Q9650 & chance at 4GHz vs Q6600 @ 3.6GHz

[Guest simjunkie]

I've edited this post. You're right, I was out of line. I've got to be more diplomatic.-jk

[Guest simjunkie]

OK, my first response to this comment was harsh. I take it back as written. I don't mean to 'brag' my system over your's or anyone else's. And I don't want to be mean. Sam, please understand that I wish to adopt a different viewpoint than yours. I believe that with the right parts and proper cooling, and the proper settings, a 450-460FSB will run stable and robust well beyond the point that the parts become obsolete. To me that's 2 years or more.-jk

[Guest simjunkie]

And I apologize to this forum for going off on another member like that. I don't agree with you Sam but I shouldn't have spoken to you like that.-jk

[Guest Nick_N]

RhettIf after you feel you have established the correct vcore based on temps and tjM and mem voltage but you still get a hick-up..NB voltage either AUTO or 1.42.. if still no go, leave NB on AUTO and check your BIOS for CPU CLOCK SKEWNB CLOCK SKEWset them BOTH for DELAY 300pstry that and see if it calms the beast. If it does, increase it to DELAY 100ps DELAY 100pslowest value which is stable.. 800 or more may be -more- unstable with that kind of a delayyou must change them both equally or close to equal.. try not to place more than a 200ps gap between them if you try to work an offset betwen the CPU and NB.If you do not have those settings available under the jumperfree settings or advanced menu, verify you have the most recent official BIOS installed. I do not remember if those settings are available on your board or not however Asus has been adding them to all the BIOS files recently for the clocking boards from what I heard.

[Guest D17S]

Sj. no worries. The mods are allowing some heinously bad examples 'round lately. To that degree, we're on our own. Thank you for taking the right-road. We all appreciate it . . . and thoroughly enjoy the discussion. This is how we All learn. My observations is that (imHo, for goodness sakes!!) the P35 was built to run at a 400Mhz FSB from conception. All this 38/43/45/48 "upgrades" are just revenue extracting fine tunes (from their lay customer base) to that original build. A standard P35 mobo will run at 400Mhz and be operating within its design spec. The drill then becomes to match a CPU to that 400Mhz design spec. That's why I like the 9X multi . . . and Intel does NoT. Notice there were no cheap 9X quads in this latest batch? if I play My game, I'm not stuck buying one of their new (and improved!???) low multiplier CPU. and - ThereFore - , I won't need one of Intel's tweaked-out X48 super-deluxe chipsets from one of their mobo partners that is stuffed with hi-res voltage controllers . . . and whatever other fluff they can stuff it with. This Entire latest-gen has been No improvement in my book, but it is a Huge improvement in Intel's books. This "upgrade" revenue flows directly from that top-line Revenue number to an Income Statement's bottom-line Net Income. Wasn't much Expense in between, other than to to change the print on the chipset's labeling press. Intel is a business, and an InteLligent one at that. We like that.However, there's no need to play-to Intel's technical/financial strategy. Let someone else pay Intel's bills. There on the other hand, with one's eyes fully open toward achieving that illusive and final 10% of FS performance (and potential stability/system damage issues), tweaking computers is a terrifically FuN hobby.

[Guest Nick_N]

Yes that was a bit harsh ... I had a specific reason for dealing with that the way I did however its done and it needs to stay that wayHowever I do understand the frustration when one is trying to carry on a conversation about real clocking to have that be posted.We are discussing real clocking that works and this thread was not started to debate or discuss how dangerous it is or why Intel made chipsets or procs.. that posting is off topic.but hold the comments please and report it to the mods as a topic hijack if you feel its being aboused and leave the rest to them.

