I'll steal this from a previous post:"This is the new line up:Dual Core---E8500 - 3.16Ghz - 9.5 multiplier - 333 Base system clock - 6MB cache - $266 - Jan 08---E8400 - 3.00Ghz - 9.0 multiplier - 333 Base system clock - 6MB cache - $183 - Jan 08---E8300 - 2.83Ghz - 8.5 multiplier - 333 Base system clock - 6MB cache - $### - Jan 08---E8200 - 2.66Ghz - 8.0 multiplier - 333 Base system clock - 6MB cache - $163 - Jan 08Quad CoreQX9650 - 3.00Ghz -unlockedmulti - 333 Base system clock - 12MB cache - $999 - Nov 11---Q9550 - 2.83Ghz - 8.5 multiplier - 333 Base system clock - 12MB cache - $530 - Jan 08---Q9450 - 2.66Ghz - 8.0 multiplier - 333 Base system clock - 12MB cache - $316 - Jan 08---Q9330 - 2.50Ghz - 7.5 multiplier - 333 Base system clock --- 6MB cache - $266 - Jan 08"You're right, Intel is able to lower the multiplier by raising the FSB. The FSB is never going to go back down because it will be going away Entirely in one year. That means from now, till then, we are stuck with the current lineup. Looking beyond FSX (remember, it's probably just one more year till FS11), the big gains will be about the number of cores. These will be doubles (or quad-ruples) over what we have now. Spending big bucks to get an extra 20% (i.e., $1000 to get to 4+ ghz quad) seems like a wasted resource. If I was going to wait till Jan, the cheapest bet will be the E8200. It'll run the current FSX as well as any quad, and be the cheapest interim solution. But remember, you are going to NeeD a quad for FS11, but the Penryn is NoT that quad ( -- unless you can spend 1000 bucks for just the CPU). Looking ahead, the next-gen Nehalem will have this Quickpath interconnect replacing the FSB. I have no idea how that's going to work for we overclockers. But until that occurs, the Q6600 will continue to be the smartest longer term solution. It'll go to 3.6 (now for FSX) and be ready to provide a free performance double )that's 200%) for FS11. 20% increases are what we looked for in the old days. The CPU after Nehalem will be a shrink to 32nm and have room for up to 32 cores . . . and the software guys will have finally figured out how to use them. That's where we're going, in only several years. How do you say the difference between 4 cores and 32 cores? I don't even have a word for it. See why 20% increases are sooo unimpressive. Let Penryn go on by. Get the Q6600 and plan your next upgrade in 2 years with Nehalem's big brother. The 1st CPU of the 32nm generation will be at least an octi-core.I have a quad now and will wait 3 years for the 32nm's 2nd gen. That ought to be a 32 core version. It's not quite my 10X (1000%) increase threshold, but it's close.

>But until that occurs, the>Q6600 will continue to be the smartest longer term solution.>It'll go to 3.6 (now for FSX) Sam - My experience getting Q6600 to 3.6 is exceptional, most I've seen hit 3.0 with 3.33 on the top end. At least with keeping the volts in spec.IMO, for a value FSX high performance setup the E8500 is the one to wait for. Low cost, stock at 3.16 and a high 9.5 mult to make OC much easier, that one should hit 3.8-4.2 pretty easy.>Dual Core>---E8500 - 3.16Ghz - 9.5 multiplier - 333 Base system clock ->6MB cache - $266 - Jan 08 ...

Sam, I do not agree for that one:FS11 is IMO not 1 year away. They shift major ressources towards Train Sim. And now that new ESP program, uses more workforce. While the Train Sim thing will help to work towards new (DX10?) graphic features, it won't be quick enough to return to FS11 and make it for "holiday season 2008". We have seen what ACES could do in 1 year (SP1 + SP2). With all testing, shipping, 2008 would mean a rather small number of new features. So it will rather be "Holiday 2009". And then let's wait for Hardware to catch up, stretchin it into 2010.-> Just my bet.People like me (P4 2GHz) can't wait that long. I have the choice of E/Q6xxx at no dropping price, or Penryn's at a higher price and more absolute clock plus 10%+ other perf. At least it's the biggest progress since 1.5 years.Could you give your opinion again about the memory divider on some motherboards? Would it help to clock the E8500 higher than 3.8 (400x9.5)?THX Bigean

