Sam, Gary et al:I installed my Tuniq Tower 120 on my Gigabyte P35-DS3R with an e6850, cranked the fan to high (still very quiet in my Coolermaster Stacker) and in the Gigabyte BIOS, changed the voltage to AUTO (as per the BIOS recommendations) and increased the mulitiplier from 333 to 400. Steady 3.6ghz, no issues and core temps measured with "Core Temp" never climbed above 55c during a quick pattern around EGLL.Ok, now what- leave it as is, or with these low temps, try for a bit more? The RAM is CORSAIR XMS2 2GB (2 x 1GB) 240-Pin DDR2 SDRAM DDR2 800 (PC2 6400) Dual Channel Kit Desktop Memory.Chime in guys, I'd like to get all I can out of this system in a 24/7 STABLE configuration- I do not want to sacrifice stability for any more performance at this stage.......Best-Carl F. Avari-Cooper BAW0225http://online.vatsimindicators.net/980091/523.png

The only real way know is to again bump the multiplier (thus increasing the speed) and stress testing your computer for a 24 hour period and make sure you don't have any errors.If you make another increase in speed pay attetntion to your temperatures. Easiest way to test your temps prior to stressing is to download Intel's Thermal Analysis Tool and set it for 100% loading of each core and then load both cores and see how your temps are.

Carl,IIRC, the problem with the auto setting for voltages on the gigabyte motherboards is that it scales most, if not all, the voltages up as you overclock. When I tried this (a long time ago), I found the RAM, PCI and CPU voltages as reported by EasyTune5 were all sky high for what I wanted to see.Try using manual voltages and keep everything at stock as you work your way up from 333 to 400 in say 30MHz steps and test for stability. If you make it to 400MHz without having to increase core voltage then there's usually a bit more overclocking in it yet. Put another way, as soon as you start having to increase core voltage above stock, you are usually within the last 400MHz resultant CPU speed of stable operation. eg. with my Q6600, I had to increase core voltage slightly from 1.235V Vid to get 3.2GHz stable and ended up at 1.4V for 100% stable 3.5GHz operation. 3.6GHz @ 1.4625V works about 95% stable for me, but I'd rather a little less OC and 100% stability any day.Gary

I don't know how the auto-strategy works in da Gigabite, but I have my P5K-e manually set to a Vcore of 1.425v and a FSB of 400mhz. Everything else stayed in auto. The Asus auto-version just worked (thank goodness). Nothing's caught fire yet. The Q6600/GO CPU became an immediate happy-clam at 3.6Ghz. Tjs (average) 75-80C under a Prime 95 load. At 3.6ghz, a Tj @ 55C means that you're either underwater (or Wesson oil) or providing a very light load. You're doing air and flying the sim. Look over this and get familiar with the difference between Tc and Tj. http://www.tomshardware.com/forum/221745-2...mperature-guideNow find the Tc/Tj/Vcore specs for that CPU. Use Coretemp to monitor Tj, Speedfan for Tc. But be aware, it has gotta make sense. Coretemp (Tj) will be your most reliable reading. Tc must be about 10C Cooler than the Tjs (averaged). If it's not, ignore Tc. You are Only interested in Tj anyway.Use CPUZ to monitor your CPU volts. Be aware that even a manual bios setting is no guarantee of what's gonna come out the other end (My bios sets Vcore .1v low. That's a lot). Watch CPUZ as you do your power runs. OK. Let's kick it. Ever do a power run in a real 747. Maybe a 737 (twin) is more appropriate here, but I've never run one of those. But in any case, here's how it goes. It be da same deal. CPUZ will be your fuel flow indicator (fuel pressure, actually). Coretemp will be your EGT indicator.Muscle that beast over to mid-taxiway Bravo and get the brakes set. Call the tower and tell 'em to keep your tail clear. Settle in. Now push 'em on up (Get Prime95 and fire it up). Watch that your fuel flow stays steady and doesn't drop off as the power comes up (no Vdroop). Watch EGT on your 'ittle twinjet. If Both engines (cores) stay online, great. Go some more (increase your FSB). If not, bump up fuel flow (Vcore). Check your specs. For the Q6600, FF is 1.5v, tops. These CPUs have temp limiting just like a real jet engine. However a wise runner doesn't just kick the thrust levers to the wall and hope for the best. The Q6600 cores will begin to throttle back (de-clock) at 95C and auto-shutdown at 100C. But don't let it get there. Keep your hand on the thrust levers at all times. Watch that EGT. But especially, have fun.

