The Official Opteron 939 Overclocking Thread

[dinos22]

so you can't run those sticks at 2.5-3-3-6 1T or......

that OC (3140MHz) with really optimised settings should land you low to mid 23m times

[gundamit]

so you can't run those sticks at 2.5-3-3-6 1T or......

I lose TRCD 3 somewhere in the 270s. This is not super TCCD, its merely good.

gundamit which bios are you using?

Its one of the 702s, can't remember which. Lol.

[Vapor]

Hmmm, I'm using 704-2BTA (supposedly doesn't have the TCCD problem, w/e), oh how I want to retry my luck with my 146 but I really need to do work

[gundamit]

Gun> I might catch ya TCCD with 2gig's

I'm anxious to get may Patriots back in there but I need to figure out whats up with my TCCD first.

You've already posted a sub 24 minute 32m run. Makes me want those Crucials so bad!

[Vapor]

MM, can you give us a Tweaker shot?

I think my Ballistix aren't as bad as I once thought.

[gundamit]

check the timings out on the SS i jsut posted tho LowLatency at ddr500

That is just sick. Didn't know Crucial could do that as well.

[Ugly n Grey]

Great thread you guys, almost 12,000 views on this thread alone at the time of this post. I can see these easily being the chip of coice for AMD guys on the board here.

While I rarely take the opportunity, I'm going to say "I told you so" to the guys who claimed I was off my rocker when I stated the AMD and IBM sales droids assured me Opty's were tested and binned differently. I see you here in this thread jumping on the bandwagon

I feel so vindicated I could weep....but I'd rather have a couple new optys and a hot nurse to make up for it.

Great stuff, I'm going to try the dualies as soon as I can get my hands on a released one.

[jahjahbinks]

You are the one who ordered dual core Opteron?

[Ugly n Grey]

I already have dual core Opty's but they are all 940 pin chips, I mean I want the 939's

[jahjahbinks]

Oh, because I read in a post a couple of days ago that someone ordered a 939 pin dual core opteron, so unless I read it wrong, it's 939 not 940.

[dinos22]

165/170/175/180 are all Socket939 Dual Cores

[Vapor]

I didn't know there was a 180 available yet?

[dinos22]

I didn't know there was a 180 available yet?

announced yes.....available NOT sure

one of our retailers has 3x175 CPUs left in stock...damn i wish i was a loaded damn it

[mongoled]

Per your suggestions I've changed the timings and dropped some time from the 32M run at 3.14ghz. Thanks.

Hey gundamit,

I took a look at your tccd timings and their is something I would like to ask you. Why do you use these values:

Max Async: 10ns
Read Preamble: 8ns

I have two sticks of Gskill tccd, LC's (the poor ones ) and they do 2.5-4-3-7 up to 305mhz stable. I currently have 'Max Async' set to 7ns and 'Read Preamble' is set at 5ns. Their are a few other different settings which im going to change to yours to see if the mem will still run at 305mhz rock stable.

I report my results in a bit

[Vapor]

I need a really high preamble to even get the Opty above 2.8GHz. Max async was also higher than what I used in my 4400+ but I needed it to get stable above 3.1GHz.

[mongoled]

I need a really high preamble to even get the Opty above 2.8GHz. Max async was also higher than what I used in my 4400+ but I needed it to get stable above 3.1GHz.

Yes, ive suspected for a while tht these settings need tweaking according to CPU type. At the moment im using my old Winny, it seems to need different settings to the Venice I had. The Venice was alot more particular to these settings!!

Ofcourse I will be doing alot of testing once my 146 gets here and these settings will get alot of my attention.

Anyhow ive just tested with gundamit settings, I made changes in my bios for trc/trfc/trrd/twr/by pass max/idle cycle. Tightening Trc and Trfc has given me a nice increase in bandwidth in memtest (havnt tested elsewhere) if i were to try and get this same increase by using a Trcd value of 3 (I have 4 set in bios) I would get alot of errors in memtest.

With the tighter timings 305mhz gives me single bit errors, 300mhz is fine, im going to try 'burning in' at 300mhz with the tighter timings, then move the mhz up later on. Im very happy with my LC's, i believe the constant tweaking has helped break these in nicely.

Gundamit,

Can you tell me why you have 'By Pass Max' set to 7x? Ive never played with this setting, nor seen anybody else talk abt this setting, so im curious to know wht your experiences are with this.

[Vapor]

7x Bypass Max speeds things up in 32M and according at least one guy on Team Japan, adds stability.

[afireinside]

I have no clue what bypass I was running at 3527... I DID need higher async/preable to break 3050 though.

[dinos22]

a lot of you have been wondering what motherboards these Opteron CPUs are actually made for..............well here it is

http://www.tyan.com/products/html/tomcatk8e.html

http://www2.amd.com/us-en/recmobo/De...queryID=100327

all this trouble over one freaking motherboard......might as well let all the other manufacturers support these.......

[Zeus]

W000T!!!
Managed to track one down-- with the magical stepping!!



Gonna torture this baby tonight!!!

[Vapor]

W000T!!!
Managed to track one down-- with the magical stepping!!



Gonna torture this baby tonight!!!

Good job....have fun with it, it's a good one (the production number is VERY close to mine, too)

[gundamit]

Well I finally gave up on my TCCD at 314x10 9:10 divider. My best run was at 24:06.922. Oh well.

I took a look at your tccd timings and their is something I would like to ask you. Why do you use these values:

Max Async: 10ns
Read Preamble: 8ns

No reason in particular. I'm usually just thrashing about or going with auto settings. I did drop the Read Preamble to 7ns on my final TCCD run which was 2 seconds faster than my last posted results, so dropping it more may help.

I put in my BH-5 and tried to find my max clock. I think its 3.15ghz but unfortunately I have to use the 7:10 divider and that runs my BH-5 a lot lower than their top speed.



My BH-5 won't go to> 258 so i had to drop the speed to 3.1ghz in order to go with the 3:4 divider. First run came up at 24:01.719. Tried defragging the drives (couldn't hurt) moved the SuperPi program off a partition to the small root drive, turned off DLL in Tweaker and moved DQS Skew Mode from Slow to fast in Tweaker. Took another run ...



and i finally get to under 24 minutes in SuperPi 32M.

Later on this week I'll get my Patriots back in and do some more testing. Perc has kindly offered to send me the bracket I need to get my Prommie going. Hopefully next weekend I'll be sub-zero. Anyone care to guess what I should expect to see with mild sub zero cooling?

Finally remembered to get A64MaxTCase.



Dang. I was hoping for 67c rating!

[Zeus]

Please allow me to brag a little...

How's 3GHz stock voltage watercooled sound?



Didn't do much else tbh, but still nice.

Have done 32M@ 2850 stock voltage, no problem but the time was very bad, i think windows is borked.

Will try some more, haven't been passed stock Vcore yet, all seems to work flawless with my Gskill TCCD although ClockGen won't work proper if i try to raise HTT more than 10MHz so i have to overclock oldschool in Bios.

[Vapor]

About CG, I've learned to never make adjustments more than 2-3MHz HTT at a time...it sucks but it's what's necessary I've learned.

Good looking OC there, bud.

[gundamit]

Please allow me to brag a little...

How's 3GHz stock voltage watercooled sound?

Sounds good! How high do you have to raise the volts to get it 32M stable at that speed? Post #422 has a guy at 3ghz 32M stable at 1.39v.

Seriously though, please post tweaker settings so memory noobs like myself can be brought up to speed. Good luck. :thumbsup: