Question For Nickn Or Oc'ing Experts - Finishing An Oc

[cmpbellsjc 2,960]

Your CPU voltage is almost certainly too low if set to 1.32V, especially once you take vdroop and normal board variations into account. Not many Wolfdale chips will hit such high clocks on such low volts, despite what some extreme overclockers may report on some forums (not speaking about this one). I run between 1.44-1.46V on my E8400 to achieve 4GHz. I do not recommend you run this voltage, as it will definitely shorten the lifespan of your CPU. Bump your voltage up a notch and test again. 1.35V is the highest you should go to stay within Intel recommendations, 1.45V is the maximum the manufacturing process used to produce your processor is designed to handle. Also, switch to OCCT for stress-testing. It will find errors much quicker than Prime95 (unless your o/c is horribly unstable, which it does not appear to be). As an added bonus, OCCT displays the CPU voltage in real-time so you can see what the voltage is actually reading, not just what it's set to in the BIOS. On my board I have vdroop of .2-.4V which is pretty bad, but pretty normal for P35 chipset boards I've seen without any LLC feature.

Yeah I was set at 1.4 CPU Volts when I couldn't pass the test, not 1.35. I was at 1.35 when I was at 3.6 ghz. I was set to Jeff's mobo setting when I couldn't pass prime. Being set at 1.4v with LLC enabled I was running at 1.38 volts. Thanks for the link for OCCT.

[cmpbellsjc 2,960]

Yes sir.. the UTX issue with their 1.3 patch is a bugger. I do not run any higher than 8% car traffic even on my systems after that patch. My systems can run the traffic that low but many will find they must disable the UTX Moving Traffic option in the UTX tool in order to gain back that CPU and memory resource especially flying heavy metal in larger hubsThey added traffic to side streets with that patchAs for your instability, as mentioned you will most likely need to trim some settings. AI Clock twister.. set it to LIGHTEST and see if that makes a difference first. Also, if the BIOS has the setting DRAM STATIC READ CONTROL set it to DISABLEDSee if those do not trim in stabilityNB @ 1.55 is probably good however there are some that may need to increase that to 1.6.. I would however be careful above 1.55 and check NB temp. If it remains below 60c under a full load its OK. CPU PLL and FSB Termination are usually ok on AUTO with dual core procs... same with GTL voltage trim.... typcially only quads need trim on those settings and I would leave them on Auto. Sometimes a touch of Vcore helps and when clocking Load Line Calibration in the BIOS should be ENABLED which dampens the effect Vdroop has and allows what is set in the BIOS for Vcore to remain a constant. That means you do NOT overshoot Vcore in the BIOS to obtain a lower value in Windows... ENABLE LLC will help stabilize a clock but you must also make sure your Vcore setting is not going to exceed the proc max. I would need to look up the values for the proc as I do not have any clocking notes on it here.

Last night after I posted this and before I got your reply I reset the RAM back to the stock 5-5-5-15 and went from 8.5 X 450 back to 8 X 450 for 3.6 ghz with the RAM still set to 1081. Maybe I should stick with 3.6 ghz and not try bothering with 3.85 ghz. There probably isn't much difference between 3.6 ghz and 3.85 ghz is there? Maybe 1 FPS increase? If I stay at 3.6 with the RAM at the 1081mhz and the 333 strap with 450 FSB do you think or have any recommendations of tightening the RAM settings? I will leave it alone for now until I hear back from you. Oh, one other question before I forget. If I set the RAM timings in bios do I need to fiddle with memset? I noticed that if I make a change to the settings in memset they dont seem to show up in bios.

[TechguyMaxC 516]

Yeah I was set at 1.4 CPU Volts when I couldn't pass the test, not 1.35. I was at 1.35 when I was at 3.6 ghz. I was set to Jeff's mobo setting when I couldn't pass prime. Being set at 1.4v with LLC enabled I was running at 1.38 volts. Thanks for the link for OCCT.

Can you turn your GTL voltage down a notch and test again? Some 45nm chips seem to like lower than reference GTL voltage.

Last night after I posted this and before I got your reply I reset the RAM back to the stock 5-5-5-15 and went from 8.5 X 450 back to 8 X 450 for 3.6 ghz with the RAM still set to 1081. Maybe I should stick with 3.6 ghz and not try bothering with 3.85 ghz. There probably isn't much difference between 3.6 ghz and 3.85 ghz is there? Maybe 1 FPS increase? If I stay at 3.6 with the RAM at the 1081mhz and the 333 strap with 450 FSB do you think or have any recommendations of tightening the RAM settings? I will leave it alone for now until I hear back from you.

