DFI Infinity/Lanparty B vcore issue?

[Iolao]

Try this simple test:

with default cpu voltage run memtest86 at maximum rock solid fsb (use cas2 2-2-2-11 --> tighest timings) with max vdimm (3.3v), for example 250x9. Once you reach maximum stable fsb, try again but this time with cpu voltage at 2v. Do you get tons of errors?

Please specify:
1) Ram used (brand and ram chips)
2) Board production place (taiwan or china) and possibly northbridge production date
3) Eventually custom cooling applied to the mobo (NB, SB, mosfets and pll for example)
4) Bios used
5) Psu used
This is mine:

1) 1x256MB twinmos bh5 + 1x256MB OCZ PC3500 bh5
2) Made in Taiwan (don't know nb production date)
3) Heatsinks on mosfets, nb liwuid cooled, sb with stock nb heatsink
4) Beta 31/1
5) Antec PX412PP with all rails trimmered to the max

...and I get tons of errors this way; I cannot run high fsb and high vcore together.

[Dojo | Warlord]

I got 240fsb with 2.0 vcore fine. No errors.

1) - 2x 256MB Mushkin PC3500 Level II Black, BH-5
2) - Made in Taiwan, unknown production date
3) - No heatsinks on mosfets (yet), Swiftech MCX159-A on NB, stock active Leadtek nForce2 Ultra400 Deluxe Limited NB cooler on SB
4) - Beta 1/21 BIOS
5) - Antec TruePower 550-watt; stock rails - 4.99(5v), 3.29(3.3v), 12.08(12v)

[Evil_Spork]

i have similar setups(dont have time to type it out)

at 2.0vcore in bios i get about 1.95ish real world with vdimm mod and all voltages maxed.

[SAE]

Originally posted by Evil_Spork
i have similar setups(dont have time to type it out)

at 2.0vcore in bios i get about 1.95ish real world with vdimm mod and all voltages maxed.

Hey Evil.

That's not the point of what Marco is trying to say. He has issues with high vcore and high fsb getting several mem errors...

And I think it's my problem too... cannot even get a stable 250

P.S. Marco, I ordered another khx3200 from the shop I told u... the older one I have cannot cope with the other new one. It's on ebay now

[Evil_Spork]

ahh.... sorry i didnt read teh whole post.. =-\

i seem to actually get some instability with voltages maxed.. ive lost usb/lan/sound at 250fsb.... normally 260 only did that after a while. 255 used to be totally stable.. i dunno what the problem is... and now this morning it wouldnt boot up at all lol.. hit power.. and fans spin but nothing happens. now i got to to say "missing os".. not a big help.. buuuut im dumping this board. trying my old one and probably buying a new one.

[SAE]

Oh. That's bad.

Some board's quality seems to suck... cause I already had 2 bad boards and one good (the 1st that died)...

Mine was pretty stable at 255 too and then did not even complete the 1st test of 3dm the next day

Furthermore there are only very few people with very high cpu and fsb clocks... so maybe all the boards aren't that good. Who knows?!

[Iolao]

Originally posted by SAE
Hey Evil.

That's not the point of what Marco is trying to say. He has issues with high vcore and high fsb getting several mem errors...

And I think it's my problem too... cannot even get a stable 250

P.S. Marco, I ordered another khx3200 from the shop I told u... the older one I have cannot cope with the other new one. It's on ebay now

Thank for explanation I'm in the mood of click "confirm order" but I might wait first a new board to try to sort it out this issue.

Evil_Spork: I'll try to be clearer in what I say:

let's suppose your ram maxes out at 250 with tightest timings and 3.3v; this is memtest stable for hours and obviously 3dmark stable too. If you test at a low vcore set (let's say 1.65v) for example 250x9=2250Mhz (which most cpus are able to) everything is fine. Try now to up the vcore at 2v (or more if you voltmodded) while staying with the same settings (250x9 tighest timings and vdimm set to 3.3v).... are you always error free?
In my case answer is no, memtest gives me a lot of errors and 3dmark behaves even worse....

