Brutality Posted June 2, 2011 Posted June 2, 2011 (edited) Twice have I been forced to reduce overclock due to instability after a week of use. First time I got a 12h prime95 stable OC with the following settings: Ai Overclock Tuner: X.M.P. (for the dimms to have optimal timings and speed) BCLK: 100 Turbo Ratio: By Per Core Core Ratio Limit: 47 Internal PLL Overvoltage: Auto Memory Frequency: DDR3-2133MHz (according to X.M.P. profile) EPU Power Saving Mode: Disabled I didn't change anything in DRAM Timing Control, because I assume X.M.P. will handle that Load-line Calibration: Ultra-high VRM Frequency: Manual VRM Fixed Frequency Mode: 350 Phase Control: Extreme Duty Control: Extreme CPU Current Capability: 100% CPU Voltage: 1.32 DRAM Voltage: 1.65 (according to X.M.P. profile, and http://www.xtremesystems.org/forums/showthread.php?t=265735) VCCSA Voltage: Auto VCCIO Voltage: Auto CPU PLL Voltage: Auto PCH Voltage: Auto CPU Spread Spectrum: Disabled Then after a week or two I took a 12h prime95 test and it failed I then started again from scratch with overclocking and I couldn't get it 12h prime95 stable at 4.7 anymore, so I went with 4.6 with the following settings: Ai Overclock Tuner: X.M.P. (for the dimms to have optimal timings and speed) BCLK: 100 Turbo Ratio: By All Cores Core Ratio Limit: 46 Internal PLL Overvoltage: Auto Memory Frequency: DDR3-2133MHz (according to X.M.P. profile) EPU Power Saving Mode: Max Power Saving I didn't change anything in DRAM Timing Control, because I assume X.M.P. will handle that Load-line Calibration: Ultra-high VRM Frequency: Manual VRM Fixed Frequency Mode: 350 Phase Control: Extreme Duty Control: Extreme CPU Current Capability: 100% CPU Voltage: 1.315 DRAM Voltage: 1.65 (according to X.M.P. profile, and http://www.xtremesystems.org/forums/showthread.php?t=265735) VCCSA Voltage: Auto VCCIO Voltage: Auto CPU PLL Voltage: Auto PCH Voltage: Auto CPU Spread Spectrum: Disabled After a day of folding it wasn't 12h prime95 stable... Here is my procedure for overclocking btw: 1. Find highest overclock 2. Find lowest vCore for this OC 3. Try to get EPU working with this OC So is 12h not long enough for stress testing or is it because I use 1.65V dimms? Currently I am doing a p95 stress test with stock (or ASUS Optimized) settings cuz I am totally confused lol But unless anyone got some better ideas I am going to put my dimms to 1.5V with 1333MHz frequency and suitable timings and see if that will cause a degradation with the same overclocking procedure... Heck, what do I know about overclocking :/ Edited June 2, 2011 by Brutality Share this post Link to post Share on other sites More sharing options...
DanTheGamer11 Posted June 2, 2011 Posted June 2, 2011 Temps? Share this post Link to post Share on other sites More sharing options...
Brutality Posted June 2, 2011 Posted June 2, 2011 Temps? Oh yeah... never went above 65*C, so that is not the cause Share this post Link to post Share on other sites More sharing options...
DanTheGamer11 Posted June 2, 2011 Posted June 2, 2011 (edited) Oh yeah... never went above 65*C, so that is not the cause well thats really the only thing that causes degradation apart from crazy voltage... wait for all the Americans to wake up as they seem to be the majority on this website... I am noticing the better overclock was at 1.32 compared to the lesser overclock at 1.315... but is that too small a difference? Also I thought that SB could only use 1.5V ram? Edited June 2, 2011 by Dan The Gamer Share this post Link to post Share on other sites More sharing options...
Brutality Posted June 2, 2011 Posted June 2, 2011 (edited) well thats really the only thing that causes degradation apart from crazy voltage... wait for all the Americans to wake up as they seem to be the majority on this website... I am noticing the better overclock was at 1.32 compared to the lesser overclock at 1.315... but is that too small a difference? Also I thought that SB could only use 1.5V ram? Well after the first "degradation" it wouldn't even pass at 1.35V at 4.7, so I just went for 4.6 and got it stable agian there But yeah, my only thought of issue is the 1.65V dimms, although they are made for SB and seem to work given a stable OC, Intel has not recommended them... Thanks for your input though Dan =) Hehe, I bet you were waiting for me to post some high temps so you could point your finger and say, there! This really is a mystery to me too. Either it's not enough stress testing or the dimms - my guess Edited June 2, 2011 by Brutality Share this post Link to post Share on other sites More sharing options...
DanTheGamer11 Posted June 2, 2011 Posted June 2, 2011 (edited) Well after the first "degradation" it wouldn't even pass at 1.35V at 4.7, so I just went for 4.6 and got it stable agian there But yeah, my only thought of issue is the 1.65V dimms, although they are made for SB and seem to work given a stable OC, Intel has not recommended them... Thanks for your input though Dan =) Hehe, I bet you were waiting for me to post some high temps so you could point your finger and say, there! This really is a mystery to me too. Either it's not enough stress testing or the dimms - my guess I wan't really expecting you to say that you had high temps cause you seem pretty experienced... does it run stable at 1.33 @ 4.7? or whould that be too high? its 12:00 in some parts of America now so... were you using any motherboard tools to overclock? Would resetting the BIOS be too much? Edited June 2, 2011 by Dan The Gamer Share this post Link to post Share on other sites More sharing options...
Waco Posted June 2, 2011 Posted June 2, 2011 It's possible you just lucked out on your first Prime95 run. I tend to use Linpack as a quick test since it doesn't need much more than an hour or two to root out nearly all instability. After it's Linpack stable I move on to Prime95 for 12 or 24 hours. Sometimes you'll get a pass when you normally wouldn't by shear luck. Share this post Link to post Share on other sites More sharing options...
Nuclear Posted June 2, 2011 Posted June 2, 2011 It's possible you just lucked out on your first Prime95 run. I tend to use Linpack as a quick test since it doesn't need much more than an hour or two to root out nearly all instability. After it's Linpack stable I move on to Prime95 for 12 or 24 hours. Sometimes you'll get a pass when you normally wouldn't by shear luck. I'm agreeing with Waco on this one. You didn't run enough volts or get high enough temps through your 2500k to come close to causing degradation. I've have chips pass 8+HRS of prime95, and the next day fail after 30 minutes. I don't use Prime anymore and never use LinX over 25 runs. Prime95 fails to really stress these SB chips in a way that fits their common instability points and takes too long to find problems. I can run a quick 5 pass LinX test in under three minutes and decide from that what I need to tweak before going after a 20 run stability test. BTW: I have a 2600k that I've pushed as much as 1.6v through for hours and days under a water setup and it's still running just as strong as the day Intel sent it to me. It's vdimm as high as 1.72v, VTT at 1.3v, System Agent over 1.1v, and PLL at 1.9v. I beat and beat on that chip and it never falters. Share this post Link to post Share on other sites More sharing options...
Brutality Posted June 2, 2011 Posted June 2, 2011 (edited) I'm agreeing with Waco on this one. You didn't run enough volts or get high enough temps through your 2500k to come close to causing degradation. I've have chips pass 8+HRS of prime95, and the next day fail after 30 minutes. I don't use Prime anymore and never use LinX over 25 runs. Prime95 fails to really stress these SB chips in a way that fits their common instability points and takes too long to find problems. I can run a quick 5 pass LinX test in under three minutes and decide from that what I need to tweak before going after a 20 run stability test. BTW: I have a 2600k that I've pushed as much as 1.6v through for hours and days under a water setup and it's still running just as strong as the day Intel sent it to me. It's vdimm as high as 1.72v, VTT at 1.3v, System Agent over 1.1v, and PLL at 1.9v. I beat and beat on that chip and it never falters. Wow. Sandy Bridge really is freaking awesome... How many days have u had it? So what should I use for stability testing instead? Can you link me to the program and perhaps instruct me in its use? Btw, thanks guys. I am so happy that I really didn't degrade it, hehe - almost had a depression lol Edited June 2, 2011 by Brutality Share this post Link to post Share on other sites More sharing options...
IVIYTH0S Posted June 2, 2011 Posted June 2, 2011 Just try lowering the memory speed down and see if that is it, or raising your north bridge volts some (VCCSA I guess?? look it up) Share this post Link to post Share on other sites More sharing options...
Nuclear Posted June 2, 2011 Posted June 2, 2011 Just try lowering the memory speed down and see if that is it, or raising your north bridge volts some (VCCSA I guess?? look it up) VCCSA is System Agent Voltage...definitely don't won't to go playing with that unless you know what you are doing. For 5.0GHz I don't need to touch mine from the stock levels, so 4.7GHz should be okay at 0.925v. You can safely try up to 0.975v, but I wouldn't go that far yet. Sandy Bridge really is freaking awesome... How many days have u had it? So what should I use for stability testing instead? Can you link me to the program and perhaps instruct me in its use? Click here for LinX with AVX Support Download that and unpack it. Set it to use all of the memory you can and set the Run option to 20. I have 4GB of RAM and idle around 3100MB free...so I use 2800MB as my memory size and click start. If you have errors, it will either stop or result in a BSOD. If it completes all 20 passes, then you system is as stable as it gets and call it a day. Let me know if you have questions Share this post Link to post Share on other sites More sharing options...
Brutality Posted June 3, 2011 Posted June 3, 2011 VCCSA is System Agent Voltage...definitely don't won't to go playing with that unless you know what you are doing. For 5.0GHz I don't need to touch mine from the stock levels, so 4.7GHz should be okay at 0.925v. You can safely try up to 0.975v, but I wouldn't go that far yet. Click here for LinX with AVX Support Download that and unpack it. Set it to use all of the memory you can and set the Run option to 20. I have 4GB of RAM and idle around 3100MB free...so I use 2800MB as my memory size and click start. If you have errors, it will either stop or result in a BSOD. If it completes all 20 passes, then you system is as stable as it gets and call it a day. Let me know if you have questions Thanks =) How long does 20 passes take? and how many passes needed when "stepping up" (i.e. like 30 min p95) I assume the 20 pass is the "final stability test", like the 12h p95 test 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