The best advice you're going to get for adjusting your RAM timings is at XtremeSystems, in this thread particularly as it deals with your RAM specifically.

Oh, one other question before I forget. If I set the RAM timings in bios do I need to fiddle with memset? I noticed that if I make a change to the settings in memset they dont seem to show up in bios.

This is by design. Memset does not adjust BIOS values because if you set something improperly and your system hangs (and restarts) it will just keep hanging at boot.edit: typo

[Guest Nick_N]

Last night after I posted this and before I got your reply I reset the RAM back to the stock 5-5-5-15 and went from 8.5 X 450 back to 8 X 450 for 3.6 ghz with the RAM still set to 1081. Maybe I should stick with 3.6 ghz and not try bothering with 3.85 ghz. There probably isn't much difference between 3.6 ghz and 3.85 ghz is there? Maybe 1 FPS increase? If I stay at 3.6 with the RAM at the 1081mhz and the 333 strap with 450 FSB do you think or have any recommendations of tightening the RAM settings? I will leave it alone for now until I hear back from you. Oh, one other question before I forget. If I set the RAM timings in bios do I need to fiddle with memset? I noticed that if I make a change to the settings in memset they dont seem to show up in bios.

The memory you are running is pretty much locked at stock timing and probably will not go over 1093 @ 2.1v which is why I posted what I did about the corsair sticks. 1081 @ 5-5-5 is pretty much their hard limit. You wont get any more out of those.If you are stable at 3.6 call that your base for 450MHz. In order to go higher I would say it will require more vcore over any other setting but do check the 2 I posted and set them as such. The E84000 is known for hitting 4GHz and I have even seen reports of 4.2-4.4 so the process0r will do 3.8 if it is a E0 slugThe difference between 3.6 and 3.8 is what I would call a nice little bump and if its possible stable and safe I would try for it. Its not going to equate to frames and this 'more frames is better' thing everyone talks about is not what its about... its about improving the visuals and how smooth the flight flows. In that you can be at 24-27 frames and never need one frame more. I have seen 45 frames look terrible with jitters and 28 smooth as glass. FS9 was much more sensitive to higher frame rate need than FSXSo yes 200Mhz would be nice to get if its possible but it will not provide what 4-4.5 doesThere are 2 steppings in that processor... C0 and E0E0 is the high clocker/low voltage... Some C0 chips wont reach the high clocks, some will but require 1.4v+ to do 4GHz so if CPUz shows stepping as C0 that icould be why you may be limited.If this is a E0 chip it should not have any problem with 3.8GHz @ somewhere between 1.34-1.40v and do be aware I consider the hard limit on a 45nm CPU with Vcore as 1.42v and would never run one higher than that. Check your stepping in CPUz and let me know what it is.and do make sure C1E/C2E and all other CPU advanced BIOS settings ere disabled (except PECI if its listed, leave enabled)If you are not using Memset to SAVE a change for Windows boot then Memset will override the BIOS at Windows boot only... Memset is only needed for testing and do note you can not change CMD or CAS in Memset on the fly, only sub timing values. As Max mentioned anything you change in Memset will not effect the BIOS and only remain active for Windows until a reboot unless you selet to SAVE the settings which will enable them at boot. Otherwise Memset is not needed if the same settings are used in the BIOS

[Guest Jeffhew]

Sean,That thread that Max mentioned at the Xtreme Systems site is the one where I got the initial settings for this RAM when I started overclocking my system. It's a long thread but, a good read. I have a thread there as well about the specific combination we're using and, while I didn't keep updating there too long, there are a few tips that may or may not help you. http://www.xtremesystems.org/forums/showthread.php?t=197097 . My setup is quite different now from the pics in that thread when I was starting out. Sorry it didn't stay stable for you. I ran OCCT on mine today and it performed fine. My board has the C0 stepping so I don't push beyond 1.4v. On a whim, I loosened my timings back to 5-5-5-15 and pushed the FSB to see how far I could go. I didn't touch the vCore, and bumped the NB up to 1.57. My best was 485 x 8.5 for 4.1 GHz but I couldn't run the sim there. I pulled back to a little less than 3.9 with an 8 multi and had a real nice flight. RAM ran at 1164 MHz with 5-5-5-15 @ 2.2v. I'm not saying you should do this. I'm just letting you know for reference. Nick has mentioned that these sticks should be at 2.1v. My specs say that these are rated to run 533MHZ at 2.2v so Nick, could you help me out with that one? I don't doubt your word, I'm just confused about the discrepancy. JEDEC @ 270 and 400 shows 1.8v and EPP @ 533 shows 2.2v. At any rate Sean, keep pluggin'. If you can get it stable up around 3.8 GHz, you will see an improvement. Like Nick mentioned, it's not so much faster as it is smoother. If ultimately you can't get there, 3.6 GHZ is nothing to be ashamed of. It's a nice overclock.Sorry I don't have more answers but I know Nick and Max do so, good luck!Jeff*EDIT* Incidently Nick, do you have any suggestions for improving the data transfer from the hard drives short of running my system faster? Are there any other settings, outside of OC'ing the system further, that will help maximize throughput? Any ideas you could offer would be appreciated. Thanks.

[cmpbellsjc 2,960]

Sean,That thread that Max mentioned at the Xtreme Systems site is the one where I got the initial settings for this RAM when I started overclocking my system. It's a long thread but, a good read. I have a thread there as well about the specific combination we're using and, while I didn't keep updating there too long, there are a few tips that may or may not help you. http://www.xtremesystems.org/forums/showthread.php?t=197097 . My setup is quite different now from the pics in that thread when I was starting out. Sorry it didn't stay stable for you. I ran OCCT on mine today and it performed fine. My board has the C0 stepping so I don't push beyond 1.4v. On a whim, I loosened my timings back to 5-5-5-15 and pushed the FSB to see how far I could go. I didn't touch the vCore, and bumped the NB up to 1.57. My best was 485 x 8.5 for 4.1 GHz but I couldn't run the sim there. I pulled back to a little less than 3.9 with an 8 multi and had a real nice flight. RAM ran at 1164 MHz with 5-5-5-15 @ 2.2v. Nick has mentioned that these sticks should be at 2.1v. My specs say that these are rated to run 533MHZ at 2.2v so Nick, could you help me out with that one? I don't doubt your word, I'm just confused about the discrepancy. JEDEC @ 270 and 400 shows 1.8v and EPP @ 533 shows 2.2v. At any rate Sean, keep pluggin'. If you can get it stable up around 3.8 GHz, you will see an improvement. Like Nick mentioned, it's not so much faster as it is smoother. If ultimately you can't get there, 3.6 GHZ is nothing to be ashamed of. It's a nice overclock.Sorry I don't have more answers but I know Nick and Max do so, good luck!Jeff

Hey Jeff. I was just getting ready to reply to Nicks remarks when I saw you posted. Yeah, I started get errors on prime, although I am going to download OOCT and start using it. I got to a funny point that after it would fail prime when I restarted the computer it would reboot, but only to a black screen. The keyboard lights would be on, but nothing on the screen. The only way I could get back into BIOS was by powering down. Needless to say, I started getting worried and set everything back to 9 X 400 with the 400 strap and 1066 ram speed. I will continue to plug away, however scary as it is. Well, really the only scary thing was that under the prime95 blend test, which I keep failing, is that the NB temps got to 55-56 degrees. for some reason, even with all the fans that NB temp seems high to me, but Nick said that as long as it stays under 60 it should be ok. The only thing is, when I am flying I can't keep exiting FSX to check temps, although I don't know that FSX generates temps like a prime95 blend test does.Well, I am going to reply to Nick now. I will keep you posted.Thanks - Sean

[Guest Jeffhew]

Hey Jeff. I was just getting ready to reply to Nicks remarks when I saw you posted. Yeah, I started get errors on prime, although I am going to download OOCT and start using it. I got to a funny point that after it would fail prime when I restarted the computer it would reboot, but only to a black screen. The keyboard lights would be on, but nothing on the screen. The only way I could get back into BIOS was by powering down. Needless to say, I started getting worried and set everything back to 9 X 400 with the 400 strap and 1066 ram speed. I will continue to plug away, however scary as it is. Well, really the only scary thing was that under the prime95 blend test, which I keep failing, is that the NB temps got to 55-56 degrees. for some reason, even with all the fans that NB temp seems high to me, but Nick said that as long as it stays under 60 it should be ok. The only thing is, when I am flying I can't keep exiting FSX to check temps, although I don't know that FSX generates temps like a prime95 blend test does.Well, I am going to reply to Nick now. I will keep you posted.Thanks - Sean

I've found that when the system posts fine but won't boot to Windows, a notch of vCore will solve the problem. Don't go beyond 1.4v though. As for the NB temps, I've seen mine as high as 65c on a really warm day with no adverse effects but, yeah, keep it as cool as you can.Talk to ya later,Jeff

[cmpbellsjc 2,960]

