1333 FSB Mod, Confirmed To Be Working By Me.

Updated for Easier Pic.

This is showing the bottom of the CPU as if you had it upside down, facing you.

Original Post.

got bored tonight and was looking around at VR forums and came across a old pic Shamino posted awhile back and thought... why not give it a shot...

so I took one of my spare conroes, and got my conductive ink pen, did the mod, let it dry, took my PC apart and popped this in... and it almost worked.

it booted at 1333, then I rebooted, and it went back to 1066... and was jumping back and forth.
so I took the CPU out, and saw, I didnt have a nice clean connection between the two BSEL pads... so I traced over it again, and let it dry, blah blah blah...

10 minutes later... it works.
now, it's working just fine.

boots up everytime at 7 x 333.

no big deal really since I know this CPU can do over 590, but its still fun to have a higher default FSB for the hell of it. :D

I wanted to see if this works, so I could try it on my quad core CPU...
hoping theres a slim chance in hell, that if it boots at 1333, the chipset strap will be looser, and give me some extra OC room.
thats a 1-100 shot... but if I never do it, how can I say it wont work??

atleast this one worked.

what would be kinda interesting, is if you took a E4300, and did this mod to it, assuming you knew the CPU could do 3 ghz, it would boot up at 9 x 333 for 3 ghz.
that would be pretty cool for those cheap boards that been coming out at Frys and such, with the CPU.

anyways, next try is to make my Quad Core do the same thing...
not really holding my breath on it, but its worth a shot.

heres the boot up speed I got from CPUZ after the mod.

http://valid.x86-secret.com/show_oc?id=165085

total down time, about 20-30 minutes, to take it apart, do the mod, let it dry and put it back together.

last note...
before doing this kinda mod, make sure you have a bios on your board that will allow a 1333 FSB CPU to be used in that board.

interesting post

ah interesting to see how the quad core cpu does :)

heres some better proof that it works. :D

you can't fool the "Auto" setting.
set it to auto in bios, it should go back to default 1066...

but as you can see here... it doesn't. :D

cool Kunaak. nice idea
most interesting to me would be to see if this has any effect on fsb walls when using 6x.
all the conroes i've had have walled somewhere below 500fsb :/ (some as low as 460 even on p5b-dlx)
so here i am buying and selling cpu's just trying to find one that will go to a decent speed on LN2 without having to buy the hellishly expensive unlocked series.

(intel i hope you read this...fix your fbs wall issue please ;) )

I honestly dont believe CPUs have FSB walls.

2 years ago, if you claimed to be able to get a Pentium 4 to 500+ on the FSB. people would have just said you were a liar.

today, people do it all the time, in these latest 7 and 8 ghz LN2 tests with pentium 4's and the latest generation boards.

ummm...well you would if you had 2 e6600 at 6X multiplier that stopped going faster at inexplicably low fsb on a p5b-dlx known to go much higher on the fsb.
swop one chip for the other and the fsb wall changes....and this all while hovering below 3ghz for the core....

fsb walls exist...you just have to look for them

Quote:

---

Originally Posted by flytek

ummm...well you would if you had 2 e6600 at 6X multiplier that stopped goin faster at inecplicably low fsb on a p5b-dlx known to go much higher on the fsb.
swop one chip for the other and the fsb wall changes....and this all while hovering below 3ghz for the core....

fsb walls exist...you just have to look for them

---

no they dont.

the P5B deluxe has a NBCC that gets higher the lower you set the multiplier.

say you have your E6600 and are setting 6 x 500 cause you think that will work.

Multiplier divide by set multiplier = X, take X and multiply that by the FSB, then take that number and multiply that by 4 and you get the NBCC.

