With multi 10 ~243fsb..
With multi 12 <240fsb..
:confused:
Printable View
With multi 10 ~243fsb..
With multi 12 <240fsb..
:confused:
Multi 12 might have more restricting ROMsip entries... there is some work being done on ROMsips, led by tictac, so maybe your issue will be semi-solved soon...Quote:
Originally posted by Rub87
With multi 10 ~243fsb..
With multi 12 <240fsb..
:confused:
How high can you get stable in Prime95?
________
Honda NPS50
225*10 is primestable..
230*10 results in lockup after a few minutes..
For comparison sake, have you tried out the D10 bios to see how high it gets you?
Mine doesn't, a vdd between 1.6 or 2.4V makes no difference.Quote:
Originally posted by kchip
Anyone else need 1.95 VDD to get their there Rev2 C1 to 230 FSB? Mine seems to like voltage...: (
I've tried:Quote:
Originally posted by uwackme
For comparison sake, have you tried out the D10 bios to see how high it gets you?
D10
D22
D23 subzero
With the D10 bios my cpu was recognized, so I could choose: 2500+, 3200+ etc.. by the others I could only choose: user define
I think the D23 subzero is the most stable, but I think 240 fsb is just a bit to low :/
Heads up... Ticcy fixed the CPC issue on the NF7-S bios. Search around for his fixed D24.
And then.... life was good
I'll give it a try:)
But could it just be the fact my stick is 512Mb that my max fsb 240 is?
Haldol gets up to 265 with 265Mb sticks..:confused:
I've tried the D24 .. it's less stable then the D23 subzero
D24--> 240*10 alot of errors in test 5
D23-->240*10 4errors in test 6 @: 0.2Mb, 493.8Mb and 2*256.6Mb
I got back my old corsair pc3200 xms v2.2 512Mb with CH-6 chips...
First on this board without mods it even didn't ran @ 220 fsb stable, now, 235*8 is almost memtest stable (11-2-2 2) exept 2 small error in test 5, always between 0.1mb and 0.9 mb, so i think one chip is overheating, I only don't know wich one..?
I put it in slot 2, the Twinmos BH-5 in slot 3 and now they're running @ 230*12.5, memtest stable...
I don't get in windows @230*12.5, I need to drop the multi to 12 cause I let just a bit too much gas out of my prommy (maybe disabling acpi would help):(
this gives me the following disappointing result:
jus curious whur people read the vcore from.
the unused mosfet pad by the cpu socket seems to give me a value that is bang the value I set in BIOS when at 1.65v... and within 0.2v at higher voltages under load.
If this is VCore it is a MUCH easier and safer read point than the MOSFET legs.. no risk of shorting out the pins etc.
The exact place im refering to is the largest of the 3pads... check it out ;)
________
ZXR250
Hmmz:(
230 dcddr doesn't seem to be stable afterall...
But i discovered something that may be the cause:
With the 2 512Mb dimms installed I measure unter load a vtt/vbt:vref = 1.6v while 3.3v/vdimm= 3.42v...
Could it be that the 2 512Mb dimms need to much power so the vref mosfet can't enough?
The problem is that the vref regulates the voltage from what the vdimm value in the bios is, with the 3.3v=vdimm enabled with 1 dimm vdimm is almost the same as 3.3v in bios, but with 2 dimms installed vdimm is only ~3.18v while 3.3v=~3.42v...
I don't know how this can because I didn't change anything to the board and the +3.3v is still directly connected to the vdimm...
Great read guys!
followed this thread and did the vtt mod to my board + the vdimm mod. (thanks to malves as well).
but i'm having some weird issues.
2.9 bios = 3.4 with vtt at 1.7
2.7 bios = 3.1 with vtt at 1.6
2.6 bios = 3.0 with vtt at 1.55
the vtt is following like it should (.05 down for each .1 in the bios) but the vdimm isn't going down appropriately...
any ideas why?
thanks!
Damn, a fight with FSB 250, especcialy with 10x250, is
really tiring..