The memory you are running is pretty much locked at stock timing and probably will not go over 1093 @ 2.1v which is why I posted what I did about the corsair sticks. 1081 @ 5-5-5 is pretty much their hard limit. You wont get any more out of those.If you are stable at 3.6 call that your base for 450MHz. In order to go higher I would say it will require more vcore over any other setting but do check the 2 I posted and set them as such. The E84000 is known for hitting 4GHz and I have even seen reports of 4.2-4.4 so the process0r will do 3.8 if it is a E0 slugThe difference between 3.6 and 3.8 is what I would call a nice little bump and if its possible stable and safe I would try for it. Its not going to equate to frames and this 'more frames is better' thing everyone talks about is not what its about... its about improving the visuals and how smooth the flight flows. In that you can be at 24-27 frames and never need one frame more. I have seen 45 frames look terrible with jitters and 28 smooth as glass. FS9 was much more sensitive to higher frame rate need than FSXSo yes 200Mhz would be nice to get if its possible but it will not provide what 4-4.5 doesThere are 2 steppings in that processor... C0 and E0E0 is the high clocker/low voltage... Some C0 chips wont reach the high clocks, some will but require 1.4v+ to do 4GHz so if CPUz shows stepping as C0 that icould be why you may be limited.If this is a E0 chip it should not have any problem with 3.8GHz @ somewhere between 1.34-1.40v and do be aware I consider the hard limit on a 45nm CPU with Vcore as 1.42v and would never run one higher than that. Check your stepping in CPUz and let me know what it is.and do make sure C1E/C2E and all other CPU advanced BIOS settings ere disabled (except PECI if its listed, leave enabled)If you are not using Memset to SAVE a change for Windows boot then Memset will override the BIOS at Windows boot only... Memset is only needed for testing and do note you can not change CMD or CAS in Memset on the fly, only sub timing values. As Max mentioned anything you change in Memset will not effect the BIOS and only remain active for Windows until a reboot unless you selet to SAVE the settings which will enable them at boot. Otherwise Memset is not needed if the same settings are used in the BIOS

Hi Nick. I went into CPU-z and checked the box under the box titled "stepping" that said "revision" and it says C0. In my advanced bios settings I have these options and this is how they are set:C1E Support - DisbaledCPU TM Fuction - EnabledVanderpool tech - EnabledExecute Disable Bit - EnabledMAX CPUID value Limit - DisabledSo I should disable all these settings? And if so, should these always remain disabled regardless of the clock that I am going for?Ok, assuming I dont touch any of the RAM settings, I can go back to 8.5 X 450 and try to get that stable. If I leave the RAM untouched at 5-5-5-15 should I go for the 333 strap with either DRAM 901 mhz or 1081 OR the 400 strap at DRAM 900? Which ever one you recommend I try what limit should I use on the NB voltage considering that mine seems to run hotter than most even with 4 fans?Another thing that had happened was that after runing the prime95 blend test I noticed that the NB temps would get to around 55-56 degrees pretty fast, but not any higher than it failed. But the weird thing was that when I would restart the computer it would hang in a black screen after Windows shut down with the key board lights still on. I would have to power off then restart. Upon restart the message would come on about the failed OC.Lastly before I go to bed, should I always leave DRAM STATIC READ CONTROL set to DISABLED and the AI Clock Twister on Auto or on light, if I keep the 5-5-51 timings?Thanks in advance, off to bed. I will check back tomorrow night when I get home.

[TechguyMaxC 516]

Hi Nick. I went into CPU-z and checked the box under the box titled "stepping" that said "revision" and it says C0. In my advanced bios settings I have these options and this is how they are set:C1E Support - DisbaledCPU TM Fuction - EnabledVanderpool tech - EnabledExecute Disable Bit - EnabledMAX CPUID value Limit - DisabledSo I should disable all these settings? And if so, should these always remain disabled regardless of the clock that I am going for?

TM = Thermal Monitoring, a feature you definitely do not want to disable unless running colder-than-ambient (i.e. water or phase-change) cooling. Vanderpool is the codename for Intel's Virtualization technology. If you don't run any virtual machine software, you don't need it. Execute disable bit is a technology supported since XP SP2 which allows the OS to stop execution of a thread if it is flagged as malicious. I suggest you leave this enabled. The Max CPUID limit is to limit the amount of CPU cores the OS sees. Definitely leave that disabled.

Ok, assuming I dont touch any of the RAM settings, I can go back to 8.5 X 450 and try to get that stable. If I leave the RAM untouched at 5-5-5-15 should I go for the 333 strap with either DRAM 901 mhz or 1081 OR the 400 strap at DRAM 900?