9/6 = 1.5 x 500 = 750 x 4 = 3000 mhz.

a 3 ghz NBCC is extremely high and in 90% of cases will not be bootable.

a CPU with a lower default multiplier has a far better chance at hitting 500 then a CPU with a high multiplier, which is why you see E6300's do it all the time, but never a E6700.

yes, the 4 meg cache will hold them back... but even a 4 meg cache CPU can hit 500 stable... so thats not the issue.
its the north bridge core clock.

a typical NBCC will not like going higher then about 2400 mhz before freaking out.

if your doing default multiplier, 2400 is enough for a 600 FSB.
but once you lower your FSB, you start to drastically decrease your CPU's ability to hit stable FSBs that are really high cause you greatly increase the NBCC faster.

funny that...as i was using 2 e6600's so as not to have the issue supposedly created by non default multipliers.
both e6600 have fsb walls
one at 460 and the other at 495fsb.
now the the really strange thing is that changing the miltiplier back to default and using -70' phase cooling made the fsb wall on both slightly worse....

so after extensive testing i can comfortably say the theory of the nb strap and default multi creating a NB fsb wall is not applicable in this case.
also i can say with 100% certainty that changing the multiplier does nothing to move the fsb wall.
and lastly two of the exactly the same cpu (using whatever multy is needed to rule out core speed as the wall) can and do have FSB WALLS at different frequencies.

if you have 2 of the same cpu like i did at the time you will be able to see the same behaviour i would think.
unless your board walls first or the cpu's are so identical that spotting the differece becomes difficult.

enough with your CPU, if you wanna discuss it further, start a topic please and dont hijack mine.

Quad Core 1333 FSB Mod Works. :woot:

very nice.. :)

I wonder if there is any way to play with Bsel pads to force the 1066 strap all the time on a 1066 strap cpu....mostly for Asus P5Bdlx that switches strap at 401 by itself, could something like that work?

oh well...this just reminded me of the lanparty revb days when ppl modded the cpu's default fsb and wound up getting higher clocks from it.

thanks for reminding me of that situation as now i can test this method you brought to my attention and see what effect it might have on conroe fsb walls.

will post back if this does indeed wind up helping max clocks in certain situations

Yeah.... :p:

it does help alittle for Quad Cores.

Been Running 350 FSB tops for the last 2 days (since I moved from the P5B deluxe to this board 2 days ago) and now, I can get 366 stable.
still no 400+ very easy... it will boot, but wont make it into windows at more then 400+ yet.

but still, this is a improvement of 16 FSB.
pretty cool so far. :)

Good to see it working on Quad, lot's of chip's may get some remodeling on the underside now. Nice one kunaak.

btw kunaak, the idea of NBCC was discarded. even by Tony ;)
but nice to see that bsel trick has an use now with those 1333 fsb bioses..
when i tried it a while back, i had asus p5wdh and of course it didnt help at all.

by the way - does the 1333 nb strap kick in with 1333 fsb cpu or does it still require booting at 401+ as usual?

very nice we should only short red line with conductive pen for FSB 1333?

Heres the Mod as I did it.

http://www.xtremesystems.org/forums/...1&d=1170759046

http://www.xtremesystems.org/forums/...1&d=1170759094

Quote:

---

Originally Posted by caater

btw kunaak, the idea of NBCC was discarded. even by Tony ;)

---

Right, lower multi doesnt mean less FSB.

Quote:

---

by the way - does the 1333 nb strap kick in with 1333 fsb cpu or does it still require booting at 401+ as usual?

---

Thats what id like to know, 401 or at another point..

Quote:

---

Originally Posted by Kunaak

I honestly dont believe CPUs have FSB walls.

2 years ago, if you claimed to be able to get a Pentium 4 to 500+ on the FSB. people would have just said you were a liar.

today, people do it all the time, in these latest 7 and 8 ghz LN2 tests with pentium 4's and the latest generation boards.

---

I think believe CPUs have fsb walls..

Why?

I tested my E6600ES in Asus P5B-Deluxe..
Max FSB test, with..

6x ratio: 530Mhz with air
7X ratio: 530Mhz with air
8x ratio: 530Mhz with air

8x ratio: 566Mhz with Vapochill LS

I think need cold (delta temp) for cpu, if we break over fsb wall.

Pre 1333fsb support added to bios this mod did not work, Post support we now have it working.

