my celeron B1 will not pass 120FSB with 1.6v
if i drop the multi and try it on 140 FSB it still fails.
my celeron B1 will not pass 120FSB with 1.6v
if i drop the multi and try it on 140 FSB it still fails.
What voltage did you use with the wire mod?Originally Posted by lktuio
Celeron M is multi locked (doesn't have speedstep) so no point in changing multi because it will always use the same factory multi.
well mike..
take a look at this
that seems to be like a banias pm, not celeron mwell mike..
take a look at this
LOLwell mike..
take a look at this
Pentium-m
I have no more ideas than those explained on those links I gave...
opps ...
sorry .. i dont know why ive been thinking that it is a celeron ..
The Bannies aren't that great overclockers. I couldn't get my 1.5 banny past 2.0 and my 1.6 banny past 1.92Ghz (stock volts)
Dissipation (max) of cores @ defaults
http://www.xtremesystems.org/forums/...533#post953533
Generaly, C-M & P_M have at same Vcore & clock frequency same thermal dissipation ......
Last edited by Spajky; 07-13-2005 at 03:11 AM.
Another Possible Solution 4 better OC proposal!
Since "existing Boot&FSB Problem" is present with Celeron_M, I got another idea to trick the MoBo (w/o bios code hex hacking!) & here it is; didn´t think it about it before!
The problem is present, because at first moments/seconds of boot, the default boot Vcore of this CPU is only around 1,2V /Vcc,BOOT one!/ [see links I previously gave] & is multplier locked & even "Vid wire pin trick" or Vcore rising (in Bios Cmos settings) does nothing at those moments [kicks-in later what is set/rised in Bios] & default FSB is only 100Fsb/400qp.
So what to do? Tricking MoBo to think right @ boot that has inserted in socket a 133Fsb Cpu instead a a 100Fsb one !!! MAYBE this COULD WORK if the Cpu is not a bad example/OC-er & CM350 will boot right @ 1,73GHz! Try !!!
There are two ways:
- one doing it on the P4 socket [wire trick as for Vid definition way] (or under the MoBo soldering same) by forcing MoBo to think that has inserted 133Fsb Cpu by using BSEL (bus_select, BCLK) pins combination (H means insulating pin, L means connecting it to nearest Vss (GND) pin! Table for P4:
pin AD5 pin AD6 Vss (Gnd) closest pins: AD4; AC5,7; AE7
Bsel(1) Bsel(0) Fsb_MHz
L L 100
L H 133
H L 200 (this you do not need!)
- the other maybe easiest way is doing the "same kind of mod" on CM/PM socket on Asus_CT-479 adapter by only grounding/shorting a Bsel(0) - C16 pin (forcing it on L!) to some nearest Vss pin: B16; C15; D15,17
So CM will boot with 1,2Vcore & 133Fsb & than Bios Cmos settings kick_in (rised Vcore & Fsb over defaults, so taking care about PCI/AGP clocks & problems with peripherrals, with good cooling should go to 200Fsb & maybe more. With preventing peripherrals problems if the Cpu can not go so high, there could be some other problem to resolve:
- modding the adapter: rising VccA [measuring it first, PLL supply voltage, max.120mA - see those my previouly linked posts for PM-dothan discussion] to 1,8V instead of 1,5V, since C_M supports both voltages. This (overvolting) would help internal Cpu generated_clocks to work better/at higher clocks, same as does overvolting pure core!
This last may be needed, because I found 2 datasheets for P_M, older ones states VccA 1,8V, newer one 1,5V & Asus IMHO took for all CPUs lower one not thinking much about extreme OC-ing!
Damn, if I would be so good in money earnings like I am with modding ideas, I´d be a millionaire .....
Last edited by Spajky; 07-16-2005 at 08:11 PM.
I'm only confortable in doing the u wire mod in the adapter, i don't wanna lose the warranty of my components.
What do you guys think of 1.7vcore? is it too much for Celeron M? or too much for Asus P4P800 SE?
I will try it but i'm afraid of burning my precious components.
IMHO going over Intels max. recommended is not a good Idea if not having subzero (or at least good water) cooling, because you do not gain much, just build heat; this first started with Tualatins mainly IMHO & 1,6Vcore is already 27% over C_M defaults for core!!! Thats why I recommended VccA & VccP voltage mods on adapter, rather than too much overvolting pure core!
Cpu internal PLL generator for other clocks has not enough juice to oscillate properly if OC-ed too much ... to much overvolting can kill components easier than heat (remember the Northwood sudden death syndrome?)
NSDS .. has long vanishedIMHO going over Intels max. recommended is not a good Idea if not having subzero (or at least good water) cooling, because you do not gain much, just build heat; this first started with Tualatins mainly IMHO & 1,6Vcore is already 27% over C_M defaults for core!!! Thats why I recommended VccA & VccP voltage mods on adapter, rather than too much overvolting pure core!
Cpu internal PLL generator for other clocks has not enough juice to oscillate properly if OC-ed too much ... to much overvolting can kill components easier than heat (remember the Northwood sudden death syndrome?)
