7x0 dothans are 533 bus C0 steppings - deffinitely the ones you want.Quote:
Originally Posted by Sume4
Printable View
7x0 dothans are 533 bus C0 steppings - deffinitely the ones you want.Quote:
Originally Posted by Sume4
Pretty sure the 730,740,750 are 533FSB while the 735,745,755 etc are 400FSBQuote:
Originally Posted by Sume4
nah that's not 1:1. i don't have ram that can do that. not now anyway...Quote:
Originally Posted by charlie
with 1:1 the highest i got so far was 248.. guess i need to offer my ram some
more juice. to bad the 3.3v rail skyrockets doing that. that 311 was with ram
@3:2, i managed this once before with the ram at 5:4 but never 1:1 :(
btw charlie, do you still use the booster or did you vmod your p4c? i'm asking
out of curiosity, as so far no one answered my question whether hooking the
3.3v line direct to the dimmslots offers a higher useable vdimm compared to
doing the oldskool vdimm mod using lm358 pin 4 to ground at same 3.3v line
settings.
i imagine it *could* be like this:
- 3.3v soldered straight to the dimmsocket offers 3.3v vdimm (all the time).
- pin4 to ground @ 0 ohms on lm358 offers 3.3v -0.10-0.12v (due to chips)
is it like that? if so, i'll do that 3.3v to dimmsocket mod, because 3.3v @3.7v
just to get a useable vdimm of about 3.47-3.5v under load is bad. and this
makes benching a nightmare.. you never know what kind of voltage your ram
gets at any given time :(
even with most of the 400 fsb parts crapping out at fsbs of 16x-18x ish, theyQuote:
Originally Posted by illmatik
are still quite fast :). the only thing you need are the higher multis to get the
overall clocks up and some good memory to make up for the low fsb. regarding
overall max clocks, charlies former 755 (part of my collection since end of last
year) is capable of doing 28xx mhz and pulls quite decent spi1m numbers that
way. my current 760 does ~2800mhz with that same cooling, but can do way
more fsb (almost twice that of the 755 :D).
overall the newer ones are the better choice, but (imo) only for the higher possible
fsb. the chips i have don't differ much in terms of general overclockability (see
above)
After some tweaking i loosened up the preformance setting wich in my case meant the clock by clock speed was reduced slightly but the cpu could clock 50/60MHz higer (other 20MHz were from vmod).
Screens (cpuz voltage reading is wrong was using 1.67v):
http://home.kabelfoon.nl/~vnrswk/spi24063dothan.jpg
http://home.kabelfoon.nl/~vnrswk/dothan3105pi.jpg
Some pics of the issolation i used in this run (it did a great job ran 4 hours straight without any problems):
http://home.kabelfoon.nl/~vnrswk/rig4.jpg
http://home.kabelfoon.nl/~vnrswk/rig5.jpg
http://home.kabelfoon.nl/~vnrswk/rig3.jpg
Trying to break the 24s in 1M and 37s in pifast tommorrow.
Quote:
Originally Posted by The Doctor
So you have done this mod? On what board? P4P800SE?
http://www.xtremesystems.org/forums/...chmentid=30658
And where do you measure your Vcore if CPU-z gives the wrong value?
Have you tried the Wire-trick (aka pin-mod)? Did it work on your board?
Jup thats the mod. board is a P4P800SE. Havent tryed wiretrick, this is easyer imho :)Quote:
Originally Posted by vegeta
Vcore measure (same spot):
http://home.kabelfoon.nl/~vnrswk/vcore1.jpg
http://home.kabelfoon.nl/~vnrswk/vcore2.jpg
Bios shows the vcore right too but i prefere using a multimeter.
Why do most people do the Vdroop like it is pictured here? The other method is a lot easier imoQuote:
Originally Posted by vegeta
http://malvescorner.net/images/p4cdroop.jpg
Thanks 4 the info :), I will do the mod shortly :DQuote:
Originally Posted by The Doctor
And how did you managed to get PAT enabled in 1:1 mode @ 258?
Which bios are you using. On my board (same as yours) it's boot @201 = PCI/AGP lock & no PAT. Or boot @200 = no lock and PAT enabled.
Are you booting with Pci card ?
