Here's the scenario and I am clueless as to what is happening.When I first built this machine, after a 2-day burn in, I decided to go for an overclock to 4.00GHz. I bumped up vCore from default 1.25 to a paultry 1.287, and voila, no problem running this beast at 4.0GHz on stock cooling with the case cover off. This was done with the simple muliplier change to 12.0. I ran it this way for about a week, until I realized I just don't need this horsepower as I am not using FSX yet. So, put her back to 3.5Ghz at stock vCore, and have run it there since. Last night I decided to go back to 4.0Ghz for the fun of it, and maybe to see where the limits are on this rig. Well, it would not boot to XP. So, up with vCore all the way to my personal max limit, 1.36 volts. That's still less than a 10% bump. No joy. I jacked up the cooling which is really effective, still no joy. I could not boot to the desktop at > than 3.66 GHz. Hmmm, really odd . . .So, here's where it gets weird. I decided to boot to my cloned copy. Booted straight away to the desktop at 4.0GHz, back at 1.2875 vCore. Ran fine! OK, you tell me, what the heck could possibly be happening? QX9650 w/ Retail HSF|8GB Muskin PC-6400|ASUS P5E|EVGA 8800GT @700|Seagate SATA 2 x 4|Seagate Cheetah 15K.x|XP Pro|Vista 64--soon to be installed

OK now? CHKDSK runs are magic when things get weird. Also, I have constant problems on my P5K-e with the SoundMax sound driver. I can immediately tell that it got "whacked" when the FS intro music is choppy and model animation is hoppy. The computer virtually stops. I have to uninstall the SoundMax driver and reboot. Vista recovers the driver and then all's OK, for a while. I've called Asus, but the conversation always ends with, "So there fancy pants, what FSB ya running?" "Ahhh, never mind (but that's not it, rootie toot)," I mumble. Anyone else?

I used System Restore, went back about 5 days or so, and voila, am booking along at 4GHz at 1.2875v. And guess what I was poking around with the Soundmax stuff: control panel and audio wizard. As a former Win2K user before this machine, I can say it's nice to use the system restore feature in XP. I'm very glad my CPU didn't turn out to be a dud! One of these days I need to push it up and bit and see how she'll run. I mean think about this: 33% overclock with only a small bump in vCore so really on essentially room air, it really should o'clock more. I think Anand was concluding 4.25Ghz was about the point of diminishing returns, or something of that sort? What was your take on that Sam?QX9650 w/ Retail HSF|8GB Muskin PC-6400|ASUS P5E|EVGA 8800GT @700|Seagate SATA 2 x 4|Seagate Cheetah 15K.x|XP Pro|Vista 64--soon to be installed

Time for another science project. Anand suggests the new CPUs have (will have) preemptive thermal protection. This protection will throttle individual cores to maintain temperature control at a clock limit . . . not necessarily at a thermal trigger point. The thermal protection would therefore be preemptive. Intel has probably had enough of these gang-stacked server farms burning up CPUs.So, get Prime 95 going and run up the clock. The throttling will occur by one (or more) core(s) skipping Prime cycles. Gotta watch carefully, but Anand saw this happening with a Q9something the site was testing a while back. That would be the diminishing return to which Anand was refering. Why O/C if the actual work done was not increasing? Hummm.(Maybe we could start giving out science badges.)

Interesting, is this throttling noticeable using cpuz to measure the cpu speed? Or is it "invisible".

>OK now? CHKDSK runs are magic when things get weird. >>Also, I have constant problems on my P5K-e with the SoundMax>sound driver. I can immediately tell that it got "whacked">when the FS intro music is choppy and model animation is>hoppy. The computer virtually stops. I have to uninstall the>SoundMax driver and reboot. Vista recovers the driver and then>all's OK, for a while. I've called Asus, but the conversation>always ends with, "So there fancy pants, what FSB ya running?">"Ahhh, never mind (but that's not it, rootie toot)," I mumble.>>>Anyone else? After doing the XP system restore to a week ago or so, I went for it today. 4.4Ghz boots to Windows fine, however this took ~1.375 vCore. It errored out at this volt in Prime95. Temp was not a problem (relatively speaking). Never went past 54C at this clock in TortureTest with my cooler set on max. I was reading Anand's article again and seems like it's really kinda not worth much as the power draw is so steep at this o'clock I may be starting to hit the limits of my PS, which was 750W with a huge 12v rail. From Anand's article, probably would have no troubles bumping up vC even as high as 1.38 or even higher, as he was saying it's hard to hurt a Penryn even with vC due to some built in protective schemes. Even so, it's kinda moot but as you have said good for braggin' rights. 1.38 is exactly a 10% overvolt, which IMO is never an issue. I noticed with Anand's water cooled rig showing CPUid 1.55v showed some very hefty increase of vCore, or at least actual vC. Not sure what he set it at. 1.6 probably. Anyway, kinda scary to throw that at a $1,000 processor. I'm guessing if I ran this guy at 4.33Ghz or 1.25Ghz x 1.39 or 1.40 vC that would fly. Sam, why does Anand run his FSB at 425? Is there NO reason to increase the FSB bandwidth, ever? That doesn't make sense either. It's in part about how fast data can move from CPU to memory or NB, no? If that's the case then perhaps it IS an issue, running the FSB as fast as possible that is. Not sure WHEN it would become an issue, but not sure . . .QX9650 w/ Retail HSF|8GB Muskin PC-6400|ASUS P5E|EVGA 8800GT @700|Seagate SATA 2 x 4|Seagate Cheetah 15K.x|XP Pro|Vista 64--soon to be installed

This (alleged) throttling only happens on One core. CPUZ wouldn't see it. Anand had Prime running and noticed something "funny" going on. One of threads was falling behind. He theorized this was one of the cores throttling.My impressiion is at Anandtech, a 425Mhz FSB is no big deal. And so it's not with the 38-48x mobos. The 35s are good to 400, but that's as far as I would trust them. There are no benefit of a faster FSB, ever, but Anand also did an article that Slower FSBs could get better performance. FSB megahertz is not a limiting factor. However, timing factors can be. By slowing down a FSB, the northbridge will auto-shift to faster timings. This is called the NB "strap" and can be manually set in the bios, regardless of FSB speed. These tighter NB/FSB timings increase performance by a scientifically measurable, 'ittle-bit. Unless you just like to play, it's not worth the trouble.

FSB megahertz>is not a limiting factor. This is where the rubber meets the road. OK, that is nice to know.Now, what do you say about vCore max for my processor? Am I being overly careful keeping below 1.38 or what do you think?QX9650 w/ Retail HSF|8GB Muskin PC-6400|ASUS P5E|EVGA 8800GT @700|Seagate SATA 2 x 4|Seagate Cheetah 15K.x|XP Pro|Vista 64--soon to be installed

The 6600's limit is 100C (Tj) and 1.5Vcore. It also has a 71C (Tc) limit, but that's pretty meaningless. Tc is just an average of the Tjs - 15C. The 45nms are a different animal. Really haven't done the research on those.

Im keeping my 45nm at 1.45v, which is the absolute max stated in the datasheets, maybe im a bit crazy but time will tell if thats too much. Im hopign in the coming months we find out these can take more voltage, then we can really overclock these cpus, right now theres pretty much no need to watercool with a 45nm because the temps are so low at these low voltages.