Kunaak:
We've been doing this over in the WCG section for a couple months now on the woodcrest and clovertown rigs.
The clovers are essentially Kentsfields with some added instructions to run in duals. Keep in mind that the boards we work with don't have the OC features that the gaming boards have so we had to get a little creative.:D
I've run my dual clovertown 2667/8mb/1066 up as high as 3330/8mb/1333 using the 10 multi and to 3271 using the 9 multi.
Thats with 1.35v.
We're also handicapped using FBDimms and there's a 370 "wall" in the boards but it's still impressive to see 8 cores at those speeds.
As to what it will do, go look at HWBot on Sisoftsandra or wPrime..;)
All on stock volts but there are volt modds available up to 1.55 for these boards. I get a little skittish on moving the volts over 1.45 as these are $500.00 boards with $3000.00 in chips and $700.00 in memory.

amazing I will try with my E6300 and E6600

Quote:

---

Originally Posted by Kunaak

I honestly dont believe CPUs have FSB walls.

---

Try to tell this to all the people who have had various E6400 cpus which mysteriously stop at, for example, 480x8, 480x7, while another E6400 cpu can do 480x8 or 525x7, on 1333 strap, all the other hardware being the same.

Also, the GTL reference voltage mod existence confirms this as well as some users who have been banned for posting things about the FSB wall I wouldn't like to discuss openly here.

Other than that, nice find on the 1333fsb mod :) Might try it on my 6400to see what does it change ;)

Kunaak I assume you are doing this on your P5N32...if so, what bios are you using ??

Very nice 4300 will go to heaven ^_^

Quote:

---

Originally Posted by mrlobber

Try to tell this to all the people who have had various E6400 cpus which mysteriously stop at, for example, 480x8, 480x7, while another E6400 cpu can do 480x8 or 525x7, on 1333 strap, all the other hardware being the same.

Also, the GTL reference voltage mod existence confirms this as well as some users who have been banned for posting things about the FSB wall I wouldn't like to discuss openly here.

Other than that, nice find on the 1333fsb mod :) Might try it on my 6400to see what does it change ;)

---

send me that CPU that has a wall...

watch me break it.

PS.

I am sick of coming in here and seeing people trying to take over the post with other topics...

if you wanna start a topic about FSB walls... do that.
I already asked one person to stop trying to hijack my thread already...

Kunaak, nobody said "FSB wall" couldn't be broken :) The point is, it exists on unmodded hardware and kicks in, sometimes not allowing to max out the potential mhz for the cpu it is actually capable of.

Now your 1333fsb pinmod might help here a little as well - what you showed with your Kentsfield right here - so I wouldn't call this discussion "hijacking" of your thread, and I think it would be on topic if people reported in here if this mod helped to achieve them higher fsb on various Allendale/Conroe/Kentsfield cpus.

However, if you think otherwise, I believe, you have the moderator power to make your corrections :)

that's pretty good. Did you have to increase the voltage? A 500 Mhz overclock is a pretty big deal.

Quote:

---

Originally Posted by Kunaak

I honestly dont believe CPUs have FSB walls.

2 years ago, if you claimed to be able to get a Pentium 4 to 500+ on the FSB. people would have just said you were a liar.

today, people do it all the time, in these latest 7 and 8 ghz LN2 tests with pentium 4's and the latest generation boards.

---

You need to play with a Yonah :)

Nice thinking on the 1333 mod on quadcore! I need to go out and get a conductive ink pen now :(

Quote:

---

by the way - does the 1333 nb strap kick in with 1333 fsb cpu or does it still require booting at 401+ as usual?

---

I also want to know, and would this change ram dividers on 680i like it would with a 975(I think?)?
Would you be able to run a 1333 cpu on the 1066 strap? What about the performance implications?

Can you run some 32m for me ? :)

Very nice... i also saw that mod some time ago but i've been afraid of using it...

Now i will try it on my X6800... thanks a lot :)

awsome discovery...
thanks Kunaak

by the way i have bumped an old fsb wall thread to continue offtopic discussions over there. post#60 page 3. http://www.xtremesystems.org/forums/...81#post1997281

Quote:

---

Originally Posted by flytek

well a lot has happened since this thread started (like gtl vmods and boards that can tune it) but i have a reason for dragging it up.