Ok Guys- first- thanks!I'm not quite sure whether I have understood all of this correctly. Sam, the run-up brought back a lot of memories- thanks!ok- here's a screen shot to gove you a progress report. Everything seems to be running smoothly and I am at 3.6Ghz- have been for about the last 18 hours- without any apparent issues (fingers crossed :))Best-Carl F. Avari-Cooper BAW0225http://online.vatsimindicators.net/980091/523.png

Go into coretemp's options > setting screen and uncheck the "Show delta to Tj temp." What you are showing there is the difference between the actual Tjs and Tj(max), 100C. You want to see the actual Tjs.For that CPU, do you know Tj(max) IS 100C? Coretemp might just be guessing.That Prime screen doesn't look right for 2 cores. You may need 2 instances running. Looks to me like you only have one core loaded. I have a Prime 95 that is 'special built' for the quad and have 4 separate instances clearly annunciating, but my old HT/P4 needed 2 instances of Prime 95 running separately. There was some command line tinkering to get em both to go. I think you need something like this.

Sam:That is a bug on Core Temp, so the delta shows the balance to 100c- subtracting gives the individual temps. I can try running Prime95 twice concurrently and see if that makes a difference. The voltage seems ok in auto and so far so good with stability at 3.6ghz.......maybe I ought to quit while I'm ahead? Thanks!Carl Avari-CooperBAW0225

The bug in coretemp is with B3s. Their Tj(max) is 85C. Therefore, with coretemp's Tj(max) at 100, the B3s give an inaccurate delta. With the GOs, the delta function works as advertised. But forget about delta. Work with actuals. If you are considering these deltas as your actual temps, you are actually working with temps that are way low. Call the tower. Have the fire trucks standing by.

Sam and others:I have no idea how to get the actual temps other than with CoreTemp and with that showing Tj 100c always- the only two other temps lower down show idle at about 36-37c. With Intel's TAT it idles consistently at about 23-24 c and the hottest it has ever gotten with BOTH CPUs at 100% workload is 56C...........I don't seem to be having any issues at 3.6ghz and am inclined to just leave well enough alone......thoughts?Best-Carl F. Avari-Cooper BAW0225http://online.vatsimindicators.net/980091/523.png

Coretemp's Tj/100C is a fixed value showing a CPU's maximum allowable temperature. It may not be Your CPU's maximum temperature, though. Unless you know your actual Tj(max), disregard that 100C number. Intel's 24C makes no sense for an Idle reading. That's room temperature (unless you're doing this on the artic plain). Is that CoreTemp 36C still the delta to that undefined 100C? If so, that too is not helpful. Uncheck Coretemp's delta box and directly read the 2 Tjs. Get Prime95 and run that as your stress tester. Otherwise "load" temp is not helpful for troubleshooting purposes. The whole point of going with a same priced Dual over a Quad was to get another 150Mhz of dual core speed. The Quads go to 3.6. The duals go to 3.75. The argument is that FS only uses 2 cores anyway, so the additional 1.75mhz is worth giving up those extra 2 cores. I never bought it, but that's the argument. Now that you have it, you might as well get it. Here a link that describes all that temperature stuff: http://www.tomshardware.com/forum/221745-2...mperature-guide

Ok-Running MANUAL voltage control set at 1.425, FSB set at 411 and indicating 3.7Ghz stable with Prime95 for 6 hours so far. Will continue- but all appears as though this is going to work!Thanks Sam and Gary and others- I really appreciate all the advice. FS9 is awesome and FSX runs well also!I think this is where I will stay until the reporting issue with Core Temp is worked out. With that HUGE Tuinq 120- I might go for 4Ghz once I'm sure what the Tj Max really is!HAPPY CHRISTMAS AND NEW YEAR TO ALL!Best-Carl F. Avari-Cooper BAW0225http://online.vatsimindicators.net/980091/523.png