My theory is the following:
Board lacks of caps or has bad voltage regulation. We all know that vcore is up to +5v line, new thing on this board is that also vdimm is depending on that rails, so combining high vcore with high vdimm may result in those behaviours.... If the board was a good quality one, you could try to trimmer up your +5v line from psu as I did; unluckily it seems there's a HUGE DIFFERENCE between psu's molex readings with DMM and mobo's mosfet readings (in my case about 0.3v which is A LOT), moreover rail tends to be very variable under load (while the reading from psu's molex confirms that my antec has a steady +5v line). This whole story sounds to me like caps missing or bad voltage regulation. Playing with DMM all over the mobo gave me evidence that dfi infinity/lanpartyB series uses tiny pathetic caps and above all they seem just a few compared to what you can have on an abit board (that's another POS but it's a totally different story)... and I'm missing comments on the little mosfet version used here rather than the larger ones as I cannot be sure if there's any difference. All this comes more realistic when I see only "shots" with dfi, all people tend to give great importance to a shot at 260 (for example) while having ridicolous cpu frequencies. Apart from sae's and lastviking's really amazing 3dmark scores, I see NOONE is able to push both the limits of the cpu and ram, that's the reason I started this thread

[SAE]

Well done

I maybe try out my 1700+ (unlocked ) to sort out the problem too... I think if I could stay below 1,9V VCore ther should be no problem with stability... maybe

But why do I have a prom??? To have 2500MHz clocks??? That's a waste of money and resources (cooling)... Too bad. Do not really want to change/add caps

[Evil_Spork]

i see what you mean. thats does seem to hold true. i think my current board is one of the low quality ones. as my last that i almost killed seemed to hold voltages and everything much much better.

im selling this one and testing old one, then maybe getting another. ill report if i can boot my comp this afternoon. i couldnt this morning.. really no idea why, it worked perfect last night, today, no post.. this mobo is getting moody.

though i beleave i have seen similar results on my current crappy board. with all voltages maxed i get soem heavy vcore dip, and vdd is pretty low. and there is some instability, though i cant say 100% sure if ti was caused by vcore being high or high clocks. sry i cant be of a whole lot of help.

[Iolao]

no support on amdmb/dfi forum... tell them you have a problem and they think you are already ok for what you have tell them you might have experienced a sort of issue and they say what they read on the book: check psu, be sure to load optimized default, ecc. They think it's fine having ram set to cas 2-3-3-11 @ 240, so problem is not present... bah

[SAE]

Originally posted by Iolao
no support on amdmb/dfi forum... tell them you have a problem and they think you are already ok for what you have tell them you might have experienced a sort of issue and they say what they read on the book: check psu, be sure to load optimized default, ecc. They think it's fine having ram set to cas 2-3-3-11 @ 240, so problem is not present... bah

Yeah. They do not really do overclocking in our "Xtreme" way. That's why THESE forums are called XTREME, aren't they?

And/or they don't take our needs/probs serious enough

[Lastviking]

But if i try higher vdimm it´s jumping to much...

I have tryied to cap modd this board 2 vecks ago(8 caps). But then the board didn´t post...so i removed them i don´t know why the board didn´t post maybe i try it another time(need to buy new caps..).

To get so high like this i cold down the board with -11c air . But i don´t get any higher fbs with lower vcore.. The vdimm is jumping also with low vcore.(vdimm over 3.6->)

Edit:

More things...higest in Sisoft memmory is 278fsb with -20 out temp. and thats only 1 mhz more than -11 out temp. So the temp is not the problem....And highest windows screen fsb is 283.6fsb with one 512 stick @ 11-4-4-4-2-5..it cane be high vdd and vdimm = jumping to much = some caps have big problem..

Sorry for my bad english.....

[Iolao]

Originally posted by Lastviking


But if i try higher vdimm it´s jumping to much...

I have tryied to cap modd this board 2 vecks ago(8 caps). But then the board didn´t post...so i removed them i don´t know why the board didn´t post maybe i try it another time(need to buy new caps..).

