El_Capitan Posted November 10, 2011 Posted November 10, 2011 Well, here's my test review... Introduction: One overlooked aspect for increasing graphics performance is the version of the graphics driver being used. Not all drivers are created equal, but does newer necessarily mean better? Some driver versions make performance increases specific to a certain line of graphics cards, other times they’ll only increase performance for specific games. After Nvidia’s release of the GeForce 500 series of GPU’s, much of their time would be dedicated to updating drivers to help that series, as opposed to that of the GeForce 400 series. This poses a question. Do newer drivers increase or decrease performance in the GeForce 400 series if they’re aimed more towards the GeForce 500 series? Today, we will look at how much newer drivers compare to older drivers in overclocking stability, and performance using one of the GeForce 400 series of GPU’s, the GTX 460 1GB. Closer Look: There will be seven Nvidia graphics drivers being tested: 260.89 WHQL 260.99 WHQL 266.58 WHQL 270.61 WHQL 275.33 WHQL 280.26 WHQL 285.62 WHQL Testing: Testing of the different Nvidia graphics drivers will consist of testing each driver with a GTX 460 1GB, and two GTX 460 1GB’s in SLI. The card(s) will be overclocked using MSI Afterburner and benchmarked using Unigine Heaven v2.1 to find the optimal overclocked settings. Unigine Heaven v2.1 existed during the release of the GTX 460 1GB and the drivers used in this test, which cannot be said for Unigine Heaven v2.5, which is why it is being used instead of the latest version. The system specifications will remain the same throughout the testing. I will test the card at the highest overclocked speeds for Core Clock (by 10MHz increments) and Memory (by 100MHz increments) at 1087mV’s Core Voltage for each driver version. I will raise the Core Clock by 10MHz until it fails a full run through Unigine Heaven v2.1 at the pictured settings: The previously successful overclocked setting that has no adverse Frames per Second results will be recorded as the optimum overclock for that driver. I will do the same for memory overclocks by 100MHz increments. Testing Setup: • Processor: Intel Core i7 950 180x24 4.32GHz • CPU Cooling: Swiftech Apogee XT Rev 2; Swiftech MCP355 pump with Swiftech MCR320-QP radiator and 6 Noctua NF-P12’s in Push/Pull at 5V’s U.L.N.A. • Motherboard: EVGA X58 SLI (E758) • Memory: 12GB (3x12 GB) DDR3 PC3-12800 9-9-9-24-1T G.SKILL Ripjaw s • Video Card: EVGA GeForce GTX 460 1GB • GPU Cooling: Swiftech MCW80; XSPC Rasa 750 pump with Koolance 120mm radiator and 2 Noctua NF-P12’s in Push/Pull at 5V’s U.L.N.A. • Power Supply: Corsair AX1200 • Optical Drive: Samsung CD/DVD Burner 22X • OS: Windows 7 Professional 64-bit • Case: HSPC Top Deck Tech Station (14x9.5) Testing Results: Single GTX 460 1GB Stable Overclocks: While difficult to understand at a glance using the table provided above, red highlights mean a failed run when higher overclocked settings resulted in success. Green highlights mean the optimal overclocked setting based on unusually lower than average Minimum FPS scores. Yellow highlights mean there was some unusually lower than average Minimum FPS score. Using the 270.61 graphics driver resulted in a successful Core Clock overclock of 980MHz, 10MHz higher than using any other graphics driver used in this review. Once using 275.33 and higher graphics drivers, memory overclocks resulting in unusually lower than average Minimum FPS. Below shows a cleaner comparison of stable overclocking results: Overclocking stability and performance continues to increase until the 270.61 drivers. However, after the 270.61 drivers, memory overclockability becomes an issue, but overall performance takes a slight hit. Dual GTX 460 1GB in SLI Stable Overclocks: Comparing two GTX 460 1GB’s in SLI’s stable overclock settings between the 266.58 and the latest available driver, 285.62 shows pretty much the same performance difference. Conclusion: While there’s no significant difference in Core Clock overclocking ability, there is a big difference in Memory overclocking ability. In fact, it could even mean instability at certain Memory overclocks. Anyone with overclocked Nvidia graphics cards would do well to retest their overclocked settings after a graphics driver update, especially for Memory Overclocks. What this means for factory overclocked Nvidia graphics cards, particularly for Memory, I’m almost too afraid to say. Pros: • Latest support for games and applications • No decrease in Core Clock overclocking ability Cons: • Decrease in Memory Overclocking ability and stability • Possible slight decrease in performance or instability for Factory overclocked Nvidia graphics cards • Slight decrease in overall performance after the 270.61 graphics driver Test Bench pictures below: This one is with one GTX 460 1GB watercooled. This one is with two GTX 460 1GB's in SLI watercooled. I moved the top Noctual fan out of the way so you can see how I'm cooling the VRM's. This one is another angle with two GTX 460 1GB's in SLI watercooled. I have 10 Noctua fans on the test bench alone. I also have two separate loops, 120.3 for the CPU, and a 120.1 for the GPU(s). I did this to make it easier to swap out watercooled graphics cards when I need to, or a CPU/Motherboard. I had to modify the test bench a bit from the original design, but I can swap out components pretty easily. Share this post Link to post Share on other sites More sharing options...