[Guest simjunkie]

Still, I think if I was able to get the higher end chipset and memory...and cpu to begin with, I would be where I am now and I wouldn't have spent the money of the first set (sets) of parts. And I really have seen improvements in the sim as I've made the upgrades. And this game...this monsterous scurge on computer resources and cpu cycles...makes this extra tweaking worth while. I have all the goodie addons. I've seen only better performance as I've made upgrades. For instance, put a 26" monitor on an 8800 Ultra and drive it with a 4+GHz quad cpu and a memory bus >500MHz and low latency and low tRD, and run cool...Beautiful.Things DO get better. Scenery is clearer and there are fewer "snaps" into focus. Smoother flight and fewer stutters. And FPS spends a greater percentage of time at that locked value of 30. It makes a noticable and worthwhile difference, especially with the quality addons we have available. And more robust power handling capabilities on a motherboard is never a bad thing. All that "fluff" really does add durability and reliability. I'd gladly pay extra for for it. And when going from a 975 chipset to a P35 or X48 the choice is X48 all the way, I don't care if it's not a magnitude better, it's better.Of course Intel gets our money (is a car any better an 'investment'?). But it's FSX that's so poorly written for todays HW that makes the cutting edge worth having. And we can FLY! And the benefit is you can warp through any lesser application. All these parts and tweaking have paid very satisfying rewards.-jkHere's a thought: Why don't we all just buy the best darn stuff we can possibly afford the first time and we won't have (or want) to upgrade nearly as often.

[Guest simjunkie]

Set up 455FSB & ran 2 memtest passes to make sure I could boot into the OS, no problems. Ran Prime and it errored out in less than 20min. I upped the NB to 1.55v and ran Prime again with no errors for a little over an hour. Went to bed & woke up 5 hours later to a BSOD. So I backed down to 450FSB and kept the 1.55v on the NB. Ran prime and no errors for a little over 4hr before I stopped the test (to get online-my only computer LOL). Still at 1.272v Vcore, 1.55NB, 1.26Vtt, 1.59v PLL (lowest I can set), 2.12v vDIMM (@DDR1080). NB temps staying in low to mid 40's, CPU temps getting up there at full load, 63-66 deg.-jk

[Guest Nick_N]

>Set up 455FSB & ran 2 memtest passes to make sure I could>boot into the OS, no problems. Ran Prime and it errored out>in less than 20min. I upped the NB to 1.55v and ran Prime>again with no errors for a little over an hour. Went to bed &>woke up 5 hours later to a BSOD. >>So I backed down to 450FSB and kept the 1.55v on the NB. Ran>prime and no errors for a little over 4hr before I stopped the>test (to get online-my only computer LOL). Still at 1.272v>Vcore, 1.55NB, 1.26Vtt, 1.59v PLL (lowest I can set), 2.12v>vDIMM (@DDR1080). NB temps staying in low to mid 40's, CPU>temps getting up there at full load, 63-66 deg.>>-jk>>OK.. that sounds like you are in the zone now.there is no change on the NB from 450 to 480 or 500 unless they decide to throw in a curve on that BIOS so anywhere from 450-463 is going to provide the same CPU/MEM communication path attributes and the only thing that will change is the CPU/MEM speed.. so if 450 is the charm, leave it.As long as the NB remains under 60c its fine.. 80-90c is termination temp on the x48. If you notice it creeping up take steps to find out why. Its always best to try and run the lowest voltage possible.If you find glitches as I mentioned, work that from Vcore first. The memory controller can be stabilized from vcore, especially on quads. If you are still getting a hiccup try a bit of clock skew as I mentioned to Mace.The following contains solid tech facts as an FYI...Manufacture spec life and useful life of a CPU, chips, die or boards are 2 different things.. clocking and heat will reduce the life of any electronic component (even at 400MHz... 450 makes no difference if temps are right) however that loss when compared to the useful life is nothing. IF clocking is done correctly and there are no defective components in the system you can expect a 24/7 lifespan on that with proper cleaning maintenance for airflow at right around 5-6 years minimum and 7-10 years max. If cycled and not left on all the time, that life cycle goes up. So are you using the computer you used 5-6 years ago for FSX? NotI have a few old Duron and P towers that still chug along just fine clocked to the max so this business about 'instant' damage is hogwashYou can be getting close to a 1000 bucks worth of free performance when you consider the default ability. And doing this they way you are you are now jk you are indeed getting the money out of the components Intel or whoever designed the board intended for that price you paid. They do not put a premium price on those components for nothing or a scam. That

[Guest Nick_N]

OpppsssAlmost forgot...Check with RamGuy over at the Corsair forums on the max those memory sticks are covered for.. if you see a glitch and still have a touch of room in vdd, use it.. Corsair typically needs their max rated voltage to run their clocked productsNo biggie.. because you are not really clocking those sticks @ 1066 to 1080... Corsair designed them from the bin based on the chip ability and most 1066 sticks will do 1150 to 1200. They just limit that in the circuit design. What you end up blowing from over voltage is NOT the memory chip... you typically blow the regulator circuit they design on the sticks to govern the ability and to sell them at a lower price because the bin may have tested a bit 'iffy' at 1150-1200.

