OC'ing the Q6600 on the Intel D975XBX2 board

[ShawnP]

I was able to get the Q6600 a bit cooler by changing the fan on the Kingwin RVT-12025D Direct Touch 4 Heatpipe Heatsink. Probably because it is not the G0 Stepping.

[ThunderRd]

Yeah, the B3s are significantly hotter than the G0s. Actually, with any of the 65nm quads water is the way to go if you really want to see high FSB speeds.

OCCT will really raise the temps in a hurry if you aren't set up with good cooling.

Clocking a quad is quite a different animal from a dual, though. On my X38 board, my quad is not stable past 455FSB, although it is completely stable at 450. I run it 24/7 at 438x8, and gain .05V from 450x8. Just for laughs, I tested my board with a friend's dual (forget which one): I could run it stable all the way to 530FSB. Now that is a big difference.

[ShawnP]

Yeah, it is incredible the difference. I dropped it down to 2.9 Ghz now and OCCT is still shutting the test down after 3 mins.

[MUff1N]

Originally Posted by ShawnP

Yeah, it is incredible the difference. I dropped it down to 2.9 Ghz now and OCCT is still shutting the test down after 3 mins.

Shawn, in OCCT you can set the shutoff temp to higher settings if you want or raise it if its lower than it should be for your processor!
You should be able to safely run it 5c higher than normal with no damage done to the CPU since it will only go to these higher temps for very short amounts of time.

For example the MAX temp for my E6700 is 60c, so I set it to 65c & it's fine an allows you to set higher CPU speeds without it shutting off.
The setting is here in the screenshot below.

[ThunderRd]

@Shawn
Are you certain that the failure of occt is due to temperature? It is a good, and therefore very sensitive stress test. If something else is causing it to fail you could be in for a long oeriod of frustration barking up the wrong tree.

Why not set everything stock, 266x9=2.4g, memory on the 1:1 divider. Set the threshold temperatures in OCCT at 80C, and then see if it completes. If it does, drop the divider to 6, and see how high the FSB goes before it fails OCCT. That is an estimate of maximum FSB. If Muff is right, it should crap out at around 400 or so. You still will have the same potential to run 3.5+, but you'll have to go with the 9 multi because your board isn't capable of running the FSB as high as mine can. Then, go back to 266x9 and see what the memory limit is. Any, or all of the above could be failing OCCT; not just high temperature.


@Carlos
With 4 sticks of memory you are most certainly limited to 2T command rate on the memory; are you sure that you aren't set to 1T? That would almost without a doubt make you unstable.

[ShawnP]

I will try that, setting everything to stock and see what happens.

[ShawnP]

At stock on the Q6600 B3 OCCT temp reads 68 C and Core Temp 0.99.5 reads 53 C both across the board.

[ShawnP]

Now at:
Memory at 800mhz -- 5-5-5-15 -- 2.12v
Cpu Voltage=1.2875V
EPS=enabled
FSB VO=1.250
MCH/ICH VO=1.525
Host Clock Freq. Multi=281

What I get is 2.52 Ghz and it last for 10 mins before reaching 80 C in OCCT.

[ShawnP]

The Memory is set to default except changing the voltages.

OCCT is setup to 85 C.

[ThunderRd]

OCCT will ramp up the stress in intervals; it will run hot for awhile, and then cool down somewhat before rinsing and repeating. Take a look at the log graphs and you'll see what I mean. 66 isn't too bad, it's a bit on the high side but that chip will take it.

I have to ask, though: are you absolutely sure the heatsink is mounted well and you used the correct amount of paste? If you are, continue the test until it finishes at least 2 hours or craps out, and make a note of the cause of failure for us if it does fail. If you aren't sure about the sink, redo it.

Doesn't look like you have a lot of temperature headroom to overclock if you are getting those kinds of temps at stock. That is unfortunate.

[ShawnP]

Funny you bring up the heatsink, LOL. I thought that myself so I disassembled everything cleaned and lapped it.