As so far, I had broken 10x240 1.5h stable in P95 and
I'm still fighting, but I got problems with 250
I've tested a lot bios,D10,D22,D23 and D24, generally
tictac's
I'm not sure which bios is for me the best, I didn't tested
them exactly and I forgot what a version I had ;)
IMO D24 series is very good, I can use high vcores (1.9V
and higher) and FSB240 without big problems
Actually I'm rebuilding watercooler and working on
block's mouting, maybe I will finish it tomorrow, also
I must improve cooling mobo (NB and fans on mosfet's
radiators)
And my rig - Thorton 2000+, 2x256 one sided hynix D43, L12 modded NF-7S
okay... i can't seem to run higher than what i could pre-vtt mod...
how important is it to have the vtt @ 1/2 the vdimm?
or should i try another bios?
i've tried d10 and d23 fire rev 1 and no luck :(
If vtt isn't 1/2 of vdimm and if if much less or more your board won't even be stable @ stock speeds...
Rub87... volt mod your PSU 3.3v rail.. give it 3.6V
Is 3.6v healthy for your dimms?:p:
I'll give it a try..
not for daily use but for benching it is safe till 3.7- 3.8, just make sure you have a fan blowing at them
Bh-5 can take 3.6v daily, just make sure you have some air going across them. 3.6v is where I stop for daily use, but for benching I have gone as high as 3.9v and still gained fsb.
Now I'm running 512 CH-6 and 512 BH-5 in dcddr for daily use @ 3.3v..
@ 230fsb 11-2-2 2 , some weird thing: @ stock voltage without mod even 200 fsb 11-2-2 2 isn't stable..:rolleyes: :D
that CH6 is really holding you back I think
Nope, it does 238 fsb 11-2-2 2 in dual channel with another stick of 512mb bh-5 , when I replace that stick with a single sided bh-5 256Mb I don't get a mhz higher, the fukkn mobo is holding me back, even with only 1 stick of 256 mb bh-5 single sided 245 fsb 11-2-2 2 isn't stable... I tried every bios i think..
I suppose 1 stick bh-5 don't need more then 3.4v for doing 245fsb single channel..? does it?
so replacing the CH6 with BH5 does gain you 8MHz FSB? Seems like the CH6 is holding you back...Quote:
Originally posted by Rub87
Nope, it does 238 fsb 11-2-2 2 in dual channel with another stick of 512mb bh-5
3.4v should get BH5 to and beyond 250FSB for sure... with 1x512MB stick I am stable at 250 2-2-2-11 @ 3.2v and can bench at 255 2-2-2-11. But with 2x512 I am only 100% stable at 232 2-2-2-11... although I can run 3DMark '01 and '03 up to 242 2-2-2-11. All this is with CPC ON and BIOS D10.... the best D24 version limited me to 229FSB for Prime & 3DMark.Quote:
...when I replace that stick with a single sided bh-5 256Mb I don't get a mhz higher, the fukkn mobo is holding me back, even with only 1 stick of 256 mb bh-5 single sided 245 fsb 11-2-2 2 isn't stable... I tried every bios i think..
I suppose 1 stick bh-5 don't need more then 3.4v for doing 245fsb single channel..? does it?
Try running a single BH5 stick in Dimm3 with 3.4v and see how high you get. Disable "APIC" and "Enhance PCI Performance" too ;)
Good luck dude
Non U'r getting it wrong, with:
1 stick 512mb XMS pc3200 rev 2.2 with CH-6
1 stick 512mc Twinmos BH-5
together in DCDDR I can run super pi up to 238 fsb 11-2-2 2
But when I replace the stick XMS with a small 256mb twinmos bh-5 stick so I'm running
1 stick 256mb Twinmos BH-5
1 stick 512mb Twinmos BH-5
In DCDDR I don't gain a single mhz:confused: so I tihnk the CH-6 is capable of more, when I only use the small 256 mb BH-5 stick even 245Mhz isn't memtest stable:help:
also see: http://www.xtremesystems.org/forums/...318#post481318
Have you tried the CH6 stick on its own?
Test all sticks individually and see which one performs the best... from my experience, putting the best performing stick in Dimm3 and the other in Dimm2 will gain some FSB... but this was on a 2x512MB setup.
Also, try BIOS D10 if you'r not using SATA... like I said above, it helped me a bit... I'm 232FSB solid with D10 and 2-2-2-11 timings.
Yeah that was for me also the best combo, the 512mb BH-5 stick in dimm 3 and the XMS CH-6 stick in dimm 2...
give BIOS D10 a try
I did, 245*8 with 1 single one sided stick bh-5 256mb gave me 2 errors in test 5 .. pff, even with 3.4 vdimm...
did you use Dimm3, with D10 it seems like Dimm 3 performs worse than Dimm1 and Dimm2... Ive heard of a few people that can get to 250 in Dimm1 or Dimm2 with a single stick in D10, but Dimm3 stops at 245... jus a suggestion
I was using dimm 1...
apic disabled?
