For Those Whose "CoreTemps" may be a bit high with the Q6600

[BonzaiDuck]

FLASH!! BIOS UPDATES MAY ADD uCODE ENHANCEMENTS THAT EITHER REDUCE TEMPERATURES, RAISE POTENTIAL OVER-CLOCK SETTINGS, OR BOTH

by BonzaiDuck, your AQ news service ["Aff - Quack!!"]

This is a heads-up -- and its accuracy is tentative, but promising nevertheless. I cannot see why it would not have general applicability to motherboards of various manufacture -- compatible with both C2D and C2Q processors.

In recent days or weeks, those of us cavemen who choose to push forward on air-cooling have noticed that the C2Q is a real pop-tart-toaster. Our water-cooling guru agrees, noting the thermal output and how certain radiators are not adequate to keep the load temperatures in the low-50s or below.

I've proven that you can reduce these temperatures by 5 to 7C degrees with motherboard and CPU-cooler ducting. We've also had quite a spate of discussions about the TR Ultra 120 Extreme, synthetic diamond thermal paste and other things which promise marginal improvements in core temperature values.

Basically, I do not believe it can be refuted -- given the relationship between temperature and electrical resistance, and the relationship between voltage and temperature, that it is possible to keep the cores, chipset and other components cool enough so that stable settings can be maintained for the same clock speed at lower voltage. This, in turn, means even more moderation in temperatures.

I decided to risk Cadillac money with uncertain hopes that I'd get a Cadillac motherboard. So I broke with my tradition of choosing modestly priced boards of good manufacture, made my own interpretations of early customer reviews that had an ambivalent mix of ratings, and shelled out my clams for a Striker Extreme. I may already have said-- given some benchmarks that show that the STriker falls a tad short of some other boards on extreme sustainable external frequencies (sometimes called "FSB" which in turn is often expressed as a quad-pumped value of 4x the external frequency) -- that the STriker is a Cadillac, but it is not a Ferrari or a Corvette.

Those who purchased the SE late last year found themselves with BIOS revisions 501 through 901(?), and the 1004 BIOS upgrade did not appear until March '07. [Again, keep in mind my argument that this has general applicability -- beyond ASUS boards or particular models.] I got the 1004 BIOS, which had cleared up some anomalies in "set" voltages versus "monitored" voltages versus "actual" voltages, and the inability of the board to implement "set" voltages beyond a certain level. The board also had provision for some eight fans and corresponding temperature sensors, but users on the ASUS forums noted that they couldn't get the fan-control features to work well.

Along came three more BIOS revisions -- 1102, 1301 and 1303. The last two appeared on the ASUS download pages during this last month of July. All of these revisions listed as fixes the implementation of microcode to improve the performance of "certain" processors, and other enhancements that "assist in the over-clocking of certain processors." [I love ASUS' penchant for vagueness, like I enjoy certain kinds of weeds in my kitchen garden.]

Just for the improvement in fan-control, I had a near-disaster with a BIOS upgrade when I thoughtlessly forgot to change the settings back to the stock or "Auto" values -- or for that matter, clear the CMOS before making the flash. Knowledge will cost you money, time is money and impatience or urgency in continuing with a project may pressure you to trade off some wallet-filler against those other variables. But I'm extremely pleased that I can now e-mail ASUS and ask them to cancel my RMA request of two weeks ago.

Now -- I've finally flashed BIOS revision 1303 -- successfully.

Guess what? And I had a suspicion that such might be the case -- the thermal wattage for the C2Q / Q6600 currently in use -- a B3 stepping -- has been tamed -- or so it seems -- slightly. Peak stock temperatures at 73F I have shown in other posts to be around 52C for the hottest core -- and my temps will be more than 5C lower than those of other users for the motherboard ducting. But now, at a 3 Ghz clock rate and FSB of 1,480 Mhz /DDR2 740 Mhz, the temperatures are about 2C lower than at previous over-clocks and the same room ambients at 2xORTHOS load. Further, I had been unable to get beyond an FSB bus speed of 1,440 -- even at the lower multiplier of 8. It now appears that I can do this at a voltage about 0.03V below the Intel "Maximum" spec, and I may be able to drop it a notch or two, or otherwise push the clock speed a tad higher without raising the voltage.

Since the Kentsfield processor appeared late last year, I suggest to everyone who has acquired the B3 stepping -- make a visit to the motherboard web-site and see if BIOS revisions offer some hope of improving performance on that and other Core 2 processors.

A word of caution -- Clear your CMOS and "Flash carefully." And you might want to file this link away for insurance, just in case:

BIOSMan

 

[BonzaiDuck]

PS

"So easy . . . . a caveman can do it . . . . "

 

[drakore]

Thanks for the update...

Wanna write an article for my watercooled xbox 360 project?

 

[BonzaiDuck]

Well -- I'm not "Experienced" in either . . . . how would you want to do it?

You provide the data -- and I do with word-craft?

 

[drakore]

hmmm maybe.... it was more of a joke than anything...

I am trying to get one done on xbox-scene.com if they need me to write the article and do it... maybe i'll give it to you...