Now same setup, 10mins into OCCT test and I have CPU 0: 69 , CPU 1: 69, CPU 2: 69, CPU 3: 69.

[ShawnP]

Aaahh Man, you've got to be ****t'n me,--------------thats exactly what I said to myself when I discovered that my heatsink is not meant for Vertical applications. So I put my case on its side. This time it's actually bumped up to 2.6Ghz, past the 10 min mark and the temps haven't gone over 57 C.

[ThunderRd]

That's more like it.

OCCT on my q6600 clocked at 438x8 is 63-64C at full load. The SMP folding client runs at about the same. I'm in a hot ambient, though, so you should be satisfied with those temps.

Now try dropping the multi and see how far the FSB will go, and report back. Keep the memory at 1:1.

You might think about building some sort of brace for the HS that will keep it true.

[ShawnP]

From what I've read I don't think a brace would do it. This model of KingWin is just meant for horizontal applications. Although it doesn't explicitly state that. I tried changing the direction of the mount, stood my case up vertical and the temps wouldn't hold under 70 C. Put the case on the side again and they are just peaking 60 C. I think I will have to keep this heatsink for a Home Theatre build and swap in a different heat sink.

[ShawnP]

My heatsink is exactly same design as this:
NCIX.com - Buy Xigmatek HDT-D1264 CPU Heatsink AM2 LGA775 S754 S939 S940 120MM Fan 800-1500RPM 56.3CFM 27.2DBA - ACXTHDT-D1264 In Canada.

[ThunderRd]

Hell, the stock HS would do better than what you were getting...I wonder if it would be better if you rotated it 90 or 180 degrees. I'd still try testing a brace under it before I gave up, though.

[MUff1N]

Originally Posted by ShawnP

OCCT reads 66 C on all four, Core Temp 0.99.5 reads 51 C on all four

I think OCCT is not reading the correct temps as it defaults to the Bios readings & I don't think they are correctly read in this case!

I would go off of Speedfan temps & set OCCT to use those reading instead because they read the same as Core Temp does on the cores (screenshots) & I know these are the correct core temp readings!!!

So use Speedfan for OCCT & that will give you a -15c reading!!!
This is a big diff in temps & the correct ones as well!

[ShawnP]

Actually Speedfan and OCCT are giving the same readings on Temp. Only CoreTemp seems to be reading accurate. I base the accuracy off of one test I did where the numbers hit red in CoreTemp and that is when the computer rebooted because of overheat.

[ShawnP]

On OCCT when you start getting "Error Detected on Core 1 (or 2 etc.)" does that mean that the voltage needs to be adjusted?

Also I was running 1:1 with 4-4-4-12 at 3.2 Ghz when I went I tried something a little to extreme and the board wouldn't boot up. So I set the Performance section to default and flashed the Bios and tried the settings above that were working and they no longer work. Hmmmm.

But it is running at 3.2 Ghz 2:3 now. I will post some attachments.

[MUff1N]

Originally Posted by ShawnP

On OCCT when you start getting "Error Detected on Core 1 (or 2 etc.)" does that mean that the voltage needs to be adjusted?

Also I was running 1:1 with 4-4-4-12 at 3.2 Ghz when I went I tried something a little to extreme and the board wouldn't boot up. So I set the Performance section to default and flashed the Bios and tried the settings above that were working and they no longer work. Hmmmm.

But it is running at 3.2 Ghz 2:3 now. I will post some attachments.

Yes, if you are getting errors on the cores then you need to up the vcore.

Make absolutely sure that all your OC settings (frequencies, voltages, dividers, etc)
are set to default settings BEFORE you reflash your bios so that it flashes right.
This is the reason that settings that worked before don't anymore is you didn't completely do this &
more than likely missed a setting or two.
At least this has been the case for me...

On Speedfan, if you go into Advanced options you can adjust (re-calibrate) the temps to where they are exactly the same as Core Temp!
You can adjust the DTS sensor to + or - to where it's exactly the same. (Screenshot)