If you want to continue to attempt to overclock, go for the lowest base memory speed and start to slowly adjust the FSB upwards.

Which ever one you recommend I try what limit should I use on the NB voltage considering that mine seems to run hotter than most even with 4 fans?

This is because you're running more voltage on the NB than is likely necessary. Default NB voltage for Intel 3/4 series chipsets is only 1.1V. I only need 1.3-1.4V on my Gigabyte P35 board to reach a 2GHz FSB (same with the Abit P35 board I had previously). No way do you need the 1.49-1.55V you've been running to hit 1.8GHz.

Another thing that had happened was that after runing the prime95 blend test I noticed that the NB temps would get to around 55-56 degrees pretty fast, but not any higher than it failed.

Again, too much voltage. The chip is leaking like crazy right now because it wasn't designed to handle the voltage being pumped through it. All that extra power has to go somewhere and that ends up being heat.

But the weird thing was that when I would restart the computer it would hang in a black screen after Windows shut down with the key board lights still on. I would have to power off then restart. Upon restart the message would come on about the failed OC.

As Jeff said, likely need more CPU voltage.

Lastly before I go to bed, should I always leave DRAM STATIC READ CONTROL set to DISABLED and the AI Clock Twister on Auto or on light, if I keep the 5-5-51 timings?

The answers you seek are at XtremeSystems. Definitely leave the Static Read Control disabled. AI Clock Twister could go either way. Enabling it could help extract better RAM performance, especially if running your RAM below its rated speed. Hope this helps,Max

[Guest Nick_N]

Sean,That thread that Max mentioned at the Xtreme Systems site is the one where I got the initial settings for this RAM when I started overclocking my system. It's a long thread but, a good read. I have a thread there as well about the specific combination we're using and, while I didn't keep updating there too long, there are a few tips that may or may not help you. http://www.xtremesystems.org/forums/showthread.php?t=197097 . My setup is quite different now from the pics in that thread when I was starting out. Sorry it didn't stay stable for you. I ran OCCT on mine today and it performed fine. My board has the C0 stepping so I don't push beyond 1.4v. On a whim, I loosened my timings back to 5-5-5-15 and pushed the FSB to see how far I could go. I didn't touch the vCore, and bumped the NB up to 1.57. My best was 485 x 8.5 for 4.1 GHz but I couldn't run the sim there. I pulled back to a little less than 3.9 with an 8 multi and had a real nice flight. RAM ran at 1164 MHz with 5-5-5-15 @ 2.2v. I'm not saying you should do this. I'm just letting you know for reference. Nick has mentioned that these sticks should be at 2.1v. My specs say that these are rated to run 533MHZ at 2.2v so Nick, could you help me out with that one? I don't doubt your word, I'm just confused about the discrepancy. JEDEC @ 270 and 400 shows 1.8v and EPP @ 533 shows 2.2v. At any rate Sean, keep pluggin'. If you can get it stable up around 3.8 GHz, you will see an improvement. Like Nick mentioned, it's not so much faster as it is smoother. If ultimately you can't get there, 3.6 GHZ is nothing to be ashamed of. It's a nice overclock.Sorry I don't have more answers but I know Nick and Max do so, good luck!Jeff*EDIT* Incidently Nick, do you have any suggestions for improving the data transfer from the hard drives short of running my system faster? Are there any other settings, outside of OC'ing the system further, that will help maximize throughput? Any ideas you could offer would be appreciated. Thanks.

If the sticks are rated for 2.2v and the manufacture posts they are safe up to 2.2v, then use 2.2vNo matter what the JEDEC spec is the manufacture of the sticks makes that call, not JEDEC They design circuits on the PCB layer which define what they will allow and warranteeSome manufactures do not specify a higher voltage even though sticks may be able to run that voltage... typically that comes with the standard memory products which the manufacture does not expect the user to clock. High performance memory will usually come with the correct max spec from the manufacture. Storage performance is based on the drive/controller/driver. Other than making sure you are using the best drive, the right controller and a good driver/firmware defrag stradegy with no partitions is the only other way to obtain perf. There are reg changes that can be made for network systems and SCSI RAID however those will not help with any typical home or gaming storage system

[Guest Nick_N]