Stopped Prime95 12 hours later- all ok, no issues- am going to leave well enough alone until I can figure out what the Tjs really are. Then I might try for faster (4Ghz). Is that possible or am I maxed out now? Best-Carl F. Avari-Cooper BAW0225http://online.vatsimindicators.net/980091/523.png

You know of course that 333/400 the FSB speeds, not the multipliers. Your multiplier at 3.6 GHz with 400 FSB is 9. If your system is stable at 9, try going to 10 - that would give you 4 Ghz. I am doing exactly what you are, but with an Asus/QX9650 system. In addition to Stability with Prime95 and various other tests, I also have set up 5 FSX test flights (including full AI, my preferred add-on aircraft, etc.) to measure fps and performance. I created four o/c profiles which are all stable (48 hour Prime95). The results in my FSX setups (added all five fps measures / 5)FSB x M = GHz : Frames / second333 x 11 3.6 : 48.42333 x 12 4.0 : 48.98 400 x 9 3.6 : 48.61400 x 10 4.0 : 49.04So for an actually in-game fps gain of 0.68 (!!!), you would run your PC at significantly higher temps and double the power consumption (@4Hz = 130 W, 3,6GHz = 80W with my Penryn). No thanks. I am simming at 333 x9. It all of course depends on what YOU want to do with your PC.The only real gain I saw was at 4,4 GHz - but I don't want to run my CPU that hot before the final BIOS versions are out. Also, check whether 400 FSB is worth it. Every old-school overclocker will tell you to go for the higher FSB, but with new multi-core chips that is not true anymore. As you can see from my test results, running your motherboard warm doesn't necessarily offer much in terms of game performance. NO WARRANTIES IMPLIED. RESULTS APPLY TO MY SETUP ONLY. Oh and one last thing: NEVER try to optimize a system until at least one year after the release of the MB. There are always BIOS updates which negate any performance measurements you may have relied on. There is an excellent article on anandtech about the QX9650 which included basic information on how to proceed when overclocking, getting the memory system stable before starting on the CPU: worth a read http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=3184

Sam, Gary, Marten et al:I downloaded the latest version of Core Temp which appears to correctly read the Tjs for my chip (e6850)- and in 12 hours of Prime 95 with no errors- the highest recorded temps for the 2 cores was 54 degrees Centigrade........thoughts..... This was at 400FSB (3.6ghz) and manual 1.425v.Thanks, and best wishes for a HAPPY AND PROSPEROUS 2008!Best-Carl F. Avari-Cooper BAW0225http://online.vatsimindicators.net/980091/523.png

My idle Tjs (45/47/54/54) are about what you are reporting under load. I've got the Thermalright Ultra 120 with the Skyth fan at 1600RPM. I'm a bit suspicious of those low temperatures, but it really doesn't matter. You're monitoring EGT on engines that have built-in thermal protection. A pilot has to at least make-believe he has an actual purpose. Watch those temps, but if you're still "making power" go some more. The trick is to pretend we're actually being useful (by watching temperatures), but keep going 'till it starts throwin' cores in Prime. Then, if you're really adventurous (like me) go some more. Fire up Everest's System Stability test and go 'til even Everest starts taking 'em out. Then backoff just a notch. Throw your roughest approach (with the PMDG 744) at it. I like the VHHX, but a megescenery'd LA is the stress-tester today. See what happens. I've just been playing and can get 3.75Ghz with load Tjs in the low 80s. My #3 man keeps flaking out with a Prime 95t load, but everyone can get through both 3dmark06 and the Everest stress test. I'm currently descending through 15K inbound on the SADDE6 to the LAX ILS24R. I have my new 8800GT O/C'd to 720mhz G/core and I'm getting a pretty steady 20FPS on the 42" (1920x1080) with AA@8x/AF@16x. All 4 cores are working hard (in Task manager) and I rarely see CPU utilization fall below 50%. Ram load is 3.15G. It appears this 64 bit operating system is giving me a bit of extra ram headroom. Wish I'd bought 8 gigs. I believe my HW is about ToTally maxed out. We'll see what happens once I get a bit closer to the ground. A little ol' blue screen is a terrible thing to waste.