Huh what other method?Quote:
Originally Posted by Nazaar
easy, wellsort of anyway. select a fsb north of 200 - 201 does the trick, thenQuote:
Originally Posted by vegeta
select the ddr500 ramspeed (divider 4:5), enable memory acceleration and set
one other thing to turbo (don't remember.. ) it's on the page with multi, fsb, ...
if you can boot with that, you got 1:1 and PAT fully enabled. if you only do
fsb 201 mhz and ram 320 (divider 5:4) you get PAT enabled as well, but only
partially...
^^^ this is talking about a (/my) p4p800-se
Jup just like u sayd only running mem 400 instead of 500.Quote:
Originally Posted by sky
@ Frozendeath :
@ charlie :
About Vcore: maybe I was really wrong & that "floppy" power connector gets electricity only for added onBoard electronics /didn´t think before of that/ & not for a Cpu (well maybe for Vtt voltage supply & reference for i/o pins). Charlie, did you try to measure with uniMeter the real Vcore on outside line/close Cpu pins (AF 8,10,12,14,16,18 ones [Vcc_core] ?
http://malvescorner.net/images/p4cdroop.jpgQuote:
Originally Posted by vegeta
this one ;)
Thanks for that link Jericho,
I found the same thing you and highlander did. I cannot post at 250 X 9, but via clockgen I can run 280 X 9 spi 1m stable, 285 locked up, but I was only at 1.525v so I bet if I raise vcore I can get into 290 + FSB
So yes there seems to be a major bios issue, some sort of wall that when we hit it there is no way to post at even 1 fsb above said wall.
The strangest thing is the wall for some is 217, 230, I'm at 247....
Thanks again, hope asus looks into this, maybe that is why bios 23 is beta :p:
Isn't that the one you showed before (in your previous post)? :DQuote:
Originally Posted by Nazaar
Or do you mean why peeps do the Vcore mod instead of the Vdroop mod?
Vdroop just stabalize your Vcore when you system is under load. And doesn't let ou choose the desired Voltage.
But after I've done the Droop-mod it overvolts my Vcore with +- 0.04V (1.618V instead of 1.583V)
Exactly 1mhz is what i found too.Quote:
Originally Posted by Frozendeath
I can manage 245 , so i guess I'm also lucky.
But although the 23beta is possibly to blame, at least the multiplier option works. Hopefuly they will fix it.
The reason I believe it to be a BIOS problem, is due to the fact I also get problems with mem timings.
If I boot and run memtest- then change Cas/tras etc Its error free.
An exaple is 233 , I still can't post lower then cas3,3,3,7 with Full PAT, yet i can adjust to Cas2 in Memtest and run 9hours full runs error free :confused:
Ahh interesting info on your mem timings as well.
I have not tried that yet, but I will do so this evening. I was running 5:4 on the mem with timings loose enough to push any problems onto the CPU as I upped the FSB.
But I will try a some tighter timings tonight and see if the no post happens to me as well.
Its all very strange, other chips in the past, most times they will post but not boot windows, or crash when they hit the desktop or whatever. Ususally add more vcore and it becomes stable. Finally you max it out and back off a little. But I have never seen this, where the chip CAN do more but something (bios ? mobo? adapter?) will not allow it to post despite its stability in windows at the higher settings.
I should also point out, that I have tried with 2x256 Bh5 & 2x256 BH6 to see if it was due to the capacity of memory, as it had been said 2x256 may give better boot FSB.
Exactly the same results - no post at 246 regardless of multiplier.
Also I can adjust mem timings in Windows and run everything there stable too - so its goto be the CT-479 & BIOS>
@ Frozendeath :
@ charlie :
& @ others too ...
The problems with initial lower Fsb boot (no mater the Vcore set) but if clock is rised in Windows later goes higher is in fact Bios/design problem according to Intel´s datasheet !!!
Yes, in electrical specification there states for Vcore: Vcc,BOOT .. 1,2V (-/+ 5% margin)!!!
This voltage is controlled by 2 output control pins on the Cpu (VssSense+VccSense) that tell MoBo that all is Ok before attempting other Post procedure, no matter what actual Vcore is set-that looks like kicks_in determined by MoBo/Bios later when POST is finished! So that Vcc,BOOT too low Vcore at begining is the problem. Same is why MoBo has to be flashed before mounting P-M Cpu on it to recognize that states (if not flashed looks likes no boot at all except on ECS-P4IBASD (i845D) MoBo).