Kunaak has tested a pin/paint mod that defaults cpu's fsb to 333bus.
so far this has resulted in a 16mhz fsb improvment on his Q6400.

anyway what needs to be tested is if the fsb walls improve with this mod for conroe and allendale cpu's.

here's the mod. post #18 http://www.xtremesystems.org/forums/...d.php?t=132900

---

Looks like more of a benefit for Quad cores. Can we get some posts from people trying this out on their own?

Looks good, congrats on the mod.

awesome find Kunaak, now we need to test if 24/7 about a month or two to see if will not damage the cpu ;)

when I buy a p5bd i will do it for sure... now my crappy board max fsb is 300 :)

Tanx a lot for Pics but its possible on Gigabyte rev1.0 mobos? anyone tried with Gigabyte or Asus?

Quote:

---

Originally Posted by Kunaak

no they dont.

the P5B deluxe has a NBCC that gets higher the lower you set the multiplier.

say you have your E6600 and are setting 6 x 500 cause you think that will work.

Multiplier divide by set multiplier = X, take X and multiply that by the FSB, then take that number and multiply that by 4 and you get the NBCC.

9/6 = 1.5 x 500 = 750 x 4 = 3000 mhz.

a 3 ghz NBCC is extremely high and in 90% of cases will not be bootable.

a CPU with a lower default multiplier has a far better chance at hitting 500 then a CPU with a high multiplier, which is why you see E6300's do it all the time, but never a E6700.

yes, the 4 meg cache will hold them back... but even a 4 meg cache CPU can hit 500 stable... so thats not the issue.
its the north bridge core clock.

a typical NBCC will not like going higher then about 2400 mhz before freaking out.

if your doing default multiplier, 2400 is enough for a 600 FSB.
but once you lower your FSB, you start to drastically decrease your CPU's ability to hit stable FSBs that are really high cause you greatly increase the NBCC faster.

---

good explanation and I would concur with this based on my own observations.

Has anybody else noticed that the pins marked in the diagram in the original post and the pins modded in Kunaak's pictures are not the same?

I tried it on my E6600ES, but it doesnt seem to work..

I will try again tomorrow..

http://mekk.homeip.net/cpgallery/dis...e.php?pos=-327
http://mekk.homeip.net/cpgallery/dis...e.php?pos=-326
http://mekk.homeip.net/cpgallery/dis...e.php?pos=-330
http://mekk.homeip.net/cpgallery/dis...e.php?pos=-329

why not oc it to 1333? and you get more mhz and more fsb :P


good find

Kunaak, i'd like to ask you a single question... the higher multiplier you can select after and before the 1333 MHz fsb mod is the same?

Thank you for sharing this informations. :toast:

multipliers remain the same with and without the mod.

why not just run 1333 OC?
cause some people have boards that just won't let you... like the cheap ECS boards people been getting with the E4300 at Frys.
this kinda mod would be good for them, as I said in my first post.

also, I wanted to test if a CPU booting at 1333 instead of 1066, would some how change how the CPU OC'd on the board... and it did in my case.
so I really dont know what to make of the mod yet, other then it works.
but why it effected my OC I can't say I know...

as for "did I need to increase the voltage".

like I said already, if your CPU cannot boot at default multiplier x 333, at stock voltage, then this mod isn't for you.

I know you can alter the default voltage, if you mod some other pads, but I really dont want to go that far.

Didn't pin modding fsb to 200 help some chips with clocking back in the AXP days? I remember some sort of pin mod... Maybe this is the same. Could someone try running a stock chip 1333 strap on p5b then a modded chip?

Quote:

---

Originally Posted by p0tempkin

Has anybody else noticed that the pins marked in the diagram in the original post and the pins modded in Kunaak's pictures are not the same?

---

yeah, they are not the same...
i don't know which one is the right one...

i think this one is the right, as i can see on Kunaak's picture of modded cpu:

http://img245.imageshack.us/img245/1350/1333fsbid4.jpg


PS: i had to flip the image to the right side.

what I am most curious about now, is to see how this mod, effects boards that are usually limited to the 1066 strap, like various 975X boards...

original pic replaced, for a clearer understanding of the mod, and how it needs to be done.