This could be your big shot at being a computer info writer haha

 

[adairusmc]

Thanks for the info. It looks like an updated BIOS for my board is availble, I will flash mine and attempt overclocking again tonight, and see if I see a difference.

 

[aigomorla]

duckie, you chose the wrong platform. :X

Reason is because the striker uses a 680i chipset. That in turn is a freaken heat monster by itself. There is no cool running 680i chipset. Even DFI had to resort to a tower chipset cooler to take that sucker down.

A P35 chipsets run a bit cooler, and the board is able to shell out more heat then the 680i. Also the P35 doesnt have so many annoying FSB walls.

Also i wouldnt compare the 680i as a caddy. I would more put it along the lines of an old Jaguar. They work great for a while, but then they break down REALLY FAST.

 

[adairusmc]

I flashed my motherboard with the new BIOS, and so far I am not seeing any different reported temperatures. I am also not seeing any increase in my overclock either.

The BIOS update did kill my Razer Copperhead mouse though. I don't know if it actually killed it or not, but the machine will not POST with that mouse hooked up (works fine on my Logitech G7 backup mouse).

 

[BonzaiDuck]

AdairUSMC
[Remark to AigoMorla follows]

Wondered what happened to you, AdairUSMC.

I can't explain your mouse problem, but with over-clocking, I was pretty forthright about my own disaster.

Let me explain what I did. I pretty much followed your initial over-clock to 3.001Ghz and 1,334 FSB with DDR2 at 667. I had the voltage set too high initially, and kept setting it down until I started having 2xORTHOS errors, then found the best setting for my board -- it was about 1.31875V. This gave me a monitored value of 1.29 at idle, and I'm not sure how it "drooped" during load, but it was stable. That may have matched your 1.29 set value if there was no discrepancy, but anyway, it was close.

But with this, I had temperatures with a peak 54C at 73F -- scaling upward at 1C per 1.8 x delta-F degrees -- I think I was testing some of my settings at 79F room ambient.

But I couldn't seem to get stable overclocks at 1,440 and DDR2 720. This was fairly easy with the C2D E6600, and I had it up to about 1488 and stable with that processor. When the OC settings meant VCORE that looked as though the increases were moving up the steep side of a parabola for equal increases in speed, I didn't want to push it there. So I dropped the multiplier. I had a very stable setting at about the same temperature regime at 2.88Ghz 1440 FSB/ 720 DDR2. And I had the VCORE set at 1.30V. At about 2.92 Ghz, though, I hit a wall, and that setting was unstable. To get there, I had to run the VCORE up to about 1.31875V.

When I flashed the new BIOS, it "seemed" that my temperatures at the lower multiplier were also lower -- but remember my VCORE was now lower also. However, I've got it back up to 2.96 Ghz and 1480/740. In fact, the VCORE is only 3.125V -- lower than the setting I had to use with multiplier 9, or the failed attempt with multiplier 8 at 2.92Ghz. And something seemed to happen in the way the temperatures were distributed across the cores -- it was more even. For the lower VCORE, the temperatures ARE lower . . .

AigoMorla -- I think you had either an eVGA or nVidia 680i. The entire design behind the 680i was ambitious, because there are some 49 PCI-E lanes, and it uses a lot of power. I'd been looking at the (now old) P5WDG2-WS-Pro board in summer '06, and watching how things looked with the nVidia and 965 chipsets then. I wanted the extra PCI-E bandwidth, and the board matched the number of PCI-E slots and slot-placement that fit my plans. Besides -- I'm very "pro-nVidia" in graphics-card choice.

I still think it's possible the board you were using was weaker because if its voltage-regulation, but that's why I decided to splurge on the Striker board. I think I knew before the beginning of the year that there were OC'ing limitations -- minor shortfalls -- when compared to other 680i boards or even other chipsets, but you see, just getting to the fastest over-clocks was not the top priority in my strategy for this build.

Haven't yet tried H20 cooling with it yet, but on air [with my Xacto-knife and foam-board mock-up] it's fast enough for me. Somebody had complained that they can't get to 400 Mhz and run their DDR2 800 memory 1:1 to the full spec with it, but I've seen reviews that are practically guideline-tutorials for getting that far and farther -- although you have to drop the multiplier. One example had the external frequency pushing 500 Mhz using a multiplier of 5 or 6 -- I think -- with the E6600 CPU. But they did it. [In fact, I think I'll visit that review if it's still available on the web and take a closer look.]

I can post my temperature frequency distributions -- I just finished running 8 hrs 2xORTHOS at the 2.96 setting with 370 Mhz external frequency. At 73F ambient, the peak value is "the only" value -- pretty much -- rock-solid at 54C -- in fact, without looking at the data, the four cores were 54C, 50C, 48C and 52C. The temperatures scale up to the high range of room ambients for that 8-hour run of around 78F -- predictably 5F / 1.8 degrees C higher -- peak values bouncing between 56C and 57C. Occasionally it would clip 58C (the hottest core), but I'd look at the digital thermometer here and the room ambients were 79F-something.