Yes guys, I´ve done today my "homework" by deeply studying datasheets whole day [http://developer.intel.com/design/litcentr/index.htm] & I am prepairing a post with more possible enhancements & troubleshooting ideas to better OC {"Me brainstorming, You make modds!" :D :D :D ... way, like I did it on times of Tualatin_on_BX mod breakthru on MadOnion/Futuremark thead more than 3y ago) also for some other sites forum Celeron-M thread guys :) ...
So how to circumvent that boot 1,2Vcore problem? Hardware way Fixing Vcore on the MoBo & maybe some of those other 2 mentioned pins (or both) "fixing" by adding some pull_up/down resistor for forced termination or isolating them (modding the adapter or MoBo further-need of also P4 datasheet for last!) That way IMHO could the MoBo or adapter be tricked to supply @ boot higher Vcore immediatelly resolving that lower Fsb problem on boot!
The other solution IMHO could be by isolating in socket the E4 pin (PwrGood) & connect it via "wire trick" to a non isolated first (not so) close VccP pin, to trick the CPU & MoBo that all is Ok, so that MoBo supplies full Vcore@boot to a CPU & boots anyway. Same way we did it for Tuallies on slotkets time ago to enable booting ...
I also checked fotos of that Asus CT-479 I got from different sources: There on it are 2 linear voltage regulators: bigger one for VccP (1,05V -/+5%) for signal i/o Cpu stages & smaller one for i/o chipset GTLREF-erence (0,63V -/+1%) = electricity source for that is from that additional "floppy" power conn. like for other electronics there on it, while Vcore source is from MoBo P4 socket ...
And you know what that means? It means that Intel reverted back to 8y ago (to P-1 family way) for supplying Cpu with 2 main voltages-differential supplying (like last ex.P233mmx & before) again !!! One for strictly main core & one for cores i/o signal stages! Thats way you need that Asus adapter (also because too may pins are differently placed on P4 & P-M Cpu, especially a bunch of Vss and Vcc pins interchanged). P4 & other last 8y Intel CPUs have single power supplying (like AMD too, starting with classic Athlon till now) ...
Did I make my homework or didn´t I ? :p:
--
Regards, http://www.spajky.vze.com :)
Now that we know what it is maybe the bios code could be extracted and injected into IC7's or other boards? Or vmods might be the easier way to go since it seems they might need one anyways? Nice work Spajky ;)
:woot:Quote:
Originally Posted by Spajky
Impressive... very interested to see more :D
C
Nice work Spajky, what this basically means is that the adapter is not able to supply atleast 1.14volts the instant we call for it to boot at higher FSB (each adapter probably varies slightly, and thus the various "FSB Walls" we are seeing).
Thats great info, now where do we look? The bios is what is monitoring this, but the adapter is the source. hmmmmm..... Bios modding or Vmodding....
Maybe a resident bios guru could take a closer look ;)
See my previous post again, I right edit it (added another trick!) :)
Spajky
If what you say is true, then it must also be try the inital load with the default multi since changing the multi has no effect on the FSB wall.
If the max boot OC for a 730 is 12*210mhz = 2520 mhz
then changing the multi in the bios "should" yeild 10*250 = 2500mhz as a new max..... should it not?
Of course, I may be out in left field here since I have no idea what I am talking about :D
Quote:
Originally Posted by bhoward
If I follow what you are saying.... I can max boot at 237 X 12 = 2840.....
Changing the multi to 9 does not allow me to boot at 315 X 9 = 2840.
In fact I cannot even boot at 250 X 9. So that does indeed confirm that the multi in no way affects the "wall"
The question this brings up to me is:
When does the multiplier get applied to the cpu? If the set vcore does not kick in immediately, how long does the multi take to be put in?
For example, trying to boot at 9 * 265 is actually like 12 * 265 causing the initial load to be high enough to pull the 1.2v "vcoreboot" to a no post flag?
For that to make any sense then the wall must hold for every multiplier, thus for me, I can boot 12 * 237, lets say that 240 is the magic wall number..... I step it down to 9 * 250, it still won't post as its stitting at 12 * 250 the second I power it up, causing to much load for the voltage regulator to keep up until the vcore set in the bios kicks in...
Just thinking out loud here ;)
Quote:
Originally Posted by Frozendeath
Now that makes more sense, and then it CAN be fixed by a BIOS update.
I too think this, and this is why ppl are able to boot with different FSB values before they hit the wall. I guess it will depend upon
a)cooling
b)CPU's default multiplier.
E.g I can boot with max 245mhz FSB with an m740, if I were to change to a m750 (that topped out at exactly the same mhz) I might be down to 210mhz wall.