The Official Opteron 939 Overclocking Thread

[fullup3]

Ok, after testing my 146 CABNE 0530APMW overnight, it seems to need a little more volts than some of the other ones like it to clock really well. This might be due to my board, or my cooling.
My max case temp of the chip is rated at 65degrees. I'm wondering how hot is too hot to run this chip safely.
I need more than 1.5v to prime any higher than 2900mhz. At that voltage, prime causes my chip to run at about 57 to 58 degrees under load.
I want to go higher with my clocks, but I also am afraid to damage the chip.
what do you guys think.
Should I go ahead and push a little further, or should I stay where I am until I can get some water cooling in a month or two.

[Deus Falsus]

From what I understand CABNE=fx quality, CABGE=sd quality, thus cabne is best.

[koosjebram]

Did you guys that are running 1.7 and up even try with stock and then try to ease it up or did you just pop it in and start at 1.6? I'd like to see some scaling exercises done on those. Why don't you go back to 1.4 and try again before you blow it up?

With cabge went to about 2450mhz on stock vcore....
After that it needed upping al the time.
I seemed to be stuck on 2800 now cause my mobo can't give more vcore
(neo platty with 1.36 bios)
Temp barely exceeds 50c on load with thermaltake beetle
(45 with bios 1.8 up)

[ozzimark]

From what I understand CABNE=fx quality, CABGE=sd quality, thus cabne is best.

here's my thought. i don't have any evidence to back this up other than the sightings of some FX CABNE's and the temperature issues..

you know how the turions have "low power" transistors that run cool, but also don't clock quite as well?
instead of putting in more low power transistors, replace them all with "high power" transistors that clock better, but draw much more power. this explains both the scaling, and the temperature issues with the CABNE's.

[187(V)URD@]

It seems to be very weird. More and more people with the 536vpaw reporting high volts and considering the high volts a low temperature.

In 2 weeks another batch of opterons will arrive, we'll seen other results very soon.

[mongoled]

The amnt of voltage tht tht CPU can take is nothing short of amazing, I never thought tht I would see the day were an AMD CPU would return to the high voltage loving tht this particular batch of CPU's seem to like.

While many of us r struggling to keep the temps down, it seems it is an effort to get the temps up with tht particular batch!! Keep us updated with those results, i think I will continue with the burn in, my CPU dosnt seem to want to run 3Ghz prime stable, but thts another story.........

[Deus Falsus]

here's my thought. i don't have any evidence to back this up other than the sightings of some FX CABNE's and the temperature issues..

you know how the turions have "low power" transistors that run cool, but also don't clock quite as well?
instead of putting in more low power transistors, replace them all with "high power" transistors that clock better, but draw much more power. this explains both the scaling, and the temperature issues with the CABNE's.

I see you have 144 cabne but are awaiting 146, I am looking to hit 3ghz prime stable with an opteron. Is the overclockability of the 146 that much higher than the 144? What's your current prime stable max oc? TIA

[ozzimark]

I see you have 144 cabne but are awaiting 146, I am looking to hit 3ghz prime stable with an opteron. Is the overclockability of the 146 that much higher than the 144? What's your current prime stable max oc? TIA

those are questions i would like to know myself
as stated before, my board is out. i never got around to doing any long term stability tests, especially due to my suspicions of having a bad board anyhow. i know i could do stock speeds with 1.025v rock solid

i'll have the answers soon though

[mundi]

I donnt know if s it here write,
bur the most important is TMaxCase
much of CABNE is 65 and you can go to 3GH and over
but a lot of CABGE have 57 and max is 2,7-2,9
better TMC = better OC
and question is if this set will by still same at time

[dogsx2]

I see you have 144 cabne but are awaiting 146, I am looking to hit 3ghz prime stable with an opteron. Is the overclockability of the 146 that much higher than the 144? What's your current prime stable max oc? TIA

I know that this was to ozzimark but I would like to address it as I do have both a 144 & 146 same stepping cabne. Here is my take so far.

1. Both cpu's hit 3000MHz at OCCT stable with ease. The 144@1.4v and the 146@1.45v.

2. With very limited testing on good air the max so far is 3050MHz-144 and 3080MHz-146.

3. I think these are no different then the venice in that the top oc on the very good chips are all about the same max,ie,the x doesn't mean greater oc.

The other thing Ive notice that has nothing to do with these cpu's is that there are sure a lot of dfi mb's that go bad. Mine went bad right after I put the 144 in it and another new dfi board lasted a day and it went bad. Now that I have a bad mb, I notice that there are a lot of them go bad.

While I am expressing my thoughts, here's another. I'm not sure it makes a difference if you burn in at stock MHz@ lowest volts or highest MHz@ lowest volts. I've always used the latter and will do so until I see hard evidence otherwise. I think it's a ymmv type thing. However, I do think it does help to do one or the other.

[computersmsa]

CABNE is more hot at the same Vcore and frenquency that a CABGE ...
It's maybe not only a diff of quality (Silicon) ! I don't know more ... but more me there are a other diff.

[gundamit]

Thanks to Perc for sending me a mounting bracket for my Prommie. My "modded" bracket was having a hard time getting a good seal on my DFI which lead to the death of my DFI SLI mobo. I'm working on modding my non-SLI DFI so I can get some SLI action going. In the meantime I decided to try my EVGA mobo that I got in a deal that you can read about here. If you read that thread you'll understand the limitations of that mobo which makes for a theoretical OC maxium of 3.45ghz when the 148 is used. The 314HTT limit really makes putting the 144 or 146 in this board pointless. Maybe a bios revision will fix this.

Right now I'm working on getting it stable at this speed. I'll check back in when I find the max stable.

