How well does your Opteron overclock?

[benbaked]

i'm really interested in seeing others overclock results with the CABNE 0544. i can only find very little about this stepping through google, and hardly any results.

i know some of you have this stepping, what are your results?

i lowered the vcore on mine to 1.45v, it's still handling 2.7ghz well.

 

[roy_o_banion]

You need two installs of p95 for dual core. Then you need to set each one at an affinity to a core. 24 hours should be good to tell if you are stable. Also, prime doesn't really complete, it is just a continously looping program.

 

[aeiou]

Type: single 144
Stepping: CABNE 0544EPMW
Best stable OC: 2.6Ghz (293x9)
Voltage: 1.45
Temps: 36/43

Edit: Updated- stock cooling

Edit2: Updated again....I think I've maxed out the stock cooling now.

 

[kovboi]

CPU Type: DC 165
Stepping: CCBWE 0534MPMW
Best Stable OC: 2601 (9 x 289) / 266 divider
VCore:1.425
Temps: 31/44

 

[Super Nade]

Post #1 updated, 8 December 2005.


My DC Opty 165 is here. From the sparse results available, this seemsto be a decent stepping.

CPU Type: Opteron 165 (Dual Core)
Stepping: CCBWE 0534 MPMW 131606AK50026
Best Stable OC: 2.55GHz, still testing @2.6GHz
VCore: 1.33V
Temps: 27/38 C

 

[kovboi]

Post #1 updated, 8 December 2005.


My DC Opty 165 is here. From the sparse results available, this seemsto be a decent stepping.

CPU Type: Opteron 165 (Dual Core)
Stepping: CCBWE 0534 MPMW 131606AK50026
Best Stable OC: 2.55GHz, still testing @2.6GHz
VCore: 1.33V
Temps: 27/38 C

Looks like we have the exact same stepping. The highest I could go on stock voltage (well, 1.35v) was 2538 (9 x 282). I was able to just barely hit 2.6 without too much more voltage, but only just; anything above 290 and one of the cores started to fail. 'course, i haven't tried your lengthy burn in process...yet.

edit: just read your other thread. looks like you were a bit luckier with your opty!

 

[Super Nade]

Well actually, StressCPU failed in about 20 sec @ 2.62GHz.
That's why I'm burning it in with CPUBurn (error-cheking off). I hope 2.6 would be stable when I get home. Now mate, tell me everything about what you did and what happened. Why are you hitting a wall at 2.6? Note that my RAM is always at DDR400.

I suspect it could be your BIOS. ASUS make good boards, but the BIOS support is crap. I remember when the Winny came out, people were having a real rough time (check the A8V guide in my sig for details). Note that Motherboard makers CANNOT officially support Opterons. I suppose AMD has been pressurizing them to this effect.

 

[kovboi]

ok, here's what i did:
initially i set vcore to 1.3125, LDT to 4x, and left the memory at 1:1 with relaxed timings (2.5-4-4-8). i bumped up 10 mhz from stock and tested with two instances of sp2004 for 15-20 min.s.
210 @ 1.3125 = fine
220 @ 1.3125 = wouldn't boot up; tried upping voltage
220 @ 1.3375 = still wouldn't boot, so tried 333 mem divider and it worked fine.
230-250 @ 1.3375 = no problems
at this point i lowered LDT to 3x and mem divider to 266.
260 @ 1.3375 = wouldn't boot into windows, increased voltage
260-280 @ 1.35 = no problems
290 @ 1.35 = failed to boot, increased voltage
290 @ 1.4125 = core0 failed after 49 seconds in sp2004
290 @ 1.425 = fine
300 @ 1.425 = failed to boot
300 @ 1.4375 = fails while loading windows
300 @ 1.45 = core0 fails after 3 seconds in sp2004
300 @ 1.4625 = core0 fails after 13 seconds
300 @ 1.475 = core0 fails after 35 seconds
300 @ 1.4875 = core0 fails after 1:50
300 @ 1.5 = core0 fails after 2:32

now, i decided to back down slightly from 290 because the core0 failure was worrisome. i ran 2 instances of sp2004 overnight and it is stable. BUT the memory controller seems fussier than my venice. i ought to be able to use a 333 mem divider with relaxed timings since my memory will do 480mhz, but i can only get it stable using the 266 divider with tight timings (2-3-3-6). you may be right about the bios, however, as asus keeps putting out new versions without a changelog. anything you would recommend trying?

 

[Super Nade]

Mate,

I think you need to burn that bad boy for a long time. I would go back and burn it in at stock for about 12-14 hrs. Looking at your low stock VCore leads me to believe that your chip may be better than mine. You have to coax it first before upping VCore. This is precisely why I have not increased the VCore despite hitting a wall at 2.6GHz.

StressCPU failed in 5 seconds on both cores @ 2.6GHz. After an 8 hr burn, it failed opn one core in about 5 min and the second in about 30 min. So, there was some improvement. I ran CPUBurn last night. Lets see what happens now. If StressCPU fails again, I'll bump up the VCore.

 

[kovboi]

yeah, that's pretty much what i decided to do. by the way, the default vcore on my chip (that is, what the bios defaults to) is a bit higher at 1.33v. i updated my motherboard's bios with a new beta on the asus website and cpu-z now reads it as varying between 1.32 and 1.34.

good luck with yours. i'll post back after a long burn.

