Phenom II X4 965 BE Rev C3 @ 4.0GHz Locked in tight

[lucasa]

So far I am running this stable, but I'm still in the testing phase and wanted to see what people thought. My monitoring information comes from CPU Z and HWMonitor and settings were configured via the BIOS:

ACC Enabled, Manual Core Control Disabled, +2 to every core

BUS : 250MHz

MULTI : X16.0

CORE SPEED : 4000MHz

HT LINK : 2000MHz
WIDTH : 16 BIT

NB FREQ : 2000MHz

DRAM SETTINGS : Dual Channel; Unganged; 1333MHz; 1.5v; 7-7-7-23, TRC 31, 2T

V CORE : 1.525v-1.55v (it runs at 1.525 under load and is set to the same in the BIOS)

other voltages were set to recommended specifications


My max temps are ~54-56 C for the Per Core temps, and the TMPIN diode reports a max of 48 C, while running Prime95 Small FFTs test. I will run the other torture tests after this runs for at least 6 hours stable without a spike in temps. My idle temp so far has been 33 C, in a fairly cool room due to recent storms ( no A/C though :/ ).

Anyway, let me know what you think, if you have suggestions or feedback, etc.


My Setup :

Antec Three Hundred Mid-Tower Steel Case, all fan slots occupied and running at max usually.

Asus M4A79XTD EVO Mainboard

Processor is in the title

Zalman CNPS 9900MAX 135mm Circular Composite heatsink, using the thermal paste it came with, good stuff.

Nvidia GTX460 1GB @ 900/1800/2200 (with flashed BIOS and slight volt mod)

WD 1TB 7200RPM HDD, WD 250GB 7200RPM HDD (both SATA II)

Corsair GamingSeries 700W PSU

 

[Rustyy117]

I don't think you could really go any higher with your CPU speed as your temps are alittle high. Maybe try and get that RAM to run at 1T?

 

[henydiah]

4ghz vcore 1.52 - 1.55 to high for X4!
my vcore X6 1.49 -1.52 v for 4Ghz when disabled 2core use vcore 1.42v-1.44v for 4ghz. Cpu nb jus raise at 0.025v

 

[spentshells]

How are your temps (in the case ) ? otherwise looks awesome

 

[lucasa]

4ghz vcore 1.52 - 1.55 to high for X4!
my vcore X6 1.49 -1.52 v for 4Ghz when disabled 2core use vcore 1.42v-1.44v for 4ghz. Cpu nb jus raise at 0.025v

Oh I knew that those voltages were high I just wanted to ensure stability, I hate it when I have system lockups LOL. I've gotten 4ghz before by just upping the multiplier, and only had to use 1.4375v, but I decided to lower the multiplier and raise the bus by 50mhz, and figured that this called for a larger increase in volts. Right now I am running it on 1.4500 in BIOS, been running stable all day, ten hours while I was working. My idle temp only dropped to 34 C. Which actually brings up another question, will CNQ work with a bus overclock or is it disabled just like raising the multi past x18?

Load temps are now around 50-52 C core temps, and the processor's temperature diode does not go past 46C.

I was surprised, and maybe this is because I don't run DDR3 1600, but I didn't have to overclock the northbridge frequency at all, I just lowered the multi's till they went back to 2000 even, and I have not noticed anything wrong with keeping it that way. However, most of the OC's that I have seen run the NB @ 2400-2600mhz. Why is it that they overclock it but I don't need to?

 

[lucasa]

How are your temps (in the case ) ? otherwise looks awesome

Oh yeah, the motherboard diode reads 28C max, 24-25 with system at idle. Not sure of the exact temps, but based on the quality of the fans I have, the fact that there are 8 active fans in my case, and me always having my window up to get a nice cross-breeze, I'm probably ranging from 68 to 70 degrees fahrenheit. Once it gets hotter I will stop overclocking or turn it off during the day, until we get our A/C fixed.

 

[henydiah]

most of the OC's that I have seen run the NB @ 2400-2600mhz. Why is it that they overclock it but I don't need to?

NB very important for phenom II .. maybe this Article can explain you question :

 

[megamanx00]

With those temps you may be able to push for a little more, perhaps 4.1 or 4.2. Of course, it won't make too big of a difference but if you want to get all you can out of it, and don't mind it having a bit of a shorter life, then go for it

 

[lucasa]

NB very important for phenom II .. maybe this Article can explain you question :

Awesome thank you for that information, it makes more sense now as to why it might be necessary to up that when overclocking.

