Callsign.Vega

Good lord that's painful to read.

AMD 64's don't like 4 slots, will force you to use 2t timmings. That is what I have to use because I bought 4x512. I did some testing swapping them out and lost about 4% overall performance by going 2t instead of 1t. As for 2gb over 1gb, especially for lock-on, I am not so sure. When I run lock on and check memory useage under a high-stress replay, I still have over 1gb free and available. It's most likely right on the border where your system would start looking for a page file with 1gb. I like the two GB, but if I had to do it over again I'd go with 2x1gb sticks, they have some newer ones out now that have almost as fast timmings as 512 sticks and pretty cheap.

Never run vsync, the only reason it is there is if you have tearing. It can actually lower your FPS.

Interesting. I know of a user on another forum that tried my method and recieved almost 30% increase in FPS. I wonder why it is not working for everyone. He also has dual 7800GTX's.

Which driver version are you running? Did you by chance test SLI AA and see if that worked?

Must be done with Lock-On profile deleted and a new one NOT made at all. Then, under global profile select SFR. Only way I got it to work. More specifics in my original post.

20% increase aint bad if you already have the second card. I average about 26% increase on my particular machine, I guess all the money I spend was not a total waste! ;)

I also never saw any FPS increase between one 6800 Ultra or two SLI until I used nHancer to delete the lock-on profile. The thing is, with that profile there the driver will override any setting you say to use. There lies the problem, users think changing the settings under the control panel even with coolbits is doing something, but in fact is not. You most likely don't need SLI with a 7800GTX but it would be interesting to see if your Lock-On is totaly CPU capped and it's real easy to test. Up to you of course.. :)

Evil, did you truely try SFR? You HAVE to go through the steps I posted about or it won't work. I kid you not, I did get my claimed performance boost, almost when I gave up hope.

When you have "show gpu load balancing" checked, running lock on did you have the single solid green horizontal line moving up and down the screen balancing the load in split-frame rendering? I have seen it where it does not move, that means SLI is doing nothing.

http://forum.lockon.ru/showthread.php?t=11506

This thread is about discussion of LOMAC:FC and SLI. I will provide some details as to what I have found spending many hours trying to optimize my SLI system for LOMAC. First, I will post my system: FX-55 2GB Ram Dual 6800 Ultras Widescreen LCD at native 1920*1200 resolution Forceware 81.87 beta drivers, coolbits 2.0 and nHancer I started this trek just like most people do, to squeeze out every single ounce of performance from a machine that cost a lot of money. Before I started tweaking, performance on my SLI machine was less then stellar. I started reading all of these LOMAC benchmarks out there and some of them did not jive. Example: http://www.xbitlabs.com/images/video/2005-17gpu/lockon_candy.gif All I care about on this benchmark is the 4x AA, 16x AF 1600x1200 comparison (red). If you are running any lower then that with SLI, you are doing something wrong. First thing to note is: a 7800GTX in SLI is actually slower then a single 7800GTX. Also note, a 7800GTX SLI setup is only 10% faster then a 6800 Ultra SLI setup. This would indicate more so then not that LOMAC is CPU limited. Example: http://www.firingsquad.com/hardware/athlon_64_geforce_7800_gtx_scaling/images/lomac1600.gif Same exact settings as the previous benchmark, yet a totally different outcome. Here, the graphics card is by far the limiting factor. An A64 3000+ having the exact same FPS as a FX-57 using a 6800 GT? An A64 3500+ only .7 FPS off of a FX-57 with a 7800GTX? What is really going on here? Example: http://www.firingsquad.com/hardware/nvidia_geforce_7800_gtx/images/lomac1600.gif From the firing-squad just like the one above, yet once again total different results. Here now a 6800 Ultra SLI has a 32% speed increase over a single, and a single 7800GTX has a 41% speed increase over a single 6800Ultra on a CPU limited game? Just by judging from those three benchmarks above, if you were trying to either upgrade your’ CPU or graphics card for Lock-On, there is no clear answer. Now playing around with my system with 81.87 beta drivers that does have a Lock-On profile, I found that DISABLING SLI and running one card actually gave a few % points FPS boost. This is running LOMAC at 1920*1200 res with highest quality set under display properties. The thing is, you can load LOMAC with “show gpu load balancing” and the SLI is barely even used. nVidia defaults LOMAC to alternate frame rendering, which did not net me any frame boost. I then tried SLI:8x AA. I then showed about the same FPS as with 4x AA. So finally I received a positive to SLI, you can run twice as much AA without a frame loss. I then tried to manually edit the nvapps.xml file to delete the lock-on profile so I could try split frame rendering. This did not work so good, as the nVidia driver has it locked down somewhere. I found this free program online that allowed my to delete the lock-on profile from the .xml and registry: http://www.nhancer.com/ With that, I deleted the lock-on profile, changed the global config settings to split-frame, launched LOMAC and did some testing. I showed a whopping 26% FPS increase across the board. The frame increase was with the same settings: 4x AA, 16x AF and drivers set to max quality. For some unknown reason, nVidia set the default profile for LOMAC to use alternate frame rendering instead of split-frame where there is a clear performance advantage. Using SLI:8x AA is nice but I do not see that big of a difference over 4x to warrant a loss of 26% performance, at least on my machine. Users with SLI, please submit your testing with SLI and see if you found similar results. Go ahead and try split-frame rendering over alternate frame and see if you received as large a performance gain as I have. Feel free to post comments and/or questions.