Yup.. I couldn't test in windows that way, but in memtest86 it wasn't stable with apic disabled.. I also tested with APIC enabled but that made no difference...
Is it possible to use vdim mod with grabebr?? I don't want to ruin my waranty
Yes sir it is.Quote:
Originally posted by SwimmerBoy
Is it possible to use vdim mod with grabebr?? I don't want to ruin my waranty
You need a very small grabber tho.
You wont to connect it to pin #19 on W83301R.
Can anyone show me pics and how to do the vdimmod by grabber??
(NF7-S rev 2.0)
I know I need to read over this lengthy thread again, but has anyone gotten any significant gains from this mod?
It seems to me that it is cauing more headaches than anything. If the Vtt isnt kept at exacttly 1/2 of Vdimm after the mod, the board is all kinds of unstable.
BTW, where you most of you checking the Vtt? At the Vtt Vreg under the dimms, or between dimm2 and dimm3 with the resistor divider?
Please ppl... I need more voltage in my mems...can anyone exlpin me how to do vdim mod to NF7 by grabber???
Here is a pic to look over. The resistor that is typically soldered to is connected to a pin on the Vreg right under it. That pin is #19.Quote:
Originally posted by SwimmerBoy
Please ppl... I need more voltage in my mems...can anyone exlpin me how to do vdim mod to NF7 by grabber???
And what about the rest!??! I'm 2 noob to understand just like that :(
snap away that pin...... as easy as that.... :)
I dont know the rest of the conectons :(
I'm being introduced to vmods :D
I'll need the full skemeto work it out...
But to @ll, thanks for your help ;)
that pin is Vsense for vdimm
OS what I have to do is to join the 1kohm POT to the grabber and sticky to the 2nd pin of that chip?
I really apreciated f you could explain me how to do all the connections :)
solder a pot to that pin... gorund the other end...
set resistance to max.. lower the resistance to get higher vdimm
Where do I get the power to increse the voltage??
@tictac... got pm ;)
when u lower the resistance most of the voltage will go to ground ... Winbond ic will get lower voltage than he expected so he trigger up Vdimm mosfet gate voltage
higher vdimm u get... simple as that
Waht about pics to see it better?? Does someone has one?
:toast:
Thanks tictac for the help
here clearer pic
Memory voltage at Bios always at 2,6v right?
[[]]
no... u still can set ur voltage from bios
but when u select 2.6V it will give much higher voltage
u can monitor the voltage from pc health status menu
okidoki.... I'll try it and reply here then...
Thanks
[[]]
set the 5k pot to max.. lower the resistance for higher vdimm voltage.. cheers
Question:
The last 2 NF7 boards I have done the Vtt mod on are acting weird as hell.
They will only boot with a stick in dimm #3..period.
This is a simple mod, and its clean so I dont think the actual mod work is causing this. I also removed the mod and reconnected the trace, but it is still doing the same thing.
The first 5 boards I did this on it worked perfectly.
Any tips would be apprec as I have tried everything I can think of.
I have tried double and single sided ram btw in all 3 slots.
The voltages at the 56 ohm resistors and the actual Vbt Vreg match perfectly as well.
Sorry I can't help, but I did notice that my board overclocks my BH5 a good 8MHz higher in Dimm3... maybe this is how it was on the stock board, but Im not sure because I modded it before I even tried it out.
Well, someone having the exact same problem I am having said that he flashed to the stock Abit bios and it booted fine with sticks in dimms 1 & 3, but not in dimm 2??
Also, soon as he changed the timings, board wouldnt post again with sticks in 2 dimms. :confused:
I did some testing a while back with different slots and posted my results here... it was using standard D22 if that matters. I have a feeling D10 doesnt drive Dimm3 as hard as the other slots, while non-D10 BIOSes do the opposite, but I have only tested with non-D10.
Sorry I couldn't help more.
this is same problem that i encountered when i decided to make "voltage station" for vdimm, vtt, vref and 3.3vio...Quote:
Originally posted by Hell-Fire
Question:
The last 2 NF7 boards I have done the Vtt mod on are acting weird as hell.