Boy, it seems to me you are overcomplicating the issue. I have to disagree with your blanket statement that "These new Penryn quads are going to be tough to O/C." The Q9450 will do 3.6Ghz just fine, which is a nice OC, and if you already have a Q6600 at 3.6Ghz why even bother upgrading this time around?Fortunatly for me I have yet to see any compelling reason to get a quad core. Not a single game shows any real benefit over duals, FSX and Crysis included. Crysis shows a measly 15% gain over dual, and for a game thats not even CPU bound, that means nothing. FSX sees virtualy no gain, and if I am stutter free with a dual, there is no reason for a quad. Sure people will claim that in a years time you will NEED a quad, but they have been making that claim for nearly a couple years, and I still don't need a quad.Luckily for me I can spend that same $266 on dual and hit 4Ghz without much effort, and since I can just set my memory to unlinked, having superfast ram is not needed. Besides that, 1066Mhz ram can be had for $130, which will even do a 1:1 FSB:MEM ratio, and you can't beat that. When FS11 comes out in 2-3years time I will already have a new system, so again there is no reason for me to get a quad right now. So if you only want to spend $266, then I;d say aviod the quad and get E8500.

I just setup a rig with that Q6600. Actually, I had my doubts too. 3.6 sounded pretty optimistic. However, I just turned that little crank and up it went. I'm sittin' at 3.6 . . . and thinking about going higher. It appears I have the room. Here's the tech detail on any " specs" concern:http://www.tomshardware.com/forum/221745-2...mperature-guideFor a Q6600/GO "spec" is not a voltage, it's temperature. I'm running 3.6Ghz at a Tc of 62C (Tjs 68-74) at 1.4v. Everest stress test (and FSX) will run for days here. (Note: "Tc" is the temp taken between the cores. "Tj" is the temp taken within each individual core. With the Q6600/GO, there is about a 10C cool down between the core(s) Tj, and the temp probe mounted between the cores, Tc. This is what the article states and what I am seeing IRL.) "Spec" for a Q6600/G0 is a Tc of 71C. The Tjs can run all the way to 95C and then will begin to 'throttle back' clock speed. If they hit 100C, they will simply shut down, then restart as they cool off. This is their built-in protection circuit. These CPUs cannot hurt themselves. They will start throttling, or simply shut down. However, if the operator stays within 'spec,' the chip will never get close to these fail-safe temperature points. Remember, Tjs only run about 10C hotter than Tc. This means that at the 'spec' Tc value of 71C, the Tjs won't be much above 85C. This is still Well below the 'throttling' Tj temp of 95C. This means I can go some more . . . so I think I will! This is all happening on a plain-jane P35, Asus P5K-e with all bios settings on Auto ('cept voltage and FSB). I just turned the little crank and up it went. I agree in the immediate term ANY of the the E8s will run FSX at least as well as the Q6600. Better? Probably 20%, at best. Remember, the Only difference between the E8s are their multipliers. At 9.5 (x 400) you get a dual at 3.8. At 9.0 (x 400), you get a quad at 3.6. Either costs $266. In either case, the FSB and/or the DDR2-800 ram will be the limiting factor. If you want more from the E8, it'll start to cost. You'll need an X38/48 mobo AND DDR2-1066 ram. All that costs more right now to trade-off a (free) double down the road for a (2 month out) 20% increase right now. That 20% right will cost me a whole new rig when it comes time to quad-up . . . and that'll probably have to be a 1st-gen Quickpath/Nehalem. Ugh! That's gonna be a mess. Let 'em sort that one out for a year. Looking ahead, there are lots of things to consider. FS11 is a total guessing game. It could be a year . . . or 3! One thing IS for sure, it will be Fully multi-core capable. The memory dividers are entirely miss-named (anymore). We used to be able to turn the memory down so we could run a higher FSB with el-cheapo memory. Not any more. The slowest any memory will run with a FSB is 1 to 1 (1:1). That means the memory will DDR (Double the Data Rate) of whatever FSB you set. Want to run a FSB of 400. Better have memory that will run at 800 mhz (DDR2-800). Want to crank that E8 on up some more? The memory must be able to run on up too. Either you gamble that your DDR2-800 ram will not explode when you run it faster than its rated speed (800mhz), or you go buy more expensive ram (that is rated to go faster, DDR2-1066, for instance). Speed costs. How fast do you want to go? That E8 is gonna cost more to make it go beyond a 400mhz FSB (3.8ghz). You'll need an X38/48 mobo plus faster (rated) ram. All this for a 20% increase (now) over free 200% increase (later)? The Q6600 just made overall sense to me.