Hi Nick. I went into CPU-z and checked the box under the box titled "stepping" that said "revision" and it says C0. In my advanced bios settings I have these options and this is how they are set:C1E Support - DisbaledCPU TM Fuction - EnabledVanderpool tech - EnabledExecute Disable Bit - EnabledMAX CPUID value Limit - DisabledSo I should disable all these settings? And if so, should these always remain disabled regardless of the clock that I am going for?Ok, assuming I dont touch any of the RAM settings, I can go back to 8.5 X 450 and try to get that stable. If I leave the RAM untouched at 5-5-5-15 should I go for the 333 strap with either DRAM 901 mhz or 1081 OR the 400 strap at DRAM 900? Which ever one you recommend I try what limit should I use on the NB voltage considering that mine seems to run hotter than most even with 4 fans?Another thing that had happened was that after runing the prime95 blend test I noticed that the NB temps would get to around 55-56 degrees pretty fast, but not any higher than it failed. But the weird thing was that when I would restart the computer it would hang in a black screen after Windows shut down with the key board lights still on. I would have to power off then restart. Upon restart the message would come on about the failed OC.Lastly before I go to bed, should I always leave DRAM STATIC READ CONTROL set to DISABLED and the AI Clock Twister on Auto or on light, if I keep the 5-5-51 timings?Thanks in advance, off to bed. I will check back tomorrow night when I get home.

Your proc stepping is the C0 version and therefore results will be limited. It will require a higher vcore than the E0 stepping proc to obtain a clock. With C0 you may be limited to 3.6-3.8GHz. Some C0 procs would do 3.8-4, some would not.Since you are not experienced in clocking, leave TM enabled the rest can be shut down, however this does affect TM1/TM2 functions and I do shut this down till I establish my clocks, then enable it to confirm stability. I also know the limits of my processors and monitor during stress testingOCCT has a setting in its properties which allows you to set the temp where the test will fail and shut down automatically.Execute Disable Bit is up to you.. I disable it however it wont interfere with the clockYou have 2 options to clock... 400 (and bumped higher than 400) and 450FSB. The STRAP sets the memory math and you dont want the sticks running 900. You wont get those sticks stable above 1092 @ 5-5-5 so that is your window to work with on the memory no matter what CPU speed ... 1066-1090MHz 5-5-5-15. tRAS, the last value, can be increased to 18 in order to check for memory stability in the 450FSB clock however if possible it should be run at the stock timingleave the 2 settings for Twister and DRAM STATIC READ I posted as they are For 450FSB set timing @ 5-5-5 2T @ 333 STRAP DDR2 1081 If these are the sticks you are using: http://www.newegg.com/Product/Product.aspx...N82E16820145197they have a rating of 2.1v however if Corsair approves them up to 2.2v then you can use that but you must verify that with Corsair. See RamGuy in the House of Help Corsair forum before attempting 2.2v I have not seen anyone who run those sticks need 2.2v to hit 1081 5-5-5As I motioned above your NB @ 1.55v is what I consider max without special cooling. I always try for the lowest NB voltage possible to run the stable clock. @ 3.82GHz 450FSB 1081 you may need to run 1.47-1.55v however @ 400 MHz I see no reason to run NB @ 1.55 and most likely it should run stable @ 1.2-1.35 or AUTO on 400MHzVcore in the BIOS with LLC ENABLED @ 450MHz should be somewhere between 1.35-1.42MAXIf the proc wont run that stable then your C0 proc is simply not going to allow the clock and remain in a safe Vcore range, period.At that point 400FSB is your only option and nudge it up in increments to hit the highest stable FSB above 400 on the STRAP that keeps the memory between 1066 and 1090 5-5-5 (333 or 400)Vcore and NB can come down at that point however you will need to find the right values that work. I would start by setting NB as a base for finding Vcore/memspeed stable at 1.45 and then once I find my highest stable FSB above 400 @ vcore, start dropping NB until stability test fails. @ Fail then nudge NB back up a bit to lock the clock and that's the best you are going to do.

[Guest Jeffhew]

If the sticks are rated for 2.2v and the manufacture posts they are safe up to 2.2v, then use 2.2vNo matter what the JEDEC spec is the manufacture of the sticks makes that call, not JEDEC They design circuits on the PCB layer which define what they will allow and warranteeSome manufactures do not specify a higher voltage even though sticks may be able to run that voltage... typically that comes with the standard memory products which the manufacture does not expect the user to clock. High performance memory will usually come with the correct max spec from the manufacture. Storage performance is based on the drive/controller/driver. Other than making sure you are using the best drive, the right controller and a good driver/firmware defrag stradegy with no partitions is the only other way to obtain perf. There are reg changes that can be made for network systems and SCSI RAID however those will not help with any typical home or gaming storage system

Thanks for the clarification Nick. Much appreciated.Jeff

[cmpbellsjc 2,960]

