@Williamh695 Very nice discovery. I just tried it out and it makes a world of difference for XCX. I never had the bad fps that the OP of this thread had but after getting into NLA a little I decided the fps dips were frequent enough for me to call the game currently unplayable. I found some videos of XCX on youtube and there was one where a guy had an overclocked haswell i5 and was getting the same dips to 16 fps in NLA. Overall his cpu did fair better than mine but with the same dips I pretty much figured CEMU was the limiting factor here.
But wow what a difference that setting made. I definitely consider the game playable now. I ran through the same area around the BLADE tower with both the setting on and off. Each time I ran around the tower once to make sure nothing needed caching. Then I started the benchmark and ran around it again. Here are the results as recorded by fraps. All settings normal except 4k graphics pack is enabled.
2017-02-01 17:22:25 - Cemu
Frames: 2511 - Time: 106954ms - Avg: 23.477 - Min: 16 - Max: 31
2017-02-01 17:29:34 - Cemu
Frames: 2452 - Time: 86015ms - Avg: 28.507 - Min: 24 - Max: 32
I tried to make the runs the same both times but obviously I'm human. I remember running into a fence on one run. But I still consider the results to be quite valid. Obviously the game feels much better with a min fps of 24 as opposed to 16. Not bad for our "Prehistorical" CPUs eh?
@crimsonedge11 If you're still reading this. I noticed a while back that you have 12gb of ram. An odd amount for a sandybridge platform. I'm guessing 2x4gb and 2x2gb? This is completely anecdotal but I've read several people say that Sandybridge had a weak memory controller and that it was far better to only run one channel(2 modules) instead of both channels(all 4 modules). I would try removing the lower set of ram here, ensure that your best set is running in dual channel and try running the game again. Just a thought. Your results are just too far away from mine and even from williamh695s non overclocked cpu.
But wow what a difference that setting made. I definitely consider the game playable now. I ran through the same area around the BLADE tower with both the setting on and off. Each time I ran around the tower once to make sure nothing needed caching. Then I started the benchmark and ran around it again. Here are the results as recorded by fraps. All settings normal except 4k graphics pack is enabled.
2017-02-01 17:22:25 - Cemu
Frames: 2511 - Time: 106954ms - Avg: 23.477 - Min: 16 - Max: 31
2017-02-01 17:29:34 - Cemu
Frames: 2452 - Time: 86015ms - Avg: 28.507 - Min: 24 - Max: 32
I tried to make the runs the same both times but obviously I'm human. I remember running into a fence on one run. But I still consider the results to be quite valid. Obviously the game feels much better with a min fps of 24 as opposed to 16. Not bad for our "Prehistorical" CPUs eh?
@crimsonedge11 If you're still reading this. I noticed a while back that you have 12gb of ram. An odd amount for a sandybridge platform. I'm guessing 2x4gb and 2x2gb? This is completely anecdotal but I've read several people say that Sandybridge had a weak memory controller and that it was far better to only run one channel(2 modules) instead of both channels(all 4 modules). I would try removing the lower set of ram here, ensure that your best set is running in dual channel and try running the game again. Just a thought. Your results are just too far away from mine and even from williamh695s non overclocked cpu.
Last edited by Xalphenos,