Is there any chance doing this can screw up your processor (aside from the inevitable voiding of a warranty)? Such as it frying out, being unable to fit in socket, or what would happen if you connectd too many of the connector points together?

For anyone curious...

I undid the mod today, to test more...

and now, 366 FSB isnt possible for me anymore.
back to the old top of 350.

366 just will not remain stable anymore, without the mod...

very informative work kunnak ^5

id love to see this mod on a e4300/ecs fry's combo
i would go out and buy it, just to try it if i weren't in the east coast

somebody do it!

I gave it a shot but it doesn't look like it is working

mion, what board are you using? Is the bios programmed to be able to read the new bsel flags and set accordingly?

Nice find. I'm actually a little bit surprised to find that it works still. I remember when I started the pinmod trend for the dothans. But then, with the meroms (and possibly yonahs), and whatever intel was using for power consumption control (c1e maybe?).

With the meroms, forcing a 533fsb to 667 resulted in proper fsb manipulation but stuck the cpu into the lowest multiplier iirc. So it's definitely good to see that it's working well with the kenties again.

Any chance you could post some mem benches like with mbench for it at 333fsb via bsel manipulation vs bios fsb overclocking? It sounds sure enough from what you've witnessed with the extra 16fsb that it's kicking into the next strap right away.

Then again, from looking at it at another angle, wouldn't 350fsb and the lower strap result in better performance than what 16fsb could give you, at the sacrifice of a higher strap?

Also, you mentioned the ecs boards. It's a shame I don't have any fry's around me or else I would have picked up one of them by now for sure. But I take it they support 1333 with a new bios?

looks nice , gone try it l8 out on my X6800 .
i think cpu's dont have fsb wall to , try'd 5 baords ;) .
2 p5b green , 2 P5W DH and a Abit AB9 pro .
last P5b maxxed out @ 545fsb stock , the first one 490fsb stock .
last P5W DH maxx out @ 420fsb with 1.9vmch , first one 410fsb with 1.45vmch :p .
the abit maxxed out @ 450fsb with 1.7vmch , this whas all done with same cpu and cooling air !

Quote:

---

Originally Posted by flytek

ummm...well you would if you had 2 e6600 at 6X multiplier that stopped going faster at inexplicably low fsb on a p5b-dlx known to go much higher on the fsb.
swop one chip for the other and the fsb wall changes....and this all while hovering below 3ghz for the core....

fsb walls exist...you just have to look for them

---

i totaly agree , FSB wall exists


i tested last week a E4300 on my p5b deluxe (570FSB with my e6600) and with my TG 5300 (DDR1200) ,and the CPU (E4300) wont boot past 370 FSB with any multi




i also had a E6400 with same problem , tested in 3 diferent boards and FSB wall was the same in all the boards with all multis



cheers

Quote:

---

Originally Posted by blind_ripper

looks nice , gone try it l8 out on my X6800 .
i think cpu's dont have fsb wall to , try'd 5 baords ;) .
2 p5b green , 2 P5W DH and a Abit AB9 pro .
last P5b maxxed out @ 545fsb stock , the first one 490fsb stock .
last P5W DH maxx out @ 420fsb with 1.9vmch , first one 410fsb with 1.45vmch :p .
the abit maxxed out @ 450fsb with 1.7vmch , this whas all done with same cpu and cooling air !

---

545 sounds like cpu wall with that 2nd P5B on air, a decent one though :D

I did mod, now go home test, and will tomorrow write result.

Sorry, photo made with low resulation webcam.
http://img381.imageshack.us/img381/7087/1333modyx8.png

Quote:

---

Originally Posted by loc.o

545 sounds like cpu wall with that 2nd P5B on air, a decent one though :D

---

killed the bord but wittekakker will try to replace the dameged parts .
now its vmodded and ill tell u ;)
http://i3.photobucket.com/albums/y90...r/530fsb8m.jpg
btw how comes do u think that fsb gose up when u increas vmch :p

Edit : tomorrow il have a Evga 680i to test with so il report max fsb to :) , also if the 1333fsb mod works with modded bios ;)

I dont believe CPU's have FSB walls...

boards can, but not CPU's.

sounds like a good mod.. need to test it on my E6300.

