Dual Core Opteron Thread #2

[Cheffy]

dont bother with burning in myself, never really believed it exists. more that time spent with hardware improves the skills employed in oc..

i choose the speed i want, do everything i can to get it, if that fails drop down untill it works.

[WigSgiW]

Thanks for the replies all.

I thought that might be the case; with the experience I have had with overclocking the 3700, a really crappy 4400 and the moderately crappy 165, all on the same motherboard, I think I should be pretty good to go as soon as I slap the new 165 in there.

Nice to see that patience isn't always necessary - I would hate to have to (for max overclocks sake) slowly move up from 200 FSB in 10 increments to 334 FSB - it would drive me bonkers!

I will keep everyone updated when the chip arrives

[QuietIce]

The only "burn-in" (not really) time I can see being worthwhile it to run the chip 16 hours a day - and off at night - for a week at a moderate OC (but still on stock volts) just to get the TIM cured before the real OC'ing starts. Other than that I just can't see any good reason to waste time before hitting it ...

[ColonelCain]

Just got a LCB9E 0651RPMW. I saw another one that was a week 13, doing 2.86 @1.19v. Think that mine could do the same thing with a DFI SLI-DR Expert, naked, under H20?

[knopflerbruce]

Just got a LCB9E 0651RPMW. I saw another one that was a week 13, doing 2.86 @1.19v. Think that mine could do the same thing with a DFI SLI-DR Expert, naked, under H20?

The LCB9E's from week 50/06 --> seem to be very good. I wouldn't expect it to do as well as my 0613, though.

3g with quite low voltage is my guess, but I never bought a LCB9E (I guess i have to someday ) so there are other guys who know alot more about these steppings than I do.

[freakBYnature]

OCing has become a very quick thing for me, instead of ramping up to the max clocks I go over max stable speeds and move down. It's easier to tell that way what the max will be because you can look at Prime failure times and make an educated guess

LOL, that's mostly how I do it to... with the boards I'm familiar with though

[freakBYnature]

The only "burn-in" (not really) time I can see being worthwhile it to run the chip 16 hours a day - and off at night - for a week at a moderate OC (but still on stock volts) just to get the TIM cured before the real OC'ing starts. Other than that I just can't see any good reason to waste time before hitting it ...

Second that, tried that sh*IT with a faulty FX60, and burning in..... or "burning in" simply didn't do the trick! Just gave me a headache

[knopflerbruce]

I usually start with 3g @ stock, and if that doesn't work, up the voltage to 1.45. And then to 1.55 if necessary.

But my first chip I did it the "boring" way, 50mhz at the time - dual prime 8h every night. Since that chip did 2.9 @ stock voltage it took a while lol.

[david901101]

I got a new Opty from RMA today, a bit better than the last one (0615 MPMW @ 2583 mhZ with 1,4 volts) and now I have a 0615 DPMW

[LIKMARK]

I got a new Opty from RMA today, a bit better than the last one (0615 MPMW @ 2583 mhZ with 1,4 volts) and now I have a 0615 DPMW

You tested it yet?

[david901101]

You tested it yet?

It's running at 2661 mhZ with 1,35 v for now, since my motherboard doesn't like fsb speeds over 295 mhZ.

Update: Running at 2673 mhZ now with 1.31 v.

[Repoman]

Cool n Quiet with CrystalCPUID is awesome

Currently doing a 1160MHz @ 0.95v with 31C load as my mininum setting, 2320MHz @ 1.15v for my medium setting, and 2900MHz @ 1.3v for the 'max' setting The program automatically switches between these depending on CPU load. 1160MHz is perfectly responsive for idle Windows browsing and using Firefox.

I might even switch it up to 1200MHz, 2400MHz, and 3000MHz settings (switch HTT from 290 to 300). This way I wouldn't feel bad about letting it run at 3Ghz, heating up my room and wasting power because it would be at 300x4 most of the time

PS the key to getting it to work is using the "Auto" CPU VID setting in the BIOS.. then if you have a DFI board you can just use the VID Startup value to actually control the initial voltage applied, until CrystalCPUID takes over. YOu can also still use the Vcore special % thing, it's always added to the VID that CrystalCPUID sets

[beerking]

ho so good

[QuietIce]

Interesting. On the ASUS A8N32 (don't know about the others) enabling CnQ locks out the VID but uses the previous setting for startup. I used it for about 2 months on my sig rig (running @ 233x12) and everything worked fine. It would shift between 12x, 9x, and 6x as needed. After 2 months I started crunching SETI so I didn't have a need for it anymore. I did change to 255x11 after that to bump the RAM a little ...

[Cheffy]

I Use CPU Rightmark V1.5
set to max fid 12, max vid 1.4, mins 5/1.1 works a charm. uses pretty much every multi available depending on load.
then again, unlike most boards, the abit ax8 seems to set v-core by setting vid within normal cpu settings then adding additional voltage on top, so most apps/utilities report 1.4v all the time, rather than the real volts. when multi/vid are dropped, the extra on top is maintained!