Well if it makes ya all feel better, I run 1.7V 24/7 :P. With load that is (folding).
Gets me an extra 60MHz over 1.6V :P (also means I got a 166FSB which is perfect for the bus speeds (that way I get 33/66)).
temps never go above 31C load. (swiffy block/pump/coolant with a 120mm black ice)
What exactly does the VccA and VccB mods do different from overvolting the Vcore? (bus signal overvolting?)IMHO going over Intels max. recommended is not a good Idea if not having subzero (or at least good water) cooling, because you do not gain much, just build heat; this first started with Tualatins mainly IMHO & 1,6Vcore is already 27% over C_M defaults for core!!! Thats why I recommended VccA & VccP voltage mods on adapter, rather than too much overvolting pure core!
Cpu internal PLL generator for other clocks has not enough juice to oscillate properly if OC-ed too much ... to much overvolting can kill components easier than heat (remember the Northwood sudden death syndrome?)
I guess 1.7v won't do any diference!
My celeron M hit a brick wall at more than 160FSB, it's stable at 160FSB at 1.4250v and not stable at 165FSB at 1.6v
Maybe something else is holding me back, but i really think it's the Celeron M, well i will sell it and buy myself a P-M 730.
Yes, kind a that.
VccP is a supply 2A power voltage for i/o (input/output) lines of core (peripherrical core circuits); rising it up a bit gives to them more "juice" when switching on and off states (logic zeroes & ones) & since a bit more current there flows, signal there is more clear (less subceptive to "noise"- digital "mistakes") so OC-ing by rising Fsb goes better. Thats why for example Tualatins on non-native boards w/o PowerLeap slot-1 adapter have been OC-ing better (higher Vtt=VccP than nominal voltage on those MoBo) since are multiplier locked, same as Celeron_M ! (same default clock there at defaults for both = 100MHz!)
That way you Over-volt & over-Current a bit same time those i/o Cpu lines supply & chipset*** gets on Cpu side a bit stronger signal (on memory side you have there Vdimm lines, on which you do the same with rising Dimm voltages, but in that case their i/o lines voltage+reference one rises automatically with Vdimm rising; on Cpu side that is not the case since you have separate adjustable Vcore & fixed VccP+reference voltage! On memory side many times in Bios you have a setting of "signal strength" (current setting for memory/chipset i/o lines!), but for Cpu side you do not have this there, so a mod is required on adapter/Mobo side to enhance stability there IF needed (at edge OC or if not achieving high enough FSB clocks). Since Celereons are worse (& cheaper) performers especially on Fsb side, IMHO will benefit from rising VccP more than PM 7x5/100Fsb/ or best PM7x0/133 ones.
The effect of rising VccP a bit (10% or so) is similar for CPU peripherrical circuits just like rising Vcore for central Cpu parts, but the limitation here is that it can never be higher than lowest Vcore to prevent data corruption. Rising a bit VccP there has some thermal impact on Cpu heat, but very small ...
Note: *** i/o (aGTL)signals from Cpu on CT479 adapter do not go directly to the chipset but thru "clamping voltage bi-directional chips" (3 there!), so rising Vccp affects them first before chipset ... There on adapter is also a voltage source for i/o signals reference between digital states (GTLREF, 0,63V) which can be also raised (mod!) to 0,7V together with rising VccP, so you enlarge also the margin "window" of proper recognizing 1 & 0 signals (more voltage "space" for i/o Cpu transistors to pull up & down signals).
In my case for my Tuallie (35% OC-ed can ride on lower Vcore than default!), on the contrary (having a BX old MoBo) I have lowered VccP on the MoBo (marked for them as Vtt instead, from on my MoBo 1,47 to 1,41V /default is 1,25V/ so to prevent data corruption since at heaviest Cpu load set Vcore drops from mine 1,47V to 1,44V sometimes ... I am planning to go toward C_M & MoBo changing before Christmas, since after last year close (less than 10m, almost direct) lightning hit (another long story), I have occasional problems with my MoBo ...
The story about VccA is the supply power for CPU internal clocks PLL generator. If higher, it helps keeping PLL circuitry running stable @ higher Cpu frequency & directly impacts on Cpu stability at high clocks (much more than VccP!). Since datasheet states for there bi-voltage source (around 1,5 or around 1,8V), I would opt for higher one, since IMHO Asus CT adapter uses single & lower one. For that to achieve, another mod on the adapter could be needed ...
Those kind of mods for someone having "guts", the will, knowledge & practice with delicate soldering (having proper equipment too) are not so complicated to make if you know what U are doing, having in mind also, that sometimes doing that, sh1t can happen finishing killing some stuff.
I am planning also to make some mild careful "lapping" the Cpu core (like all mine CPUs till now) & pre-burning it like this I did before: http://freeweb.siol.net/jerman55/HP/preBurn-in.htm together with CT adapter
Yes, I do useful mods for fun too (browse a bit my site, well, needs some updating...) & the last one was just yesterday on my car (don´t ask!)
Please "snip" quoting while replying ...