Ok, Sam and Solar,Sam already *has* a Q6600 OC'ed. It really doesnt make sense for him to do anything before Nehalem comes.Me and others who are about to get a new rig, need to check. The Penryn is the first chance for more clock speed since 2005/6. Not the 3GHz model, it only goes 10-20% faster over the "old" C2D 3GHz. Now how much more will be possible? We already know now that FSX again scales the most on clock raise. You get the biggest bang. Now for the bucks: If mobo can tolerate a >500MHz FSB, but you can set the memory back to 400, doesn't it solve the problem? And having >4GHz compared to 3.6GHz is at least another 10%, exceeding 4.2 something goes into a close 20%. 10-20% basic gain and another 10-20% clock-related gain, not bad (if it's really like that). When did we have such a gain last time? But what does it take, and how much does it cost? I don't know, that exceeds my knowledge. I'm naive there, so insight is appreciated ;)Bigean

Having a 500fsb daily overclock is way beyond the capability of most people imo. It's not easy at all. It requires an above average motherboard and lot's of cooling imo.Even passing 450 on a daily overclock is more than alot of gear will do. Up to 420 should be fairly easy. There are alot of C2D's and quads that will hit 3.5ghz, but not all of them will do it. After passing around 3.2 or 2.4 ghz the heat really starts ramping up, especially with quads.

That's just what I'm getting. The Qs running fine at 400 x 9. First, to make sure my DDR2-800 ram wasn't going to explode at 900mhz, I dropped the CPUs multi to 6 and booted/Primed at a 450 FSB. That DDR2-800 Patriot ram ran fine at 900mhz (2.3v). Nice stuff. I then brought the CPU's multi back to 8 at the same 450FSB. This was still running a 3.6ghz CPU that worked fine at 400 x 9. At 450 x 8, it Primed OK but other stuff got weird. The optical drive wouldn't read, audio got choppy and 3DMark wouldn't run. FS loaded but still that choppy audio persisted. Starting again from 400 mhz. Currently at 410 x 9 with all mobo tweaker settings still in "Auto." The Q's hanging in there at 3.7Ghz (1.4v). Temps are fine under the Thermalright Ultra.If ya'll want the drive that E8XX, your gonna need a big(er) dog mobo. As was described, the P35 is entirely happy around 400mhz, but not much more.3.8? Stay tuned.

Sam,>http://www.tomshardware.com/forum/221745-2...mperature-guide>>For a Q6600/GO "spec" is not a voltage, it's temperature. I'm>running 3.6Ghz at a Tc of 62C (Tjs 68-74) at 1.4v. Everest>stress test (and FSX) will run for days here. >>(Note: "Tc" is the temp taken between the cores. "Tj" is the>temp taken within each individual core. With the Q6600/GO,>there is about a 10C cool down between the core(s) Tj, and the>temp probe mounted between the cores, Tc. This is what the>article states and what I am seeing IRL.) >>"Spec" for a Q6600/G0 is a Tc of 71C. The Tjs can run all the>way to 95C and then will begin to 'throttle back' clock speed.>If they hit 100C, they will simply shut down, then restart as>they cool off. This is their built-in protection circuit.>These CPUs cannot hurt themselves. They will start throttling,>or simply shut down. >>However, if the operator stays within 'spec,' the chip will>never get close to these fail-safe temperature points.>Remember, Tjs only run about 10C hotter than Tc. This means>that at the 'spec' Tc value of 71C, the Tjs won't be much>above 85C. This is still Well below the 'throttling' Tj temp>of 95C. This means I can go some more . . . so I think I will!>Thanks for providing this link and summarising the Tc and Tj situation. This is the first time the whole T thing has made sense to me. I now see that at my current 3.6GHz overclock with Tc of 60C and Tjs of 71-75C that I still have a little headroom to push harder too. :-ukliam :-)A couple of questions before I do though:1. Any idea why my mobo won't boot at 3.6GHz at vcore up to 1.5V, yet if I boot at 3.5GHz and vcore 1.4V then use the Gigabyte EasyTune5 utility to crank it to 3.6GHz @ 1.44V it will run all day fully loaded at this higher speed?2. The article you linked to says don't go over 1.5V vcore. Is that the BIOS set value or what my mobo/cpu is actually reporting it is reading? The reason I ask is that with vcore set to 1.44V, I get about 1.38V at the CPU, which suggests that I could perhaps go over 1.5V vcore in the BIOS to get a resultant CPU vcore of 1.5V max. What do you think?Gary