To get so high like this i cold down the board with -11c air . But i don´t get any higher fbs with lower vcore.. The vdimm is jumping also with low vcore.(vdimm over 3.6->)

and what running torture tests? How does your board reacts? Remember we're talking about stability with high vcore.

[Lastviking]

Originally posted by Iolao
and what running torture tests? How does your board reacts? Remember we're talking about stability with high vcore.

Yeah i know it´s torture test and more but i only helping with some info...to find out the problem.. And i don´t want to test torture test..i don´t use 2.0v in 24/7 use i need beter coling then.. and i don´t want to have room temp like 5c ...to test long torture tests..

[Evil_Spork]

how in the world did you get 2750mhz on air cooling?! thats just amazing!!!!

[Lastviking]

Originally posted by Evil_Spork
how in the world did you get 2750mhz on air cooling?! thats just amazing!!!!

Thx..But i think i get higher with the vcore modd. i´m only using 2.0v @2767Mhz and the out temp was only -11 some days it´s -20-> ))

[SAE]

Sweden is not very far away from Germany but by far colder

So if you put ya rig outside you might be able to beat a prom And maybe stability is better when the mobo is in ice

[Evil_Spork]

ahhhh that was done with supercool air.. gotcha.. i thought that was regular air!!!

i was gunna have to hurt someone severly if that was on regular room temp air.

[sjohnson]

In case you didn't see my post, I killed my original Infinity. When my new one gets here, I'll try the 2.0 volt experiment and report.

But, anecdotally, my original board had what I'd consider a lot of voltage fluctuation at high Vdimm (3.3v BIOS) and Vcore (1.9v BIOS).

And, as I recall, I could pass memtest86 up to 246 FSB but failed at 247. I always thought it was a simple need for more Vdimm but again, I'll test out my new board and report.

[Iolao]

/OT

Stefan: I found a german shop nearly identical to ebug but quite cheaper. It's called norkit, is it reliable?

/OT

I would add that when using my ati 9700 pro +3.3v rail moves a lot, while with an old nvidia geffo2mx everything is steady like a rock. Anyone can report some experience on this?

[Iolao]

Originally posted by Lastviking
Thx..But i think i get higher with the vcore modd. i´m only using 2.0v @2767Mhz and the out temp was only -11 some days it´s -20-> ))

Say hello to the penguins while benching! Holy §§§§, -20C! You are either mad either a cyborg!

[Shep]

I have some vcore issues with my Infinity. I can't set the vcore higher than 1,95, which gives about 1,90ish measured with software. Haven't poked around with MM yet.

If i set it higher i just get random reboots, looks just like the OVP issue i had with my 8rda+ board. So i hope the problems will dissapear when I do the VCore and Vsense mods sometime. But computers ain't first priority these days since i have just bought a house, so it can take some time

Btw: When I was in the armed service, it was really freakin cold. We had to sleep in tents when it was -40 C outside. It really does get cold in the northernmost part of skandinavia

[Redskin]

Originally posted by Iolao
[B]/OT

Stefan: I found a german shop nearly identical to ebug but quite cheaper. It's called norkit, is it reliable?

I think ebug and norsk-it work together. norsk-it is the cheaper shop, but beware some customers complain about delivery and service of this shop (in exceptional case) , especially with advanced payment.
I' ve never place an order there. But ebug deliver very fast.

Stefan whats your advice ?

[SAE]

I don't think norsk-it is as reliable as e-bug cause if you search through preistrend.de (like pricewatch) there are shown some shop grades and norsk do not convice there

RMAing should be a farce there...

I still prefer a good service, fast rmas and a really big stock...

[rookiekiller]

256 x 9.5 with 2v cpu 3.3v ram 1.9v chip

Barton 2500 xp wit watercooling

1) Kingston KHX3200 BH5 1from china 1from usa
2) Lanparty ULTRA B,1/2 made in china and 1/2 made in taiwan =o)
3) non mod on nb sb ...
4) 1-31
5) antec 550 tc


i think it is primestable, now it is over 4 hour prime... and its still run wit prime