horrible overclock, 4400+ not worth money!

[AlienKing]

CPC Auto is the big one I noticed from your screen shot. Wouldn't that default you to 2t with 2x1gig?

[Angry_Games]

Off the top of my head without comparing I didn't think that you normally up the chipset voltage but I could be wrong...

 

On your next post in the OC database with the OCZ the text settings don't match the screenshot for Tref & Bypass Max...

 

-pickles

guess i better go fix that eh =/

[noktekniq]

i don't get what's the difference!

[AndrewZorn]

yeah, is it the 1250 htt?

[Angry_Games]

oook....apparently i uploaded the 3500LL Pro pics twice heh...fixed that

 

the only thing that really stands out is the low memory voltage

yep...I've never had memory that would do this well at 2.60v

 

but

 

that is the whole thing...I'm actually giving it 2.72v

 

I set it to 2.60v in the bios, and after reboot, in the Genie BIOS you see it says '2.66v':

 

 

 

 

when hit with the DMM, it is an actual of 2.72v

 

after much testing (on the 4000EB, you'll see it shows 2.80v, which the bios claims is an actual of 2.88v, but when measured with a DMM, it is a true 2.92v, which is yet again +.04v over what the bios reads, and is a total of +.12v higher than what you manually set the voltage at), I can determine that this board will give +.12v higher than what I set it manually.

 

I've tested across all voltages on 3 different kits of different kind of memory and it is true with all of them...

 

but that leads back a little bit to the original point...less vdimm voltage.

 

Rgone is starting to find that TCCD has a 'sweet' or harmonic spot where it decides to go flying...its sort of like a car...you can shift gears at any time, but if you shift during the engine/transmission's "sweet spot", a certain RPM depending on which gear you are in and going to, your car will surge ahead...TCCD it seems (so far, still testing), as well as other memories, have this "sweet spot".

 

We've spent so much time throwing voltage at things, that I think a lot of us have missed an important issue, which is you have to keep playing and playing and playing and playing and playing with something to find out what it REALLY can do.

 

I can drop 3.1v on my Centon Advance 3200LL TCCD's, and get 275x10 on every single board, but I could never hit 300x9 with any consistency.

 

However, dialing in this vdimm voltage to 2.72v actually shows me better able to hit 300x9 and stable (and of course, fooling a bit with the timings always helps), and I am sure I will end up going higher once I can fine-tune to around 2.74v-2.75v. I'm not going to go into how to get that all accomplished, as it involves modding the board usually, or really going after it by manipulating the 3.3v rail on my OCZ power supplies.

 

 

Another side-effect of this, is that you can, after hitting this sweet spot with some memory at lower voltages, still have errors and problems...but that is usually because now you have too much voltage to the cpu. Rgone and I are finding that dropping the vdimm to a 'sweet spot' means that you no longer have to pound your cpu with voltage (keep in mind that an integrated memory controller is on the cpu itself, and memory voltage correlates with this, so having 3.3v on the vdimm, is going to directly effect the voltage/resistance of the integrated memory controller).

 

Now that last little part might not make a lot of sense, but it will (you can probably google all about this already).

 

That part of the Overclocking Guide where I really explain exactly what HTT is and how you calculate it, and made a good number of you understand exactly what it was all about when fooling with Athlon64/Sempron cpu's with integrated memory controllers....this is sort of like that...and once I can get my mind around it, really grasp it, then I will come back and try to explain in terms that everyone can really get their own mind around and understand it to the point where it makes sense when they are practicing it.

 

whew lol

 

 

CPC Auto is the big one I noticed from your screen shot. Wouldn't that default you to 2t with 2x1gig?

 

no, not unless it was sub-par memory, or possibly a Winchester/Clawhammer core. But I choose to leave it on Auto so it can decide what it wants.

 

 

Off the top of my head without comparing I didn't think that you normally up the chipset voltage but I could be wrong...

 

i NEVER up the chipset voltage on 2x512 or 2x256, and I think rarely up it with 2x1024, and I don't even up it with 4x512. I have been in the habit of 2x1024 @ 1.3v LDT and 1.6v Chipset which is different than my standard 1.2v/1.5v for anything but 2x1024. I honestly can't say it makes the difference...I should try @ 1.2v/1.5v just to put my money where my mouth is right?

 

 

LDT/FSB Frequency Ratio? Can it be? You are running a 1250 HTT! is that the key?

 

no, I always leave this at Auto. Oskar Wu, who designed the board and bios, told us when the board came out that LDT multi @ Auto will auto-adjust. If you look at every single NF4 database entry I have on the DFI Lanparty boards, you will see me leaving this setting at Auto.

 

what's the date code on the chip?

whenever i see anyone post a date code for a 4400, mine is usually earlier

 

on the top of my chip it says ACBWE 0517 TPMW

 

if i remember, it is 0509 or so...we (mfg's) got the first runs of X2 cpu's. Before anyone else got them. There's probably a guy or two from Asus, and MSI, etc that have X2's same as ours or even earlier. You'd be surprised how early we get some cpu's (and we can't talk about them, or even mention them in passing, but we usually have them months before reviewers have them)

 

I have to say these early cpu's are rarely as good as the later ones you guys actually purchase. My Winchester 3200+ was a pile of dog food, and my FX-55 San Diego is great at 2600-2700Mhz...but useless beyond that (thankfully its a 2600Mhz stock chip lol).

