Maximum PC

Okay, so I got my replacement Sandy Bridge board (Asus P8P67 LE) with my 2600K chip, and added a new Cooler Master Hyper 212+. Things are running well, and since I keep hearing how well these chips overclock, I decided what the hell, let's overclock for the first time.

Now, since I'm new at this, I decided to keep it simple. The Asus BIOS has a OS Tuner setting which "automatically overclocks the frequency and voltage of the CPU and DRAM for enhanced system performance." So I tried this, it rebooted, fiddled around a bit, and started me up with some new settings. Bclk is now 103, turbo multiplier is 43, for a 4.4 GHz overclock. All the other settings are standard or "Auto" (which is what the cpu voltage setting is on).

So, everything is going pretty well with this setting, so I'm pleased so far. However, I did have one question about it. I was testing with Prime95 to make sure everything was stable, and to check my temps under load. Temps aren't bad, max in the low 70s on occasion on one core, the rest generally high 60s. But I noticed that while running Prime95 (8 threads, blend), while most of the time it ran at the full 4.4GHz, it occasionally dropped down to the default 34 multiplier, while still showing 100% CPU usage. This didn't seem to correspond to any spike in temperatures, although like I said they didn't go that high anyway (I saw some low 80s with the stock HSF before).

So, can anyone tell me what is happening here? Should I be seeing these drops, or do I need to change some BIOS setting? I don't see the same drop when running other tests, like IntelBurnTest, for example.

Thanks for the help.

How long did it stay at that clock? If it is only for a brief second or 5 then I would say let it be.

It goes at full tilt for a number of minutes, then when it drops, it stays down for about 5 seconds, then revs back up to the full overclock, which goes around 15 seconds, then drops back down to 3.4GHz. This pattern continues from then on, although I did see a few times when the 4.4GHz went for maybe 30 seconds.

All good Mobos have cpu throttling settings that protect your cpu at high temps. But, most extreme overclockers turn these off for higher overclocks. I always did.

That is certainly true, but I don't think temps in the 70s qualify as "high temps". I got much higher temps without an overclock when I was using the stock cooler, with no throttling going on before.

ttfitz I did the same thing you did. And my set up is pretty similar and this is what it gave me on the fast overclock auto tune. I tried extreme first but blue screened.

4.43 Ghz
BCLK/PCIE 103
By All Cores 43
Internal PLL Voltage Auto
Memory Frequency 1648Mhz
CPU Voltage 1.216 V

My temps didn't go higher than 67C on full load about 45 min prime 95, and idle its about 37C using hw monitor. The only thing that has me worries is my system on a cold boot will turn on for 4 seconds turn off, then turn on and boot normally. I'm not sure why it does this did you have anything like this happen?

Seems like we ended up at about the same place. Did you see any slow down with Prime95 when you are running it like I did?



Yeah, my original Sandy Bridge board did that, but the replacement board didn't - until I did the overclocking, and now it does the same. I tried a number of fixes I ran across online, none of which worked, so now I just let it be - even with the delay, my boot times are WAY faster than before.

I didn't pay much attention but ill run it when I get home and let you know.



So you already replaced your board and both did it? I wonder why it does that. If I had to guess maybe its a safety feature? Or something a future BIOS update will fix. I guess I don't need to worry about it.

Supposedly there was a BIOS update that fixed it (and apparently it did before I started overclocking), but it doesn't for me. I've just assumed it was part of the overclocking - some testing of the memory at the higher clock or the like. It even occasionally does the turn off twice, which points to something like that.

When I got home I turned my computer on and it after it turned on and off a few times it told me overclocked failed. Ive been told on this forum and others that the b clock should never be over 100. I changed mine from 103, to 100 and my multiplier to 42 so I can get a nice stable 4.2 Ghz. Now that I changed my b clock I think it will run 4.4 but I don't see the point right now as its already fast. Ill push it as it gets older and I'm thinking upgrade.