[187(V)URD@]

So if I read good, the opteron are limiting DFI's HTT or am i wrong?

[gundamit]

So if I read good, the opteron are limiting DFI's HTT or am i wrong?

If you talking about my post, no. The EVGA has a limited HTT of 314 and memory speed of 250.

[187(V)URD@]

i was , thx for clearing that up

[ozzimark]

I know that this was to ozzimark but I would like to address it as I do have both a 144 & 146 same stepping cabne. Here is my take so far.

1. Both cpu's hit 3000MHz at OCCT stable with ease. The 144@1.4v and the 146@1.45v.

2. With very limited testing on good air the max so far is 3050MHz-144 and 3080MHz-146.

3. I think these are no different then the venice in that the top oc on the very good chips are all about the same max,ie,the x doesn't mean greater oc.

soon i'll be able to comment on the same thing. i'm banking on the 146 to be better than the 144 by a good margin though

I donnt know if s it here write,
bur the most important is TMaxCase
much of CABNE is 65 and you can go to 3GH and over
but a lot of CABGE have 57 and max is 2,7-2,9
better TMC = better OC
and question is if this set will by still same at time

by that theory, my 3700+ newark, which was revE core... should have been simply amazing, being rated to 95c core temps



gun, nice voltage

[Perc]

If you talking about my post, no. The EVGA has a limited HTT of 314 and memory speed of 250.

dude that mb isnt that bad. id like to read more about that mb cause i think my dfi ultra-d is checking out on me. deff like to play with that mb how do you like it...

peace perc,

[dogsx2]

soon i'll be able to comment on the same thing. i'm banking on the 146 to be better than the 144 by a good margin though

I'm talking about air cooling and nothing else. If you look at the air cooled optys, there not that much better then my 144@3050. I hope you get a cracker jack 146 that will do great but still stand by my statement that a really good 144 isn't going to do that much less then a 146. Look at the world records, there is not a lot of difference top to bottom on most.

Also, look at the thread Team Japan has. There is not a nickles worth of difference between the 144 and the 146.

[goodcooper]

so far, is it only the OEM chips that have the CABGE?

computersmas were all OEM iirc...

maybe they havn't gotten to the packaged models yet, maybe if we hurry up and snag some retails we can get the good stepping...

i just hope Vandi hasn't lost my PM :/ i havn't heard from him and its been about 4 or 5 days (can't remember now)

[dogsx2]

Thanks to Perc for sending me a mounting bracket for my Prommie. My "modded" bracket was having a hard time getting a good seal on my DFI which lead to the death of my DFI SLI mobo. I'm working on modding my non-SLI DFI so I can get some SLI action going. In the meantime I decided to try my EVGA mobo that I got in a deal that you can read about here. If you read that thread you'll understand the limitations of that mobo which makes for a theoretical OC maxium of 3.45ghz when the 148 is used. The 314HTT limit really makes putting the 144 or 146 in this board pointless. Maybe a bios revision will fix this.

Right now I'm working on getting it stable at this speed. I'll check back in when I find the max stable.

Those are still good looking clocks for a free mb. Glad to see them doing so well under phase.

[gundamit]

dude that mb isnt that bad. id like to read more about that mb cause i think my dfi ultra-d is checking out on me. deff like to play with that mb how do you like it...

peace perc,

As a bencher ... not so much. But as part of a budget SLI gaming sustem there is a lot to be said for it. Is "budget SLI" an oxymoron? I picked up the hot budget bencher, the Asus 939 with the ULI chipset. That might be a good one to get until the "expert" comes out next month. I'll start testing the Asus as soon as I can.

[gundamit]

Those are still good looking clocks for a free mb. Glad to see them doing so well under phase.

I have to keep reminding myself they were free when I'm wrestling around with one. Its an annoying board sometimes. The funny thing is ... the two 7800GTs are still in their shrinkwrap while I've spent a lot of time with one of the free mobos.

[Deus Falsus]

I know that this was to ozzimark but I would like to address it as I do have both a 144 & 146 same stepping cabne. Here is my take so far.

1. Both cpu's hit 3000MHz at OCCT stable with ease. The 144@1.4v and the 146@1.45v.

2. With very limited testing on good air the max so far is 3050MHz-144 and 3080MHz-146.

3. I think these are no different then the venice in that the top oc on the very good chips are all about the same max,ie,the x doesn't mean greater oc.

The other thing Ive notice that has nothing to do with these cpu's is that there are sure a lot of dfi mb's that go bad. Mine went bad right after I put the 144 in it and another new dfi board lasted a day and it went bad. Now that I have a bad mb, I notice that there are a lot of them go bad.

While I am expressing my thoughts, here's another. I'm not sure it makes a difference if you burn in at stock MHz@ lowest volts or highest MHz@ lowest volts. I've always used the latter and will do so until I see hard evidence otherwise. I think it's a ymmv type thing. However, I do think it does help to do one or the other.

Thanks for the response that's the info what I was looking 4. I cringe at the thought of my DFI dying on me (I'm on RMA rehab right now and happily holding strong... so far), I've heard a lot of sh&&t about DFI's dying for just about anything and was quite content that I never encountered such issues, but this... Is there a reason why this happens or does this happen with a specific bios?

[gundamit]

I thought it was time I had a little fun so I dropped my pair of 6800GTs into the EVGA mobo and did a quick and dirty run on 3DMark '01. Nothing like a big CPU to give that benchmark a bump.

http://service.futuremark.com/compare?2k1=8747506

[sierra_bound]

What's the deal with your CPU-Z screen? 3.3GHz at 0.048v has to be a record. LOL.