 

[NinjaWreck]

CPU Type: DC 170
Stepping: CCBWE 0530
Best Stable OC: 2800 mhz
VCore: 1.48
Temps: 32°C idle/50°C load

Best stock vcore stable OC:
2700 mhz
1.35v

Best super PI 1M stable OC:
3000 mhz
1.56v

 

[skane]

CPU : 148 CABYE 0543FPMW
Best Stable OC: 2920, 13+ hours Prime95
VCore: 1.54 set in BIOS
Temps: 32c / 48c in ITE
CPU will scale higher with more voltage, but cooling is limiting factor, so I settled on 2920 for now.

I also have a 148 CABYE 0543FPMW I think will do about the same as the one listed above, based on some quick tests I did with stock air.

 

[kovboi]

ok, i did a nice long burn overnight with 2 instances of stressCPU running for 13 hours at stock settings (vcore at 1.3375). now i started bumping up as per your method (50 mhz net cpu freq, burn for one hour, stress for 30 mins), but i'm new to the whole burn-in method. does it make sense to go straight to my last stable freq at stock vcore (9x250), or is there a benefit to working back up to it incrementally?

btw, i still had to lower my mem to the 333 divider past 215x9. on my previous cpu (3200 E6 venice), i could run it at 1:1 with 10x240. since neither the memory nor motherboard bios have changed since then, that means the culprit is likely to be the memory controller on the new opty, right?

 

[ROH]

CPU Type: Opteron 170
Stepping: CCBWE 0534 SPMW
Best Stable OC (so far): 10x250 (need time to go higher.)
VCore: 1.275x123%
Temps:idle=27 CPUBURN or dual Prime95 load=47 Game load=38

Been having a pretty good run with this CPU so far! The set-up I'm running seems to be liking it! once I find my max OC with the CPU I'm gonna try and match it at 9x as my ram can hit 310 @ 2.5,3,3,6 Memtest stable! Soooo, I'm hoping I can find a nice happy medium in there somewere!

I've noticed a big difference in bandwidth going from 9x to 10x when dealing with benchmarkers like memtest, everest or Sandra. I didn't realize the multiplier made that much difference! Learn something new every day! What the means to real world change is yet to be seen, though.

O.

 

[kovboi]

i've worked my way back up from stock freq and, lo and behold, i'm booting into windows on stock voltage at 9x261 (previously needed more voltage)! gonna keep pushing it upwards & see if i can catch up with you at 2.7GHz.

 

[Super Nade]

So this method works! Good good..

O.K mate, lets see if you can catch me. I'll be pushing for 2750 MHz tonight. You are gonna get smoked!

Holy cow! Ninjawreck, you are killing us man!

 

[ocZer]

CPU: Opteron 148
Stepping: CABYE 0540 FPBW
Stable OC: 2,75Ghz
VCPU: 1,475V
Idle temps:What's idle?
Load temps: 49C
Ambient: 25C
Cooling: Stock PIB
During testing,(when not folding ) I let it idle a bit and it settled on 30C, but within 1 second after starting prime or superPI it would jump 46C in a blink of an eye. I checked and rapositioned heatsink, but no change. I think this one is a good candidate for removing IHS. Idle / load temps differ to much, and I am unable cool it properly. Ran 1 hour stable at 3.03Ghz but temps are way to high for me te keep it there.

Update:
VCPU: 1,4V
Load Temps: 43C
Cooling: XP-120

 

[Super Nade]

49C is a bit on the higher side. What is your TCaseMax?
Heck, even my DC Temps are less that yours! During testing I found that Voltage alone has a bearing on temprature. There is no change in temps if I up the frequency. Are you using the stock HS? Once I'm certified stable at 2.7GHz, I'll swap out my XP-90 for the stock HSF and see what happens. I'll be running a detailed test in my other thread (see sig), so keep a lookout for that.

Finally, if you plan on removing the IHS, try dental floss. Its a lot safer than a razor blade.

 

[DaveB]

49C is a bit on the higher side.

Since when is 49C high at 100% load? I thought the flake out area for these delicate little AMD chips was 55C or so. Both my 146 Optys @ 3 GHz were fine at 52C or so under 100% load and my X2 3800 and X2 4400 ran fine @ 2.6 GHz+ above 55C with no problems. In both cases, 1.6V seem to be the best VCore for OC because they get too hot with more. I don't understand all the fuss about running at default VCore and keeping the temps so low all of the sudden. Just give it all the juice it needs and keep the temps under 60C at full load! Damn the torpedos, full speed ahead!

 

[deathman20]

Well I'll toss mine in here also....

CPU Type: 170 Retail (w XP-120)
Stepping: CCBWE 0541RPMW
Best Stable OC: 2.7Ghz
VCore: 1.54V (1.4V x110%)
Temps: 32*C/48*C

Also testing currently same OC and looked promising so far.
VCore: 1.5125V (1.375V x110%)
Temps: 32*C/46*C

Added note: All burning in was done at either 1.45V @ 2.6Ghz and later on 1.55V @ 2.7Ghz which happened still not to be stable for more then a few hours if I did prime it. Only after using special voltages its become stable at that speed. Go figure that one still has me puzzled, and putting out less heat too.