Your proc stepping is the C0 version and therefore results will be limited. It will require a higher vcore than the E0 stepping proc to obtain a clock. With C0 you may be limited to 3.6-3.8GHz. Some C0 procs would do 3.8-4, some would not.Since you are not experienced in clocking, leave TM enabled the rest can be shut down, however this does affect TM1/TM2 functions and I do shut this down till I establish my clocks, then enable it to confirm stability. I also know the limits of my processors and monitor during stress testingOCCT has a setting in its properties which allows you to set the temp where the test will fail and shut down automatically.Execute Disable Bit is up to you.. I disable it however it wont interfere with the clockYou have 2 options to clock... 400 (and bumped higher than 400) and 450FSB. The STRAP sets the memory math and you dont want the sticks running 900. You wont get those sticks stable above 1092 @ 5-5-5 so that is your window to work with on the memory no matter what CPU speed ... 1066-1090MHz 5-5-5-15. tRAS, the last value, can be increased to 18 in order to check for memory stability in the 450FSB clock however if possible it should be run at the stock timingleave the 2 settings for Twister and DRAM STATIC READ I posted as they are For 450FSB set timing @ 5-5-5 2T @ 333 STRAP DDR2 1081 If these are the sticks you are using: http://www.newegg.com/Product/Product.aspx...N82E16820145197they have a rating of 2.1v however if Corsair approves them up to 2.2v then you can use that but you must verify that with Corsair. See RamGuy in the House of Help Corsair forum before attempting 2.2v I have not seen anyone who run those sticks need 2.2v to hit 1081 5-5-5As I motioned above your NB @ 1.55v is what I consider max without special cooling. I always try for the lowest NB voltage possible to run the stable clock. @ 3.82GHz 450FSB 1081 you may need to run 1.47-1.55v however @ 400 MHz I see no reason to run NB @ 1.55 and most likely it should run stable @ 1.2-1.35 or AUTO on 400MHzVcore in the BIOS with LLC ENABLED @ 450MHz should be somewhere between 1.35-1.42MAXIf the proc wont run that stable then your C0 proc is simply not going to allow the clock and remain in a safe Vcore range, period.At that point 400FSB is your only option and nudge it up in increments to hit the highest stable FSB above 400 on the STRAP that keeps the memory between 1066 and 1090 5-5-5 (333 or 400)Vcore and NB can come down at that point however you will need to find the right values that work. I would start by setting NB as a base for finding Vcore/memspeed stable at 1.45 and then once I find my highest stable FSB above 400 @ vcore, start dropping NB until stability test fails. @ Fail then nudge NB back up a bit to lock the clock and that's the best you are going to do.

Thanks Nick, I will try and give it another go tonight and report back. Thanks guys for all the help. I wish one of you lived here in the Dallas, I would offer you a nice steak dinner or what ever you desire if you could come over and work your OC magic : :(

[cmpbellsjc 2,960]