Imo 70c is too high a temp for daily useage and long component life. Certainly passing 70c is not good at all.60c is relatively high, but shouldn't kill anything.70c is 158f, that is pretty hot imo. 80c=176f90c=194f95c=203fIt's true they have thermal protection, but high temps will still shorten component life.

Thanks for the concern, but as long as my computer doesn't catch fire, is stable and the CPU lasts the one year before my next upgrade, I say let it simmer for this simmer :-hah In all seriousness, the only time it will hit those temps is when all four cores are being hammered continuously, like when running a stress test program such as Prime95. I'm doing an FS9 flight with a bunch of addons right now that has my Tc at 53C and CPU use 25-33%. FSX + addons is a bit heftier and runs me at Tc of around 56C and CPU use of 33-50% use. I could probably squeeze another 100-200MHz out of my CPU before it hits 60C Tc running my FSX suite. Well, that's the theory anyway ;-)Gary

>2. The article you linked to says don't go over 1.5V vcore. >Is that the BIOS set value or what my mobo/cpu is actually>reporting it is reading? The reason I ask is that with vcore>set to 1.44V, I get about 1.38V at the CPU, which suggests>that I could perhaps go over 1.5V vcore in the BIOS to get a>resultant CPU vcore of 1.5V max. What do you think?>>Gary>On my board I have to set the BIOS to 1.525v to in order to get the BIOS to report a voltage of 1.49v. I have two columns, the column where I set the voltage, and a column next to it that says "Current Value". When setting the voltage to 1.525, the "Current Value" column reports the 1.49v, which is also the same value I see when viewing the system monitor screen in the BIOS, as well as what Everest reports. As far as the new chips go, I'm quite curious to see how FSX responds to the larger cache.

Curiosity got the better of me, so I decided to push it as far as it would go up to 1.5V Vcore indicated. I made it up to 3.75GHz @ 1.48V Vcore, but anything after that is no go. Interestingly, Tc didn't go above 62C during stess testing, but I guess that's all this little quaddy has in her (which is outstanding IMO). I think I'll leave it at 3.6GHz, which was the best I was hoping to get out of this chip anyway.Gary

>That's because these comparisons are meaningless. You can>easy run the 9650 at 4-5 Mhz. You can't do that with the>6850.I think this will be the key factor with the Pen's.It is especially a factor if you have a late-gen netburst P4 (aka space heater) or an Athlon 64 platform...Less so if you already have a Core2 of some sort.RhettAMD 3700+ (@2585 mhz), eVGA 7800GT 256 (Guru3D 93.71), ASUS A8N-E, PC Power 510 SLI, 2gb Corsair XMS 3-3-3-8 (1T), WD 150 gig 10000rpm Raptor, WD 250gig 7200rpm SATA2, Seagate 120gb 5400 rpm external HD, CoolerMaster Praetorian