 

[BonzaiDuck]

I'm sure you guys will give the understanding nod to my typo, but just to be clear:

"In fact, the VCORE is only 3.125V . . . . "

I mean 1.3125 . . . .

 

[aigomorla]

had the striker.

absoltuely hated it. IT was on on my first quadcore rig.

Reason why wsa because the board was borked from the start. Also the bios updates were uber slow. They couldnt keep up with the errors the 680i had.

the EVGA was a lot more stable, but its not stable compared to the P35.

In all, i gave up on nvidia for boards on intel. They admitted it themself, there still learning the intel market. NF series boards are ment to stay with AMD's.

 

[BonzaiDuck]

Actually, I had those very thoughts before I migrated to the 680i. If it's an Intel CPU, there's good reason to stay with an Intel chipset. That COULD extend to the motherboards, but the older Intel boards left so few options for tweaking that I became an ASUS customer. I realize there are many good board-makers out there -- DFI and Gigabyte among them. And if the way the market integrates itself didn't leave these little imperfections as with Intel failing to make an agreement with nVidia about SLI, the choice of an Intel chipset would've been a done-deal for me.

I told people six months ago that nVidia was new at the game, but I took this plunge after weighing the risks.

On the up side, I got a stable over-clock toward the goal of getting 3.00 Ghz with multiplier 8 so I could bump up the FSB 1:1 with the RAM. That is, I got as far as CPU-FSB of 373 Mhz and Q6600 @ 2.984 Ghz. Getting to a perfect 3.0, however, was problematic and I gave up after a system reset made part of my Event Viewer log-files "disappear." That's gotta be file corruption, and I really don't want to reinstall the OS. Here, I'm only guessing that the EV logs were the only volatile files open besides the CoreTemp log. I ran a CHKDSK and RAID-verify (still running), and can't find anything wrong -- yet.

I've yet to go back to multiplier 9 and try using a divider with the high-end Crucials at an even 3 Ghz. From the survey being run at the CPUs and Over-Clocking forum (I saw AdairUSMC's post and yours), I can see the obvious advantages to the G0 stepping. But at this point, I'm going to keep my wallet zipped for a while. I can't tell how the options will look for me when Penryn is released. The G0's should "be around" for a while, and the price on those should also tend to slide as Penryn comes along.

I tend to find a motherboard I can work with and choose to live with it for a couple years. This hasn't been a terrible experience for me, as it seems to have been for some others that bought the Striker, but of those so displeased, I thought I could tell from "customer reviews" which of those folks didn't really know what they were doing when they made their final, damning pronouncements. Some of them didn't even see that the "Linked" option would attempt to run the memory at the full FSB speed, or 1:2.

I'll probably take a look a the P5K board, but there are features related to layout and other matters that would preclude a switch for me at this time.

 

[idiotekniQues]

striker bad

intel chipset good

 

[firewolfsm]

The striker is beyond just buying a high-end board, it's a fucking joke.

A $130 P35 board will overclock better in every situation.

 

[BonzaiDuck]

I've poured over several reviews of the Striker Extreme over the months since last fall. In comparison to the other alternatives then, any shortfalls in over-clocked benchmarks were slivers of differences hardly worth the trouble to present in bar graphs.

Several of these reviews put the board through its paces over-clocking at several different multipliers for Core-2-Duo using decent memory choices.

But the price of the board was beyond what other 680i boards commanded in the marketplace, and it was marketed for an unsophisticated gaming public who might have had too much money to spend. C'mon. Who are they appealing to? With a board or model-name like "Striker Extreme?"

Now that I look at the P5K board, for instance, it doesn't offer me what the Striker does, in terms of expansion slots and slot-placement.

There's no contesting that fact that the CPU maker has been in the chipset business for years, that the Silicon Image controller on the Striker falls short of the ICH8R and later offerings, but then I didn't buy the board for the on-board RAID controller, either.

There is, nevertheless, one thing to consider, and that is how chipsets and BIOS updates get revised and improved over time. It's been said in corporate and governmental conferences on the evaluation of contract software and maintenance that you can give higher evaluation ratings to a software team that makes more frequent revisions. And there is a maturity factor in the motherboard market such that it takes many months from the time of initial release to a point where BIOS features have been improved and debugged through revision and chipsets have been revised or the boards have been revised and brought through new production runs.

The Striker BIOS' went through a period of being "borky," but the number of revisions should be noted. Unfortunately, these processes move forward at a certain rate while the technology in general moves forward at a certain rate, and Intel has long-since been tuned in to the market signal of enthusiasts who want over-clockable boards and a variety of features. And if there is one thing about a dominant firm -- just short of being a monopoly in a field with two dominant firms or a duopoly -- they do tend to push more and more capital toward the customer-base and customer-support.

The Striker was last-year's high-end, over-priced board that had a lot of promising features. Boards like the P5K and chipsets like the P35 are "this-year's" boards. But I'm pleased with the Striker's features for what I wanted from a board, and it'll be interesting to see how long it takes before technological change makes me long for "this-year's" or "next-year's."