laserred Posted December 14, 2006 Posted December 14, 2006 OK, it seems I'm stuck at the doorstep of 3G. I don't know how high I can push the voltage on my 165, is it only temperature-limited? Also, even if I drop my memory timings to 3-4-4-8 it doesn't seem to clock any higher. ~268MHz is what I can get to and not have memory problems. I'm going to stick a shot of A64Info cause I know that helps determine where I might be bottlenecked. Oh yeah- my board will boot with the 165 with 150 dividers- I'm seeing this is kinda unusual, just wondering. Thanks in advance! www.swansonsperformance.com/images/maxOC.jpg Share this post Link to post Share on other sites More sharing options...
CPDMF Posted December 14, 2006 Posted December 14, 2006 Max Async and read preamble are a little tight. Try 8.0 and 5.5, and move Tref to 3120 or 3684. My 165 does the 150 divider just fine as well.. Share this post Link to post Share on other sites More sharing options...
laserred Posted December 14, 2006 Posted December 14, 2006 CP, I was hoping you'd drop in. I had the Tref at what you had said before, but I also read that Oskar Wu recommended leaving it at Auto, which is how it's at 1296. I am definitely going to try 3120 and 3684, but can you give some insight to picking settings? I mean, if I use the formulas in the BIOS guide, I get 1/270=3.704, then 7800/3.704=2105.8, which would be much lower than 3120 or 3684, but obviously much higher than the mobo picked for itself. I'll loosen up those other settings and get back! Thx a bunch! Share this post Link to post Share on other sites More sharing options...
CPDMF Posted December 14, 2006 Posted December 14, 2006 I can tell you that there is some creative bios writing by Oskar Wu going on that allows and sometimes forces the boards to do things that do not follow formulas. I do not know enough about the internal writing of bioses to determine exactly why though. For Tref I throw any formula out the window if I'm not running at stoc speeds. After stock I tend to run on a theory of my own, that tells me the bios is written in clock cycles based on known values of 133 166 and 200, as I increase HTT I need to increase clock cycles to maintain the same amount of time between refresh intervals. So far it has worked on every LanParty board I own, nvidia or ati. Max Async and Read Preamble have more to do with the CPU than the memory. So these will need to be increased as HTT increases regardless of divider... Share this post Link to post Share on other sites More sharing options...
laserred Posted December 14, 2006 Posted December 14, 2006 Cool. It's actually worked some, as now I am into Windows at 275 memory speed, but it fails SuperPi right off the bat, even though everything so far in Windows is fine. Is this likely a voltage issue or a settings issue? I relaxed the settings to 3-4-4-10-14-17 as this is what Team says is their "special overclocking" SPD.. got me Anyway, I'll bump VDIMM back up to 2.8 as I had dropped it to 2.72 at the slower settings, and re-benchmark it. Thx! Share this post Link to post Share on other sites More sharing options...
Capt Proton Posted December 14, 2006 Posted December 14, 2006 Have you dropped the CPU Multi and tested to ensure that the Memory is stable? Share this post Link to post Share on other sites More sharing options...
laserred Posted December 14, 2006 Posted December 14, 2006 Yeah. It tested Memtest for Windows to 1500%, that was enough for me. It's also gone Prime stable for hours on end, I just don't think I'm going to hit 3GHz without water/phase cooling. The temps get high when I bump the VID up, had it at 1.58 just to see, and couldn't get it stable at 2970. Ah well, 2900 is still a pretty damn good overclock, so I will be happy with that. Working out the details to get it 100% stable. Share this post Link to post Share on other sites More sharing options...
dr_bowtie Posted December 15, 2006 Posted December 15, 2006 that chips is perfectly capable of 3G's at decent volts... If it takes more vcore than others are using it means you are pushing your ram too hard still...the looser the ram is at high fsb the easier it is on the memory controller and it will be more apt to make 3G's on less volts... just loosen it up more and get it to pass and see how it goes... I have found on my ram... that for a good stable run the ram must be memtest stable 15mhz higher than my final clock... so the ram is memtest stable at 315mhz and my final is 300x10 I know my ram will hit 320 stable and probably more.. Share this post Link to post Share on other sites More sharing options...
CPDMF Posted December 15, 2006 Posted December 15, 2006 It's not always a case of loosening timings to bring things around. With the 165's low multi and the need to use a divider no matter what ram you have loosening timings doesn't always work. I attached a screenie of what timings I have to use to be prime and all day game stable on the 166 divider with odd divisor correct enable because my RAM runs out at 272. Loosening timings on my setup will either fail prime or drop back to the desktop during BF2. Much easier to give your rig what it wants than to give it what you want. EDIT Don't use my timings, an extended CMOS clear might be in order if you do. Different board with different capabilities.. Share this post Link to post Share on other sites More sharing options...
laserred Posted December 15, 2006 Posted December 15, 2006 CP< I am guessing the .62 Beta you are using scans the system bus? I DL'ed it, and as soon as I clicked on it, I got a reboot, just like SANDRA. How do I disable the system scan for A64Info? Thx! Share this post Link to post Share on other sites More sharing options...
CPDMF Posted December 15, 2006 Posted December 15, 2006 I was pretty sure I used that prog on my expert, but not as sure now. When I get a chance I'll extract it here at work and see if the config file can be edited.. Share this post Link to post Share on other sites More sharing options...
laserred Posted December 20, 2006 Posted December 20, 2006 Hey CP, I had dropped my multi to 150, and at 318 it still booted and ran Windows fine. So I decided to get crazy, and cranked the FSB to 334 to see if I could get a boot at 3GHz. Well, it rebooted, and hung on the "Verify DMI pool" in the POST screen. I still have the LDT @ 1.2V and the Northbridge at 1.52v, which is the likely culprit that will allow my board to boot at 330+, or is this still some RAM timing problem? Thx! 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