So to some it up Spajky, it sounds like you're talking about improving the FSB signaling strength (or improving the signal/noise ratio at higher speeds). And to do this would require tinkering with the VccP pins (which there seems to be a bunch of). Does the adapter regulate the power to the VccP pins? I really don't know much of how the adapter works in supplying power to the CPU. I need to get a Digital Multi-Meter.
adapter generates all those voltages spajky mentioned.. that's for it has the power adapter
Spaiky, i've read now all your posts and You seem to have a strong point there.
Actually, when pm @ desktop madness started with dfi board, i wondered if fsb limit/oc'ability has something to do with lower io voltages and power-saving bus.
now, i think, you have proven in theory, that it is so.
i might think about modding my adapter (measuring voltages first, of course
Yes, that bigest (looks like Vcore Mosfets on MoBo) voltage linear regulator there does the job for all of VccP pins ...
mod it nowadapter generates all those voltages spajky mentioned.. that's for it has the power adapter
Spaiky, i've read now all your posts and You seem to have a strong point there.
Actually, when pm @ desktop madness started with dfi board, i wondered if fsb limit/oc'ability has something to do with lower io voltages and power-saving bus.
now, i think, you have proven in theory, that it is so.
i might think about modding my adapter (measuring voltages first, of course
Maybe ...
With native Centrino chipsets (first DFI or Aopen board on i855) there could be also a chipset or Bios problem, not just lover voltages not to be able to go much higher than 166Fsb (PCI/agp locks etc), while with Asus that should not be the problem (some people did with CT adapter even 300Fsb! as I saw, by lowering multiplier on PMs).
I really do not see why Celeron_M (dothan) shouldn´t go higher on Asus_CT, especially if C-0 revision or 350J one. The core is practically the same as last PM ones (its not Banias-"poor OC-er"), so with some overvolting should IMHO do 2,6GHz (200Fsb/100% OC) "on _Air", heat/cooling is not a problem as looks. Maybe we Are just too demanding (70% OC on air is an excellent achievement; times ago 50% was very nice result!) for this low priced CPU. But I really do not believe, that the core of C_M350 is so much worse made (maybe 2nd class cheaper silicon wafers?) than P_M ones ...
Sooner or later we will find out what is/was the real bottleneck ...
@ ibby : don´t be unpatient ...
PS.: Can someone with digital camera make me a big close high resolution *.jpg shot of Asus_CT-479 upper side & upload it somewhere posting me here URL for DL-it; my existing one is not enough detailed slide even if I enlarge it ... TIA ...
Last edited by Spajky; 07-20-2005 at 12:27 AM.
Hmm, then we'll just have to mod that linear regulator when I get my adapter back from RMA (No I didn't kill it, but if it dies again I'll be damned if it wasn't me who did it
Just to add another perspective on this - though I've said most of this before - my i855 board (the AOpen one - no mods)) quite happily runs my 350J @ 180MHz FSB but my P4P800-VM locks up around 170MHz... 166MHz is fully stable (and what I'm booting the VM + 350J at). And though the Banias chips are crap overclockers in terms of total MHz my Banias 1.3GHz PM runs happily @ 180MHz in the 855 board too so there's definitely something weird going on here. When I get a chance (I'm working away from home at the moment) I will buy a decent board to replace the VM... (thinking P4GD1 as I also want to upgrade my GFX soon and will make the switch from AGP)... Then I'll get the chance to torture my poor celery some more!Maybe ...
With native Centrino chipsets (first DFI or Aopen board on i855) there could be also a chipset or Bios problem, not just lover voltages not to be able to go much higher than 166Fsb (PCI/agp locks etc), while with Asus that should not be the problem (some people did with CT adapter even 300Fsb! as I saw, by lowering multiplier on PMs).
I really do not see why Celeron_M (dothan) shouldn´t go higher on Asus_CT, especially if C-0 revision or 350J one. The core is practically the same as last PM ones (its not Banias-"poor OC-er"), so with some overvolting should IMHO do 2,6GHz (200Fsb/100% OC) "on _Air", heat/cooling is not a problem as looks. Maybe we Are just too demanding (70% OC on air is an excellent achievement; times ago 50% was very nice result!) for this low priced CPU. But I really do not believe, that the core of C_M350 is so much worse made (maybe 2nd class cheaper silicon wafers?) than P_M ones ...
Sooner or later we will find out what is/was the real bottleneck ...
@ ibby : don´t be unpatient ...
PS.: Can someone with digital camera make me a big close high resolution *.jpg shot of Asus_CT-479 upper side & upload it somewhere posting me here URL for DL-it; my existing one is not enough detailed slide even if I enlarge it ... TIA ...
There defintely is something wrong with being disappointed with a 66% overclock though!
I am contented with my 80% overclock although I can go as high as 85% (celeron M 350J 1.3@2.34).
BUT, I am still curious whether the wall is really due to CPU or other factor. If it is other factor, I will be more than happy to get 200 x 13 (2.6G) and make a direct comparison with FX-55.
It will be very ironic if a "celeron" beat an "FX-55".
Posting Permissions
Reply With Quote
Bookmarks