We're all aware that ram has timings. Tcas, Tras and the rest. Often we will see ram advertised with a CAS2 or CAS3 rating. My current Patriot LL ram has timings (available) at 4-5-5-12 at DDR-667. These timings refer to the speed at which the chip can exicute internal processes. As I overclock the ram to DDR800, these timings will auto-relax to 5-6-6-15. If I leave the ram timings setting in "Auto," this all happens, well . . . automatically. (But don't worry, I'll be going to manual soon . . . and clamping these timings down till something catches fire. And, it will make absolutely No difference in FS. This is just for run.) Guess what? The Northbridge (NB) has timings too. Rather than Cas and Ras (and the like for ram), the NB calls its timing sets "Straps." We have the 400 strap and the 667 strap and the 800 strap and whatever-they-want-to-come-up-with strap. For instance, (in auto) when NB sees around a 400mhz FSB, it will set the 400 strap (timing set). This strap will have the quickest (tightest) timings. As we turn up the FSB, the NB will relax its internal timings (straps) on some sort of schedule. These schedules cause all kinds of problems. The bios makers just can't seem to come to grips with when to relax these timings as we O/Cers ramp up our FSB. For instance, they will leave the timings too tight at a certain FSB. This will let this CPU at that voltage run OK, but that CPU at this voltage won't even boot. Users will see "holes" (FSB speeds) where their mobo is totally inop. It can be one major PITA. Gary. I expect this is what you are seeing. That Gigibyte easytuner is allowing the NB to keep its current, working (working) timing set. This allows the 'puter to keep on trucking. A reboot/bios reset will cause the NB to reset its timings and lock you out. . . . maybe because the quad and that particular NB timing set will not play nice. Folks have tried to find exactly where these straps re-set, but it gets beyond me real fast from there. This is one of the reasons I'm sooooo pleased with my P5K. It just runs where ever I put if, so far that is.Try going into the bios and manually setting the 800mhz (or 1066) NB strap. This will force the NB to use the most relaxed timings and provide the greatest chance of success. This will cause No performance decrease. Consider: Our FSB is like having a racetrack that can accommodate 800mph cars. Our cars can only get going 100 mph, at best. Will it cause our "cars" (data) a performance hit by reducing the tracks accommodation to 780 mph? Not in the Least. (That's why AMD's hypertransport is a faster buss scheme, but makes no difference. Intel stayed with the slower FSB. Nothing ever existed that could fully utalize either one. That is still true today.) That 1.5 volt limit is just for beginners . . but still, also for all of us that would like to have our 'puters still in one piece after playtime! It's not the voltage that causes the damage. It's the temperature. If a user is seriously playing attention, volts above 1.5 are usable. Think about pushing the thrust lever of a CF6-80e2 up for take-off. You don't just jamb it to the wall. Yea, it has overspeed and overtemp protection, but you better believe . . . someone's eyeballs are glued to EGT. Voltages above 1.5 are entirely usable, as long as you pay Close attention to temp. That said, My initial bios was fine, but I had to flash (right away) to the latest and greatest. Currently, I have to set 1.5v to actually get 1.4v. It's and entire .1v off. That's a lot. But as long as I know this is happening, it's just an snnoyance. Consider, it could have gone the other way. I set 1.4 and get 1.5. If I'm not watching EGT, I'm trying out my overtemp protection system. That's pushing anyones luck with a 10 million dollar engine . . . or my personal equivalent, a $300 quad. Once again, what you set is NOT necessarily what you get. We just have to deal with an imperfect world sometimes.Is 70C too high? Hey! Are we talking Tc or Tj? 70C Tj is no problem at all. Again, consider these temperature limits work just like turbine engine temperature limits. Intel designs these CPUs to operate at a "Continuous Power" factor. That factor (like an airplane engine) is a temperature. Intel's "Max Continuous" temp is Tj. They set a Tc just for the masses. Tc can be ignored as long as Tj is not sitting on its throttling stop. The CPU will automatically protect itself from damage, but we Do Not want to lean on this "stop." Anything below the Tj throttling stop Will Not shorten the life of the CPU. Server farms would be warranty-ing these things by the truckload if running at above Tc caused problems. The Xeon is the same CPU, but has an advertised Tc of 80+. That's because it is targeted to a more savvy user group. The Qs and Es are for us.Since my bios "upgrade," my Tc reporting tools are all over the place. Speedfan says this, Everest says that, Asus Winmonitor says something else. I give up. Coretemp's Tj readings seem to be solid. That's all I care about anyway. With the Q6600/GOs, Tc will run about 10C below an averaged Tj (if I really cared). If you really want to watch Tc, look for a Tc (somewhere) that is about 10C below an averaged Tj. For instance, my Tjs are 65-75 and Tc is (variably) 70-80C. This makes no sense. I'm currently ignoring Tc. Sure how the next P5K bios fixes this stuff. What else's goin' on?