 

the 3200+ Venice I bought is supa fantastical, as are most 3200+ and 3000+ Venice cpu's that you guys send me with your boards/memory to troubleshoot, yet the ones AMD sent us were early originals and just aren't that good for overclocking.

 

I figured the X2 was going to be the same...pretty poor...and it isn't in the sense that it will easily do 2650Mhz on every kit of memory I have...but it would never budge past that (sorta like my 754 Newcastle 3400+ that will not even clock from stock of 2400Mhz to 2500Mhz...it just wont).

 

Now I think I am beginning to understand that as good as I thought maybe I was getting, I am back to being nothing again as I am finding little tricks and Rgone is suggesting little tricks that have been slowly bringing me back up past my previous bests on more than a few cpu's...

 

 

 

I guess my whole point is...don't ever assume that you have mastered your board/cpu/memory...as sometimes a little surprise blindsides you that you would have never thought to try....but once you actually try messing around with it, you start getting better results than ever.

 

 

and of course, keep in mind, that everyone's results are different, and will vary. I probably am just a chump clocker who isn't as good as I think I am, and found out that its true...I'm a chump lol. You guys are probably much better at this than I am, and so maybe you HAVE found your top limits for everything, and maybe you've already tried all of this 'lowering voltage' stuff and are laughing at me for being such a dirty noob haha...I dunno...it has renewed my interest in going back through some hardware to re-test once again to see if all these new little tricks and tips correlate across all of my hardware...

 

probably some will be better, and probably some will be worse (i cannot imagine giving less voltage to BH-5 type memory lol...but slightly looser timings maybe?)

 

anyway, I'll try to answer a bit from here, as I know you guys will have all kinds of questions...but I cannot promise I can give an answer that makes much sense until I really wrap my brain around some of the things I just learned, found out, discovered, and now am testing

 

hopefully it renews YOUR interest to fool around a bit more and see if you can see anything yourselves (which would help validate if my little discoveries and tricks from rgone work for others besides just us)

[AndrewZorn]

so... any suggestions for this VX? you are still running your FSB at 250, which your memory is rated to do anyway, right? i dont understand what has changed, how you are getting to 2.75 on the CPU...

[noktekniq]

those just blew over my head... so i dunno what vdimm i should run my tccd at.

[GunBLADE]

i dont understand mine reads 2.6v and do 270, and still cant get the cpu runs without touching the high voltages yeah he do 2.7ghz but with 1.6v on him i dont want that XD.

 

i think the problem resides on the mobo than the cpu..

more a bios problem maybe.

[Guest nomad]

I just threw a DMM on my system and I had 2.8Vdimm set in the bios and was reading 2.91v. Put it down to 2.7 and +0.3 if not 3.1, and getting 2.82v with the DMM.

 

Also had problems with the orange slots and my TCCD. Put them in the yellow and BAM. Everything is perfect.

 

Now that I have everything set the way I want and and looking super stable (even BF2) I will be putting my entry in the OC database after the weekend.

 

*EDIT : Tried to get 2.72 with the 3.3 rail pot on my OCZ and through the bios and got a hard lock on my system. Looks like I need 2.8+ to remain stable. Also I thought I was stable on my ram with a 12 hour pass of SP2004 and then ran MemTest MFC in windows and had almost instant errors. Beginning to see the merrits of that program. MemTest 86 doesn't tell you if you are stable in wondows, that is for sure!

[Pyr0]

memtest mfc?

i hadn't heard of that one before

where did you get it?

[JiminyC]

the one he is referring to is like http://hcidesign.com/memtest/ . this is HCI's and recommended to be by a Mushkin mate. The algo's are newer and way more sensitive than memtest in dos and Prime95. memtest on boot will not show winblowz errors either, bad timing will corrupt a system to the point of reinstall of os. The one by hci is very good and can run in the background, actually makes it more sensitive, but 8 hours , 0 errors, your'e good to go. I am gonna pass RGones specs to the Mushkin mate Zebbo and see if this rig will run those timings, maybe a tad diff, but what RGone says makes sense to me. I have been able to get better cpu speed with lower cpuvid [email protected] vdimm but much looser timings. Maybe it's a tweak I need to do, but RGones *essay* :nod: goes to show you that we all have a tendency to throw a boatload of voltage for stability and there quite possibly is another way to look at it. BTW, why does everyone set LTD voltage to 1.3? No1 has been able to effectively answer that.

 

Thanks

 

*found out hard way..2 blown fan headers, save power for board put all fans on 4 pin*

[dfx]

New X2 user here

Just got my X2 4400+ yesterday :nod:

How's my chip fair up :confused:

 

@ Default Vcore + Stock Cool (Sorry for the giantic pix)