Hey Nick and Jeff. I had gotten busy since my last post. Thursday and Friday I didn't have a chance to do anything. Last night after the college football games ended I got back a this. Below are my main settings:Extreme Tweaker Ai Overclock Tuner : Manual CPU Ratio Control : 8.5 FSB Frequency : 450FSB Strap to North Bridge : 333PCI-E Frequency: 100 DRAM Frequency: DDR2- 1081 DRAM Command Rate : 2T DRAM Timing Control: Manual CAS# Latency : 5 RAS# to CAS# Delay : 5 RAS# Precharge : 5 RAS# ActivateTime : 15 RAS# to RAS# Delay : Auto Row Refresh Cycle Time : Auto Write Recovery Time : Auto Read to Precharge Time : Auto Read to Write Delay (S/D) : Auto Write to Read Delay (S) : Auto Write to Read Delay (D) : Auto Read to Read Delay (S) : Auto Read to Read Delay (D) : Auto Write to Write Delay (S) : Auto Write to Write Delay (D) : Auto DRAM Static Read Control: DisabledAi Clock Twister : Auto Transaction Booster : Auto CPU Voltage : 1.375 CPU PLL Voltage : AutoNorth Bridge Voltage : 1.49 DRAM Voltage : 2.1 FSB Termination Voltage : Auto South Bridge Voltage : Auto Loadline Calibration : Enabled CPU GTL Reference : Auto North Bridge GTL Reference : Auto DDR2 Channel A REF Voltage : Auto DDR2 Channel B REF Voltage : Auto DDR2 Controller REF Voltage : Auto SB 1.5V Voltage : Auto NB LED Selection : NB Volt SB LED Selection : SB Volt CPU LED Selection : CPU VOlt Voltiminder LED : Auto CPU Spread Spectrum : DisabledPCIE Spread Spectrum : DisabledI downloaded and ran two OCCT tests. One was the Auto 1 hour mix test. It passed that no problem. CPU never exceded 51 degrees and the NB never exceed 52 degrees. Then I ran the Custom (Infinite) set to RAM for an hour. The NB was a little hotter, got to 53 degrees, but I guess it didn't pick up any errors since the test would have kept running had I not stopped it.If you noticed in my settings above I did disable DRAM Static Read Control, but I left the AI Clock Twister set to auto and the Transaction Booster set to auto, just to see if I could get stable at those settings. One reason I left Transaction Booster set to auto is because if I disable it an option comes up to set a Relax Level. It defaults to 0, but gives me the option to set it between 0 and 7. I was unsure of what these values do, so I left it on auto.After the two OCCT tests, I fired up FSX and tested aroung the KMIA area in the LDS 767 VC and the Aerosoft F-16 as these seem to be my most demanding two aircraft that I own. Flying around KMIA in the LDS VC I maintained 25-30 FSP with 80% AI and crisp textures and a thunderstorm in place. In the F-16 VC I had 28-40 FPS with crips textures and the same weather. Also in the F-16 I was able to fly 1000 to 2000 feet AGL at 450 - 550 kts and still not get any blurry textures. All in all, at 3.85 ghz and the RAM at 1081 mhz, even with my stock timings I am pleased. Perhaps, I could get it a little better by messing with the timings, but at this point the performance is really good and I don't want to take the time or risk of trying to get more out of it. Running at 3.85 ghz is way faster than I ever thought I would try to run this CPU, especially since my original goal when I first started out was to get only 3.4 ghz and the stock 1066. 3.85 ghz is a 27.5% increase over my stock clock, so thats pretty good for free.I guess my last question is, is leaving the AI Clock Twister set to Auto and Transaction Booster set to Auto ok? As long as the AI Clock Twister set to auto and the Transaction Booster set to auto isn't causing me any harm I am ok leaving it there since I am pleased with the performance now. Oh, one last thing I noticed that OCCT seems to measure the CPU temps about 5 degrees higher than Real Temp 2.70 and my on board PC Probe. Should I disregard the OCCT temps?Thanks again Nick and Jeff for the help, I will wait to see what you guys say about my last few questions I posted in this thread and I guess I will be done for now.Regards,Sean

[Guest Jeffhew]

Hi Sean,Transaction Booster is the setting that changes the PL number. For example, if you see a PL of 7 in memset, changing the transaction booster to relax -1 would give you a PL of 8. If you had it set to enabled and boost 1, you would get a PL of 6. Essentially this setting helps to improve memory bandwidith and reduce latency. The lower the number, the higher the performance. Setting it to boost means you'll need a bit more voltage. With our RAM the PL isn't going any lower than 7. It's ok to leave DRAM Static Read Control in auto as Transaction booster manipulates the static read control delay directly anyway. Transaction Booster in auto will let the BIOS set what it thinks it needs based on your RAM speed and timings. Manual settings just give you more control and lets you eek out all the performance you can from your system, as it is set up.Ai Clock Twister lets you get some smaller, incremental performance gains at the PL you're set at. This is one setting where you will see the biggest improvement in performance. We can't get to a PL of 6 but this setting will let you move a little closer to it. Going from light to strong will require a little more voltage in the cpu and NB, depending on your FSB speed. You'll see better read times and a decrease in latency as you set it stronger. Setting this to auto is ok too. You might be able to get a little more performance by setting it manually is all. Depending on your other settings, you might not. Setting it manually just means that you have to have the right voltages set to handle the increased performance, does that make sense? I can run at a setting of strong if I keep the FSB at 450 MHz or less but, I would have to bust my personal voltage limits to run strong at higher FSB speeds.Leaving anything set to auto is generally ok. It means the system will make sure you're stable. Setting everything manually means you might get a bit more performance but, you will have to tweak settings for stability, that's all.As to the OCCT temps, mine appear to be off too. Not sure where the problem lies there. My rule of thumb is to rely on a "preponderance of the evidence". Just as in instrument flying, when one instrument is telling you one thing but all the others are telling you something else, believe all the others. If OCCT says one thing, and two or three other programs say something else and are in agreement, I'd go with those.I hope that answers your questions. If I'm off on anything, I'm sure Nick, Max, or someone else will chime in and set us both straight. :( I'm really glad you have things running well now. If I ever get down your way I'll give you a shout.All the Best,Jeff