ advanced settings windows.
 
Certainly the best skyrim performance review out there. It must be because skyrim is often labeled as a crappy console port because sites never seem to want to use it a benchmarking tool so its exceeding difficult to find graphic card comparisons for the game.

Im running a i5 2500k OC to 4.3 ghz and radeon 6870 stock. I have tried overclocking the gpu but even at 945/1125 i only gain barely one fps. I get solid 60 fps anywhere thats not on the world map, 40-50 on most world map areas, and 30 fps in dense towns like whiterun. I'm at 1080p with "max" settings.

Based on the graphs it looks like going from a 6870 to a 6950 gives no performance improvement at 1050p, 4 fps at 1200p, and 2 fps at 1600p. is it safe to assume i made the right choice going with 6870 rather than 6950 and that i would see almost no gains if i were to upgrade?
 
im using sapphire 4830 hd, 2.8ghz dual core intel wolfdale and 4gigs ram, what settings are best?
 
Back