[Guest simjunkie]

Thanks Nick. You're help is greatly appreciated.I think my next step will be to try to get my tRD down from 7 to 6. I do have DRAM channel A&B clock skews available but I don't see a CPU clock skew-maybe I need to look harder.So I'll tinker some more and stress test over the next day or two and let you know how it goes. My goal is 450FSB, 1080 DRAM 5-5-5-15 tRD6. I'll accept tRD7 if I have to take it though.I do get artifacts now in ATI tool when I try to clock the Ultra (eVGA KO edition), that didn't happen with the Q66 but when the Ultra was overclocked then, it would crash out of Xplane (not FSX, just Xplane). I'm not too worried about it because I want to get a GTX280 as soon as they get the drivers worked out. Hey, I got my Ultra for 150 from a friend who got an X2. I can live with it for a while longer.Cheers!-jkedit: PCIe bus locked @100.

[Guest Nick_N]

>Thanks Nick. You're help is greatly appreciated.>>I think my next step will be to try to get my tRD down from 7>to 6. I do have DRAM channel A&B clock skews available but I>don't see a CPU clock skew-maybe I need to look harder.>>So I'll tinker some more and stress test over the next day or>two and let you know how it goes. My goal is 450FSB, 1080>DRAM 5-5-5-15 tRD6. I'll accept tRD7 if I have to take it>though.>>I do get artifacts now in ATI tool when I try to clock the>Ultra (eVGA KO edition), that didn't happen with the Q66 but>when the Ultra was overclocked then, it would crash out of>Xplane (not FSX, just Xplane). I'm not too worried about it>because I want to get a GTX280 as soon as they get the drivers>worked out. Hey, I got my Ultra for 150 from a friend who got>an X2. I can live with it for a while longer.>>Cheers!>-jk>DRAM Channel AB is similar but different... see how it fairs first and if you get any glitches talk to me before changing that skewtRD6 may not be possible due to the memory speed/ability however there is something you must understand and that is there is more going on at the NB than the PL readout when you raise FSB to 450 and set the STRAP to the right value. Its on the same lines as going from 2GB to 4GB for FSX.. you loose a CAS value but gain on another end. The absolute maximum efficiency you can get on todays NB tech is 50-60% of the bandwidth you are running.. therefore you have increased the ability in dataflow so even if you were getting 50-60% at 400MHz on the last clock you ran, you have increased that dataflow within the 50-60% wall.It

[Guest Nick_N]

Getting the hardware in line is only step 1Getting Windows and the storage system in line is step 2. After everything is installed, everything... pop over to this thread and have a read...http://www.simviation.com/cgi-bin/yabb2/Ya...?num=1208959973and be aware, you can not substitute O&O Defrag v8.6 for any other defrag.. all steps must be done exactly as I postedEnjoy

[Guest simjunkie]

Oh I did all that a long time ago! I wouldn't run any other way. My last OS reinstall was a few months ago when I got this board, before that it was when I got XP 64bit. And the next install, don't know when that will be. That tuning guide should be on everyone's list!My protocol is:1) install XP, drivers, updates, FSX stuff, etc.2) setup & optimize according to your list3) defrag w/O&O, again by your list4) overclock5) monthly maintenance as you specify-jk

[Guest Nick_N]

>Oh I did all that a long time ago! I wouldn't run any other>way. My last OS reinstall was a few months ago when I got>this board, before that it was when I got XP 64bit. And the>next install, don't know when that will be. That tuning guide>should be on everyone's list!>>My protocol is:>>1) install XP, drivers, updates, FSX stuff, etc.>2) setup & optimize according to your list>3) defrag w/O&O, again by your list>4) overclock>5) monthly maintenance as you specify>>-jk>>...and rock on as the kids say:-lol Enjoy the changes. Everything is looking good from what you posted