Overfiend Posted May 21, 2006 Posted May 21, 2006 one thing about TRAS setting......seems going from 5 to 1 (yes it works) doesnt do anything........bandwidth in memtest didnt go up even 1 mb - sandra stays the same....every benchmark doesnt react....so its better to leave at 5.....seems to works best regarding stability as well as bandwidth. on the other hand lowering TWTR TRWT delays to 1, 2 gave me almost 5% up bandwidth everywhere..... rock stable so far - 8 hours 3dmark06 loop not a single crash - seems the key is proper Tref setting (using 3120) as someone said before...before i couldnt go above 225FSB with 3 3 3 7 timings seems my FX60 wont go on 3 Ghz - even pushing voltage up to 1.525 doesnt do the trick - crashes constantly yet im happy as it is.....for now Share this post Link to post Share on other sites More sharing options...
Guest eva2000 Posted May 21, 2006 Posted May 21, 2006 seems my FX60 wont go on 3 Ghz - even pushing voltage up to 1.525 doesnt do the trick - crashes constantly yet im happy as it is.....for now Memtest will not report all gains/losses of memory settings in terms of bandwidth... instead try looping test #5 and time how long it takes to complete a set number of passes... tighter timings which don't show bandwidth gains can be timed to take less time to do the same number of set loops Ya need water for 3ghz stable.. FX-60 is a hot bugger heh Share this post Link to post Share on other sites More sharing options...
OCZTony Posted May 21, 2006 Posted May 21, 2006 Guys remember tight TRAS can be good and bad, the issue is you may be forcing the pages shut quicker than they can be read, which may be ok for memtest86 and a bandwidth test but bad during a 3D bench etc. The way to tune TRAS for best stability is like this. I use memtest86+ and the OCZ developed memory tuning from within the app. I test all TRAS values from 5 to 15 noting bandwidth against each value, after I have altered TRAS in memtest I use C then 7 to restart the test and make sure the TRAS I set is being used. Doing it this way only takes a few mins. Then I set the highest value that gives the best bandwidth...IE if 5 thru 11 give the same bandwidth i will set 11. Now once i have done that i usually play with read to write and write to read also....TRAS usually has to go 1 to 2 places lower to get the best from lowing R2W and W2R and TWR etc but at no time will i set 5 as you usually are dialing in instabillity. Tight TRAS may be better for the odd benchmark but overall you are better tuning TRAS in the way i have described here so that you have fast stable rigs usually running lower Vdimm also. Share this post Link to post Share on other sites More sharing options...
Angry_Games Posted May 21, 2006 Posted May 21, 2006 hey Tony is there a newer memtest? everytime I run it on the CFX3200-DR, it shows 3369MB/s regardless of how my RAM is clocked...the L1 and L2 cache values for hte cpu change, but the memory bandwidth always shows 3369MB/s Share this post Link to post Share on other sites More sharing options...
OCZTony Posted May 22, 2006 Posted May 22, 2006 Not seen anything later than 1.65...i will check though. Mine runs fine here though are you using the 1.65? Share this post Link to post Share on other sites More sharing options...
learners permit Posted May 22, 2006 Posted May 22, 2006 Well after two weeks with the CFX3200 I gotta say I'm VERY impressed with it. A bit touchy to get set up but it's working very well for me and I'm still learning to find more speed. My observations so far are that the most critical settings for a high HTT are setting the correct t ref, ram drive strength, and read preamble. Share this post Link to post Share on other sites More sharing options...
JimCHLin Posted May 22, 2006 Posted May 22, 2006 A day in the life with 2 Raptors in raid, it is nothing but awful, and really hope DFI fixes this. Try not to connect WD disk drives on the same branch of power connector. To see whether this solution works for you. Share this post Link to post Share on other sites More sharing options...
Angry_Games Posted May 22, 2006 Posted May 22, 2006 Not seen anything later than 1.65...i will check though. Mine runs fine here though are you using the 1.65? ya, it is built into the bios I've just done the big build (making a tutorial vid with it) and I'll check again, but I am pretty sure the integrated memtest is 1.65. it's just weird that 2 different 2GB kits gave me exact same MB/s lol Share this post Link to post Share on other sites More sharing options...
trabpukcip Posted May 22, 2006 Posted May 22, 2006 Wow that is really a lot of errors I get two m5288 errors and three disk errors every two hours at the most! Must be because the Raptors are high speed devices. Share this post Link to post Share on other sites More sharing options...
raju Posted May 22, 2006 Posted May 22, 2006 Try not to connect WD disk drives on the same branch of power connector.To see whether this solution works for you. I can't imagine that working if the power connectors are on the same rail. Most psu's have all their molex and sata sharing the same rail regardless of the branch used. Share this post Link to post Share on other sites More sharing options...
JohnMike Posted May 22, 2006 Posted May 22, 2006 I can't imagine that working if the power connectors are on the same rail. Most psu's have all their molex and sata sharing the same rail regardless of the branch used. Not same rail, he means... different branch of power connector. Share this post Link to post Share on other sites More sharing options...
raju Posted May 22, 2006 Posted May 22, 2006 that's my point all branches are connected internally to the SAME RAIL, unless he uses a seperate psu for each Hard drive, which is hardly convienient. Meaning the power comes from the same devices internally in the psu, even if they're on a different branch.. 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