They will only boot with a stick in dimm #3..period.
This is a simple mod, and its clean so I dont think the actual mod work is causing this. I also removed the mod and reconnected the trace, but it is still doing the same thing.
The first 5 boards I did this on it worked perfectly.
Any tips would be apprec as I have tried everything I can think of.
I have tried double and single sided ram btw in all 3 slots.
The voltages at the 56 ohm resistors and the actual Vbt Vreg match perfectly as well.
when i soldered a wire to pin 2 (vtt reading) on the DIMM 1, on the backside of the mobo... after i did that the board started only when i had 1 stock in slot 3... i added second stick and no boot... only one stick in slot 3 let the board start...
also when i switched video card to PCI one, i was able to "boot" with 2 sticks but only to see bootblock failure and request to insert disk with BIOS image (???).
so i removed the wire and soldered it to the cap's leg that is next to vtt IC and now everything is alright...
i have no idea why this happens on your boards, though :(
Quote:
Originally posted by bachus_anonym
this is same problem that i encountered when i decided to make "voltage station" for vdimm, vtt, vref and 3.3vio...
when i soldered a wire to pin 2 (vtt reading) on the DIMM 1, on the backside of the mobo... after i did that the board started only when i had 1 stock in slot 3... i added second stick and no boot... only one stick in slot 3 let the board start...
also when i switched video card to PCI one, i was able to "boot" with 2 sticks but only to see bootblock failure and request to insert disk with BIOS image (???).
so i removed the wire and soldered it to the cap's leg that is next to vtt IC and now everything is alright...
i have no idea why this happens on your boards, though :(
Just to make sure I understand you correctly. You soldered your monitoring wire to the cap leg under the board yes??
Weird problem either way honestly...my board and yours. Wonder if the added resistance caused by the wire is causing a drain or something.
yes sir, that is correct...
btw, this is the "new" NF7-S v2 with all those colors :) the one before died due to some issues i couldn't identify :(
http://www.xtremesystems.org/forums/...id=33552&stc=1
http://www.xtremesystems.org/forums/...id=33553&stc=1
The wire added for voltage reading is likely causing some capacitance load.
What guage wire and how long is it? There are reasons certain sized traces and capacitors are used ;)
EDIT
Pics loaded this time.. ;)
They could be causing capacitance issues, but I also remember reading in a thread yesterday about different bioses (D10 based vs. other) that changed how the nF7 acted towards each dimm slot...
tictac commented in the thread about how tight Abit had made a few settings in the bios I beleive..
EDIT 2
Hey, look what I found! :D
http://www.xtremesystems.org/forums/...5&pagenumber=3
Hi again...I've tried the thing on the pic but my board starts to beep.... and do not boot...Quote:
What am I doing wrong?
[[]]
I give up on this newer NF7-S that I did the Vtt to.
It will only boot with dimm 3 and thats it. All the voltages are tracking properly, and I have tried several different bios files.
Arghhh.
I've found a data sheet for what appears to be an overclock controller on the NF7 (the chip below the CPU socket/Vcore MOSFETS)
http://www.attansic.com.tw/english/p.../pdf/atxp1.pdf
It looks like it has 12 general purpose I/O pins, which could be being used to control a DAC to controll the VDimm/Vtt. Could a mod be based around this chip if it was being used to control the Vdimm/Vtt?
I did the VTT mod today with a 1KOhm variable resistor, and at its lowest resistance (1 Ohm), the VTT will not go lower than half of my 3.3V line (1.63V, because my 3.3V line is a tad low). I have done the VDIMM=VIO mod, but my VDIMM is usually only 3.2. I cannot get VTT any lower no matter what I do.
I've almost given up trying to get 240 FSB stable. Can my board be that bad? I have the VDD set to 1.8 and 3.2 volts running through my PC3500 EB. I have a feeling that I SHOULD be able to get to 250 FSB, but something is not letting me. I would really love any help anyone can give me. Thanks in advance.
The vtt works the other way, you set the VR at 0 Ohms and then increase the resistance to increase the Vtt, and thats if you did the Vtt mod this way:
http://s6.invisionfree.com/undercloc...?showtopic=100
Sure you've connected the VR legs the right way??? (pins 1 and 2)
I must not have said that clearly enough. Everything was set so that raising the resistance of the VR increased VTT.