Quote:

---

Originally Posted by Kunaak

I dont believe CPU's have FSB walls...

boards can, but not CPU's.

---

idd kunaak i think the same .

Good work Kunaak. This mod is very useful for those who have systems that do not allow for tweaking.

Edit:

Quote:

---

Originally Posted by Kunaak

I dont believe CPU's have FSB walls

---

Indeed.

All CPU's have a maximal limit. If one wishes to call this a FSB wall, they may do so, however they are muddying up the waters of clarity in doing so, much the same as one who calls the speed of light a velocity wall of the spaceship. The spaceship will have it's own limit and the actual speed of light is a constant (for our intents and purposes at this time at any rate). The CPU has it's own limit and the actual restrain otherwise is left to the constancy mainboard ==> chipset limitations.

Great find mate.
Reminds me a lot of the ol L12 mod. 16fsb is certainly nothing to be sneezed at.

Asus 1333 FSB Bios and Board list.

Really interesting modification ;)

Best Regards

Kunaak nice topic and mod as well as pic and documentaion and testing. Trumps to ya dude. So just to check FROM YOU ONLY PLEASE

OC puts a strap on the NB, this mod automaticly forces the FSB to the max built in OC'd FSB option of 1333 Mhz and no presure on the NB. All OC'ing after mod is extra FSB onto of the 1333 Mhz FSB already and starting to clock the NB but not as much as if we went straight in the BIOS and did it?

Or am i being totaly thick?

When my X6800 come back off warranty i think i'll be getting this mod done.

So i guess in lamens terms it

Boosts your FSB to 1333 MHZ with NO OVERCLOCKING in the BIOS or stress on the NB aka fooling it

Locks out the option to set betwen 1066 and 1333 Mhz

Back off the NB strap when over clocking.

sorry if i am being silly here and repeating but my list of mods for cpu and mobo is getting big!

I'm running a GA-965G-DS3 with the integrated graphics, which pretty much shuts off overclocking. Would this mod get around that issue?

the g965 chipset has issues in and of itself that prevent it from going into the 300's as it is. there may be a newer revision out soon to support 1333, but it's unknown. I noticed gigabyte released a 3.3 board for the 965g-ds3 but it doesn't show support for 1333 like the 965p-ds3.

Figure I'd update with my findings. I've got a ds3 rev2 at the moment with a e6400 es (b1 stepping). I did the bsel manipulation while having f10 bios, which allegedly supports 1333fsb (albeit on rev3.3 boards only..). It gave me the usual stink I've been getting from this pos board lately and I had to remove a stick of ram for it to post. It posted at 2.13. Still 266fsb. Grr. Then, I flashed back to my personal fav, f7, which doesn't support 1333, and it absolutely freaked out. In the past, when using a bsel higher than what's supported (like 533 in 400 855 chipsets for dothan etc), it'd just run at the lower bsel flag equivalent. But on the ds3, it'd turn on for a split second (way shorter than standard up-down that the stupid ds3 does), then off, then on for a split second, etc. Once I removed the mod, I again had to pull a stick of ram for it to boot. But now with normal bsel flags for 266fsb, it's posting and booting again on f7.

E6400 with mod + p5b dlx rev 1.03g 1004bios= Not working

Damn just bought a p5n-e sli.
Doesn't seem to be on that list. Hopefully a bios update will fix that.
Anyone have Core 2 instead of quad results?

Quote:

---

Originally Posted by Kunaak

I dont believe CPU's have FSB walls...

boards can, but not CPU's.

---

chipset/mobo have always been the culprit, wonder if it'll ever change..

so, Kunaak/anyone on 975x + kent tried yet?

1600fsb should be just as doable, but the mobos/bios though...

Quote:

---

Originally Posted by over@locker886

I did mod, now go home test, and will tomorrow write result.

---

The mod works well..when computer power on, bios wrote New Cpu installed, after bios entered, I saw default clock is 9*333:banana:

But this mod isn't help cpu overclockable and more fsb:(

Quote:

---

Originally Posted by I)ickie

Damn just bought a p5n-e sli.
Doesn't seem to be on that list. Hopefully a bios update will fix that.
Anyone have Core 2 instead of quad results?