El_Capitan Posted November 10, 2011 Posted November 10, 2011 Here's some comparisons with other cards. This is why I love GTX 460 1GB's in SLI overclocked. Note, the HD 6970's in Xfire was tested with the same PSU/Memory/Watercooling setup, but using a 1090T at 4.0GHz instead of an i7 950 at 4.34GHz. Updated to include the GTX 580 and GTX 580 SLI, but with an i7 2600k at 4.9GHz. I'll update once the set-ups are identical. Updated to include two GTX 570's in SLI on the identical set-up as opposed to a different set-up (included that, as well). Not sure if it's just a driver difference or a CPU difference. Sigh, so hard to do so many different tests. Share this post Link to post Share on other sites More sharing options...
Stonerboy779 Posted November 10, 2011 Posted November 10, 2011 Nice review ive found that the latest drivers constantly crash and recover in trackmania, and trackmania alone so I am about to roll back. This is on stock clocks will need to check my gtx560 overclocks again. Good work got anymore conspiracies you want to test? Share this post Link to post Share on other sites More sharing options...
El_Capitan Posted November 10, 2011 Posted November 10, 2011 Thx. Is your Memory overlocked for trackmania to be causing the drivers to crash? Share this post Link to post Share on other sites More sharing options...
Stonerboy779 Posted November 10, 2011 Posted November 10, 2011 Thx. Is your Memory overlocked for trackmania to be causing the drivers to crash? Just the stock OC I will run it with lower memory clock and let you know if it follows what you have here. Share this post Link to post Share on other sites More sharing options...
El_Capitan Posted November 10, 2011 Posted November 10, 2011 Cool. What's the make/model of your card, btw? Share this post Link to post Share on other sites More sharing options...
Stonerboy779 Posted November 10, 2011 Posted November 10, 2011 (edited) Cool. What's the make/model of your card, btw? ASUS GTX 560 Ti DirectCU II TOP Stock clocks - 900/1800/2100 Max stable oc for benching - 1010/2020/2400 Max stable oc for normal use - 1000/2000/2400 I can run the bench oc about 90% of the time without issue. single card I can run an oc of 1025/2050/2400 for a single pass of furmark and the first few tests of 3d mark Vantage and 11. Stability would increase if I could raise my voltage over 1087mV but I aint going to bother messing with it. The stock voltage of the TOP card is 1025mV. Reference clocks - 810/1620/2004 Just played trackmania for an hour no driver crashes in online play but the same crash-repair navigating the menus in the online offline single player. Knocked the memory clock back to reference and no further crashes. Then put everything back to stock because alt-tabbing repairs the game anyway and I mostly play online. Will change drives this weekend if I can be bothered. Edited November 10, 2011 by Stonerboy779 Share this post Link to post Share on other sites More sharing options...
El_Capitan Posted November 10, 2011 Posted November 10, 2011 Interesting. If you keep the Core Clock overclocks but stock Memory clocks, my bet is that you'll be fine with the latest drivers. Share this post Link to post Share on other sites More sharing options...
cery25 Posted November 10, 2011 Posted November 10, 2011 nice review Capi Share this post Link to post Share on other sites More sharing options...
Black64 Posted November 10, 2011 Posted November 10, 2011 So does this mean ATI users are safe from memory OC instability? Share this post Link to post Share on other sites More sharing options...
Speedway Posted November 10, 2011 Posted November 10, 2011 So does this mean ATI users are safe from memory OC instability? Absolutely not It means we are waiting for Capi to do this for AMD Very Nice Capi I have always questioned this, and just never took the time to find out! Share this post Link to post Share on other sites More sharing options...
El_Capitan Posted November 10, 2011 Posted November 10, 2011 Absolutely not It means we are waiting for Capi to do this for AMD Very Nice Capi I have always questioned this, and just never took the time to find out! Lol, maybe. Though I have noticed my benchmark scores for HD 6970's Xfire and Tri-fire (no more Tri-fire though), got lower with each new AMD driver. Share this post Link to post Share on other sites More sharing options...
Recommended Posts
Please sign in to comment
You will be able to leave a comment after signing in
Sign In Now