Sorry to reask a question. I'm assuming it's somewhere in the 16 pages already. But anyone able to point me to a thread discussing the individual subtimings and how they affect performance, and perhaps what certain ic's like? I know d9's can handle low trfcs but mine are in rma right now and I've got some generic ones to fiddle with at the moment. I've just been using memset to trial and error dropping things until they cause a reboot or other failure and using mbench or spi to test times. But I'm imagining that having some subtimings at low settings are preventing others from going as low, where it'd probably be more beneficial to have something higher than I've found so that I could get something else lower for more speed. Sorry it sounds confusing. Thanks
edit: Also something I've found with the new version with my p965 is that the reported frequency isn't always accurate. If I use the normal mutliplier, it's fine. But if I lower it, it seems to report "nbcc" vs actual speed. Note I'm always using 1:1. Not really an issue at all, but just figured I'd post it.
Also, I noticed cas changing has been done away with. What I've noticed with my p965 and memtest3.0 is that if I set a lower cas timing, I can set it higher via memset and set it back down. But if I start at a higher cas and try to go lower, it'll hang. For example, if I set cas5 in bios and try to set to 4, it'll hang. But if I set to 4 in bios, I can set to 5 and then bring back and forth between 4 and 5 all day long. I guess it's no big deal that cas has been locked since I wouldn't usually boot at a cas it can't handle only to raise it to regain stability. However, sometimes, I boot at a lower fsb with lower timings, and in my experimenting, I'd raise fsb, which may necessitate an increase in cas, which mt3.0 did fine.
Thanks for a great program
edit: not nbcc but rather fsb x set multi / default multi.
Bookmarks