I did run prime95 when I got home and I did not see it clock down to 1.6 Ghz during the test. Ill continue the test and let you know if this changes. I was also told that changing my b clock to 100 will resolve the on off during boot. I haven't tested this but Ill let you know.

Changing the BCLK changes the RAM speed, which I think attributes to the booting shutdown thing. I have a 2500K and an Asus mobo does the same thing.

The problem here...

Is Auto Overclocking. You can't explain what happened to tell us anything, in which to improve upon. So either go into manual mode and do it yourself or half us all go into Asus +8 auto mode or something.

I have an Asus board, I run a K at 5.4...and it ain't in auto. Let's learn the basics here. So we have a Ferrari, let's learn how to drive it ourselves. It's MPC.com! Not being an ass intentionally, but ask a question. Put it in manual and start from there.

Did you know the Automatic transmission was made for those that can't shift on their own? (They said girls, but I'm not going there...not now at least).

Sandy Bridge marks the first generation of Intel processors to include a clock generator on die. The K parts are designated for overclocking because of the unlocked multiplier since the new architecture is minimally capable of OC'ing by the BCLK. When you change the BCLK on these chips you are in effect modifying the clocks of all integrated components to include the graphics core, CPU core, memory controller, etc. because they are all tied together now. In fact, if you go back and look at pre-release reviews of Sandy Bridge, you'll see there was great concern that these parts would not overclock at all. Intel addressed these concerns by releasing K series parts at a very reasonable price point when compared to locked parts.

If you are overclocking by multiplier alone, then you are already half way to your fix. Turn off Turbo, and turn off CPU fan power saving features and you should be solid.

mega-bump, but I just registered here just to comment on this cause im having the EXACT same issue

wich im currently discussing here: http://www.overclock.net/t/1324236/over ... ome-issues

difference is: OC isnt automatic, I did it manually per core (although on this board you cant set the Vcore setting, so that's the only thing auto, but it works fine (1.3V)

here's a picture I took, good example of what the OP is talking about:




Ive been trying EVERYTHING and havent found a solution. So if someone out there has one id be inmensily grateful.

my guess: the mobo is just bad for overclocking and it hits some sort of limit (defined by either heat or voltage or something) and automatically throttles down. Ive turned on and off every single power saving feature (and every single feature in general for that matter) and still cant fix it.

my way of dealing with it: I just use the OC anyway (currently at 4.5ghz). Im yet to see a game that will stress a CPU as much as a prime95 blend test, so it shouldn't be an issue, but it would sure be nice to fix it.

Fuck the "LE" at the end of this board should have gone with p8p67 PRO

New threads for new topics are crucial. Each OC/temp discussion are very individualistic and have their own issue. Although you feel you have the same issue as this OP, you more than likely don't.

The OP here, I'd think the heatsink is crooked or not installed properly. 10C on a core sensor is huge and just tells me the heatsink is leaning a bit (60C on 3 and 70C on 1). And if this Auto overclocking doesn't put hard numbers into the BIOS like manual OC'ing calls for, it could still have EIST enabled and why temps and freqs are moving erratically.

For you, Salsa, you would not tag onto a common overclocking topic in a thread, but start your own because of all that goes on with the process. We're all different, ya know? Your graph tells me the CPU is cycling down during the tests as it switches, EIST is still on, and nothing more. The law of averages state that you still don't have bad temps but for sure have a LOT of tweaking to do inside that BIOS.

The OP hasn't posted further on this thread because some people don't really pay attention to the question, make basic mistakes about what was said, and obsess over trivia. For example, I said "Temps aren't bad, max in the low 70s on occasion on one core, the rest generally high 60s" and all you saw was "70" and "60" and came up with "10", when in fact it was less than 5 (don't remember exactly since it's been a while, but it wasn't enough different to expect it to be a problem). You fixated on the word "auto", and didn't notice the details, which explained that it was the process that was "auto", which produced actual "manual" settings.

From what I can tell from Salsa's post, it is EXACTLY the problem I was running into. I quit looking into it when I decided it didn't seem to happen anywhere else that I could see except Prime95.