[WigSgiW]

Alrighty, I received the chip, it is a LCB9E 0704 TPMW.

Not the best ones of the LCB9E stepping, however it is interesting how I faired.

I am now up to 2700MHz (300 x 9), with my ram at 225MHz DDR 2-3-3-5; 2.9v. The CPU is definitely stable, as can be seen in the below screenshot, with Orthos running at top priority for 11 hours 53 minutes. Voltage = 1.35v - 1.40v fluctuating (1.35 in bios, with around a 0.05 vdroop on this Epox board - known for it). Nice.

This chip is definitely capable of more, however I am 95% sure that my motherboard has let me down with this one. I was initially stress testing @ 312 FSB (2808MHz) and had it stable at this setting through Orthos @ 1.45v in bios (not the full 12 hours tho, woohoo!) however Windows would often randomly lock up, and I would have huge troubles with cold boots - it would get stuck in the same place every time (code 50 on my Epox motherboard).

Tweaking chipset voltage made no difference whatsoever; would lock up no matter if the chipset voltage was 1.5v through to 1.8v (maximum the board will give it, and I don't trust my PSU much beyond this point anyway).

Dropping back to 300 FSB means a full 0.1v less vcore, and the CPU is solid stable (still testing memory with Memtest 86+ right now). Interestingly; I have stress tested up to 334 FSB before on this board (with CPU downed on the multiplier) and it was generally fine - this was some time ago however so the year of general aging would probably have put a stop to that.

Either way, happy with a 50% overclock on stock voltage, with core load temps being 42c max (PWM 49c); gained myself 200MHz from the last chip with 0.1v less as well

Click below for full image

[Heikit]

very interesting i put my FSB 1:1 290, everest shows it right, but why doesn't Cpu-z show? :S

[stealth]

very interesting i put my FSB 1:1 290, everest shows it right, but why doesn't Cpu-z show? :S

What FSB and what multi the CPU tab shows on CPU-Z?

The everest gives CPU/9 and CPU-Z CPU/10.

Can you give us a validated CPU-Z?

stealth

[TMM]

10char......

[QuietIce]

Opty 165 LCBQE 0711SPAW currently running at 300x9 @ 1.3v on water. No SS yet - I'm not done OC'ing ...

[bonzo_catalin]

OPty 170 LCB9E 0704XPMW @ 3ghz, IHS removed and cooled by Thermalright Ultra-120 eXtreme w/ 1000rpm Nexus.

I had to bump the voltage up a tad to 1.50v to get it 24hr stable though

Careful with that PWM , it's a little bit high .

[dsptech]

I just swapped out my Opty 165 for a Opty 170 LCB9E 0651 RPMW.
I'm 10 hours in running dual prime on Linux at 10 x 300 1.4V.
I'm hopefully on my way to the 3GHz club for the first time.
This one is running cool at 40c under prime on water.

My opty 165 took me to 2800MHz but lost stability over time.
Was last running it at 2.5 GHz.
That chip ran hot and needed 1.6V to achevie 2.8GHz.

My 170 was bought it at Newegg and it arrived two days ago.
Running DFI Lanparty UT NF4 SLI Expert.
1GB Crucial Balistic Tracer.

I'll post more once I'm sure it's prime stable.

[TMM]

Careful with that PWM , it's a little bit high .

I have a fan blowing over it and it hovers around 75C usually . Without a fan it goes up to about 86C. These LCB9E are just very power hungry chips . My PWM never exceeded 50C with my single core opty

[gromet]

I'm running a 165 CCBBE 0617 @ 3.0 stable. 333x9 FSB. I did remove the HS for temps.

I would hope any 170+ could match this or do better honestly. I just dont see how 185's are barely getting to 3.0ghz and calling it good....I would shoot for 3.2+. I would love to get a 185 and play with to go above 3.0ghz, lowering the cpu multipler and increasing the FSB to get well above 3.0ghz seems possible..but S939 is slowly dying, and plan on doing upgrade at the end of the year.

have a fan blowing over it and it hovers around 75C usually . Without a fan it goes up to about 86C. These LCB9E are just very power hungry chips . My PWM never exceeded 50C with my single core opty

Sounds like the issue I had...My core didnt have very good contact w/ the HS. Remove the HS and temps droped from 40idle to 22idle, load 65 (3 hours orthos) to an average of 37.3 (24hours of orthos)

[stealth]

I just dont see how 185's are barely getting to 3.0ghz and calling it good....I would shoot for 3.2+.

3Ghz at 1.3V.

http://img177.imageshack.us/img177/318/89268065op5.png

http://valid.x86-secret.com/show_oc.php?id=216598

Probably this chip can do a lot more,the only problem is how to keep temps low.
When i get the watercooling i will give it a try with ice water.

stealth