Super PI 32M 4ghz efficiency challenge - Page 51
only things left on my list to try:
trrsr @ 4
trrd @ 4 or less
.. you have to remember.. your time and mine are CL9.. both with tRCD 12 .. we are within like .3s of each other. I have to drop speed to run tRCD 11 to drop a second - 7m54s [which is 46.2ns in Maxxmem for me]..
CHL - Can you run Maxxmem and post your screen shot with those timings/speed please? Im curious to see if your below 45ns or not..
From what I am seeing using my HCH9's .. it would take +2800mhz for CL9 and tRCD 12 to be below 40ns [going by Maxxmem result .. i know Aida shows differently]
Edited by coolhandluke41 - 11/30/12 at 2:35am
Guess I know what I'm doing this weekend now.
Tested my CPU with a set of Ballistix that came in the mail recently ..doesnt look too promising.. I am beginning to think the IMC is the issue. Any speed/latency setting that results in Maxxmem reading lower than 46.8ns = unstable.. getting a time lower than 7m54s will depend on lower measured latency [I am using Maxxmem for reference].
2200mhz - 9-10-9-24 = 47.7ns which is ~7m56s
2300mhz - 9-10-9-24 = 46.8ns = unstable .. ram up to 1.88v/VTT up to 1.24v/SAv up to 1.24v
2400mhz - 9-11-11-17 = 47.2ns which is ~7m55s
2450mhz - 9-11-11-17 = 46.8ns = unstable .. ram up to 1.88v/VTT up to 1.24v/SAv up to 1.24v
.. hopefully this CPU doesnt let me down in clocks .. still need to test for max validation on LN sometime.
RipjawZ 2400 c8 http://www.gskill.com/products.php?index=451
Flare 2000 7-9-7-24 and 2000 9-9-9-24
Trident 2000 9-9-9-27 1.6v
Crucial Ballistix 2133 9-10-9-24 ( about 12 dimms left to test)
G.Skill ECO 1600 7-8-7-24 1.35v (6 dimms)
Ripjaw X 2133 8-9-8-24 (6 dimms:D and all can do 1300 c8 on air depending on voltage)
Ripjaws 1866 9-9-9-24
Trident 2000 9-9-9-24 1.65v
and of course a few kits of Samsung and Hynix for Haswell:D
I think I am pretty stocked tbh lol