I changed a couple of things around, now my CPU northbridge is running @ 2600mhz with a voltage of 1.2v.

I have also changed my RAM's Command Rate to 1, and the only issue that I noticed is that Worker # 3 in Prime95 got one error when I ran the blended stress test, which tests the RAM a lot. It was a rounding issue, the rounding was 0.4944152832, and the test expected a response less than 0.4. This was on hour 11 after completing 508 tests without issues. None of the other workers experienced any errors, and completed 18 hours of testing with around 920-935 tests completed per worker (except #3). Is this an issue or not a huge deal?

 

[henydiah]

correct in the test of testing no error message, I just asked if there are error messages in 6 hours at the time of testing, because many consider if in 6 hours no error message that means stable. but the night you perform the test will be more stable.

for more details you try this test with the IBT : http://downloads.guru3d.com/IntelBurnTest-v2.3-download-2047.html will take a short time maybe 15-25 minutes is completed, and if IBT no error message then you can say the system is stable.

 

[browsingtheworld]

Run OCCT linpack overnight to see if the CPU is really stable...

 

[lucasa]

correct in the test of testing no error message, I just asked if there are error messages in 6 hours at the time of testing, because many consider if in 6 hours no error message that means stable. but the night you perform the test will be more stable.

for more details you try this test with the IBT : http://downloads.guru3d.com/IntelBurnTest-v2.3-download-2047.html will take a short time maybe 15-25 minutes is completed, and if IBT no error message then you can say the system is stable.

I found the problem. I was overclocking the cpu/nb frequency without upping the voltages, which led to my rounding errors and IBT stability warnings. Current settings:

CPU BUS: 200mhz
CPU MULTI: 20x
CPU FREQ: 4000mhz
CPU VCORE: 1.450 (1.43-1.48 in HWMonitor)
CPU/NB FREQ: 2600mhz
CPU/NB VOLT: 1.400v

HT: 2000mhz
HT Volt: 1.200v
HT WIDTH: 16 bit

DRAM FREQ: 1333mhz
DRAM Timings: 7-7-7-21 TRC 28 2t (I tried 1t but I get memtest errors).
DRAM VOLTAGE: 1.5v


I ran ten passes in IBT - Maximum Test, utilizing 2556mb of RAM to run the equations. All results match, no errors.

 

[henydiah]

Your cpu nb volt too high , Drop cpu NB volt to 1.27 and drop your NB freq to 2200 - 2400, (remember IMC risk ) and raise vdim to 1.6v and loosen timing ram at 7-8-8-21 ... Test again use 20 at maximum ram. Alway's keep eyes at temp much below 60C., if over stop and drop your oc.

 

[lucasa]

Best answer selected by lucasa.

 

[lucasa]

It's been awhile since my last post, and after months of researching, tweaking, and tweaking some more, I managed to achieve a VERY decent overclock for my processor, especially considering I run 64 bit Win7, which many say holds the 965 BE back from being able to hit 4GHz.

My 965 BE is currently running @ 4.08GHz (255MHZ Bus x 16 multi) @ 1.5v. Everyone says 1.5v is a little high, but there are no issues whatsoever with temps, and if I'm below the maximum safe voltage specified for this chip. 1.5v coupled with the ability to continue using Cool N Quiet (due to a Bus OC instead of a multiplier OC) keeps my temps in a very acceptable range, from 24 C when idle to 48~49 C at full load (Prime95 and Intel Burn Test). While gaming the temps are lower by 1-4 C.

This overclock made it so that I had to slightly overclock my RAM though, from 1333MHz to 1360MHz, but I was able to keep stock timings and voltage with no negative impact; it's only a 27MHz increase, after all.

My CPU-NB and HT Link are both running @ 2550MHz (x10 multi) w/ ~1.28v on each. I also upped the mobo NB voltage to 1.28v to keep everything near the same voltage and ensure I wasn't hurting performance by undervolting something. My mobo temps were not negatively impacted by any of this; mainboard temps hold consistently @ 21 C at idle and 26 C under full load.

I am quite pleased with the results, IBT test run @ maximum (which is a little over 3072MB if nothing else is running) shows that the GFlops increased from 39~40GFlops @ stock settings to just over 53GFlops consistently.