Upon further prodding with a multimeter, it appears that VTT is, indeed, half of VDIMM. I measured VTT on the board and I measured VDIMM using the VDIMM mosfet near the DIMM slots. VTT is exactly half of that.
Now I'm even more puzzled as to why my board refuses to do more than 227 stably. Argh.
which BIOS are you using?
and what timings have you tried? I found my Micron based memory worked better with CAS 2.5 than CAS 3 on some BIOSes... and D10 MantaRaysXT worked best for me
I am using Tictac's D26 Mantaray (non XT). The XT version didn't let me get as high as the regular one.
My board has not liked D10 BIOSes in the past... I couldn't even get above 200 FSB.
As for timings, 2.5-2-2-5 gives me memtest errors earlier than 3-2-2-5. Setting timings to 3-3-3-11 or even 3-5-5-11 will error just as fast as 3-2-2-5. It appears to be my motherboard.
Thanks for all your help, guys. On a brighter note, I have buyers for my CPU and motherboard should I give up and decide to go Athlon 64. :p: :rolleyes:
Sell it and go AMD64 DFI :D
I was in exactly the same position as you... got to a point and couldn't get beyond it no matter what... decided to sell my trusty NF7 and am now building a DFI 250GB UT... Do the same :) Your RAM should do a lot better on an AMD64 too
:toast: So how do those s754 3000+s seem to be clocking? :D
I'll see if raising VDD helps my overclock at all. If not, I'll have to weigh my other options.
You can get a cheap Mobile 2800+ or Mobile 3000+ from excaliberpc... they should clock nicely :)Quote:
Originally Posted by mcnbns
I found Vdd had to be well tuned on my old NF7-S... too low and it would freeze and crash in 3d and memtest... too high and it would BSOD... 1.6v-1.7v was my happy medium... i tried between 1.6v-19v.
I am not sure I follow what you are saying.
You want the Vtt to be 1/2 of your 3.3v rail if you have done the Vdimm=Vio mod.
You should get to 240, atleast for benching.
Quote:
Originally Posted by mcnbns
I think I may have finally figured out the problem. Today I did some Memtesting, during which time I measured my 3.3V line on the VDIMM mosfet with a good Fluke DMM. Much to my surprise, under heavy load my 3.3V line drops to about 3.1V! That was the reason VTT was dropping when running memtest. (At least it was tracking properly :p:.)
So, to remedy the problem, I am going to do the PSU sense wire mod to the 3.3V line of my PSU to get the voltage a little higher. There may yet be hope.
I have a question. I want to voltmod my nf7-s with tictac's 4th method. I want to do all three mods. From what i understand the vdd will be half of ur vdimm, wich is the 3.3 rail -0.2 fluctuation. I have an OCZ powerstream 520 with the adjustable rails. So that means i should put the 3.3 rail to 4.0 to get 1.9 vdd?? That just doesn't seem right, so do you have to do an other mod in order to get a higher vdd voltage?
You're wrong with vdd
vdimm = 1/2 vtt
vdd is a north bridge's voltage
If you raise +3.3V rail to 4.0V, it will gain nothing, vtt will be equal to 1.45V
That's why vtt mods is needed, on Abit there's vtt's limitation (to 1.45V, you can
raise +3.3V to 4.0V, raise vdimm with VR to 3.2V, but vtt's maximum will be 1.45V,
until you will mod it)
Vtt = 1/2 Vdimm ;)Quote:
Originally Posted by Tonic
hmm, my vtt/vbt seems to track up to ~3.4. if i give any more vdimm, i get immediate errors. i have a newer board (blue slots). i know someone else with an older board, and they track as well (dont know up to what point though).
i havent verified the tracking with a dmm though...
just do 3.3=vdimm mod, VTT tracks fine then.
Very good idea to put cooling on the VTT IC anything above 3.1v though, it passes 60c quite easily...
Rub87, thanks, my error ^^
Ok I think i understand know. But what about the vtt=vref mod? Does this effect the vdd voltage whatsoever? Or do i need to do the regular vdd pencil mod in combination with the 3 mods described in the 4th method?
why do all that work?
vdd does not track vtt, vbt, or vdimm. vdd is chipset voltage and does not relate to ram voltage.
Yes i know, but i need a higher vdd voltage (1.9ish) For a higher fsb and higher clock speeds. But now i know none of the earlier mentioned mods effect the vdd voltage :) So i'll just do the pencil mod.