---

I can't imagine core2 being different than core2 quad. The architecture is the same and bsel is bsel. No if's and's or but's. I guess the only problem is a motherboard that can interpret it and be able to do something with it.

For instance, in my case with the e6400 and ds3 rev2 board: with f10 bios and mod, it posts but at 266fsb. With f7, for some reason, the motherboard absolutely flips out. I would have expected it to boot but at 266 but alas, no. F8 and beyond (possibly F9 I can't remember) support 1333, but allegedly only with rev3.3 boards. If it works with kenty's, I have no doubt that it'd work with c2d's too with the right board.

Would be so very interesting to see whether theres also a FSB 1066-Mod for the Prescott2M!

See the thing is, i have a 630 pretested @ 4.4GHz, but my Asus P5AD2-E (925XE, capable of 1066MHz FSB!) often does crash when i OC it to 266MHz...

How could you figure out which pins you'd have to mark to get a 1066FSB?

Because theres gotta be an Option for that, hence the 3.73XE with 266QDR.

EDIT:

I Found the specs in the Prescott 3.73XE Datasheet. Will give it a try :)

lets take a closer look what are we doing:

I've read the Intel's Core 2 Duo/Quad processor documentation to figure out what are those pins.

On page 49 of Core 2 Duo processor, we can see that we are making a simple jumper between G30 (BSEL2) <-> J30 (VCC)

http://img64.imageshack.us/img64/5228/intel01zz4.jpg



On page 60, G30 means:
BSEL2 - POWER/OTHER -> Output
http://img64.imageshack.us/img64/5023/intel02cl7.jpg

And on page 61, J30 means:
VCC - POWER/OTHER -> ?
http://img528.imageshack.us/img528/5223/intel03zc9.jpg


So, on page 75, I've found that VCC is an Input:
http://img70.imageshack.us/img70/3426/intel05hq8.jpg


And on page 69, we can se an explanation what is BSEL[2:0]
http://img70.imageshack.us/img70/5100/intel04ei2.jpg

and it points to Table 17 (page 30) which is:
http://img232.imageshack.us/img232/1880/intel06sd3.jpg
I can't see a possible combination to 333MHz, just 266MHz (Q/X/E6XXX) or 200MHz (E4XXX). Maybe just newer processors have it (?) :rolleyes:


So, this mod is simply make a jumper between BSEL2, an output FSB frequency selector, and VCC which is a simple power pin.

By making the mod, you are taking BSEL2 from L to H. That would move you from the first row int Table 17 to the last row in Table 17. It's safe to say that the last row in Table 17 is "RESERVED" for 333 MHz FSB.

Seems pretty straightforward to me.

kunaak,


have you been able to measure any increase in temps by simply jumping to 1333. either cpu or chipset?

i know its prob. not going to happen but it would be interesting to see what the cpu and/or chipset does to itself in terms of output wattage/heat when doing this mod. i know in appearance its nothing more than changing the FSB to 333. but in the internals of the cpu and motherboard,, whats happening.
is the cpu jumping up its default voltage?
is the chipset changing the latency? is it also increasing the voltage its asking for??

Quote:

---

Originally Posted by virtualrain

By making the mod, you are taking BSEL2 from L to H. That would move you from the first row int Table 17 to the last row in Table 17. It's safe to say that the last row in Table 17 is "RESERVED" for 333 MHz FSB.

Seems pretty straightforward to me.

---

no its not safe to say that unless you can find what that registered is reserved for.
i mean obviously its happening, but its a risk,, a big one, and thanks to Big K for possibly sacrificing one of his chips to test it.

makes me want to tear apart my rig and try it.

Quote:

---

Originally Posted by virtualrain

By making the mod, you are taking BSEL2 from L to H. That would move you from the first row int Table 17 to the last row in Table 17. It's safe to say that the last row in Table 17 is "RESERVED" for 333 MHz FSB.

Seems pretty straightforward to me.

---

so, its not possible an E4XXX (200MHz) become a 333MHz, because 200MHz are L-H-L and 333MHz are H-L-L

we have to mod more pins (BSEL1 and 2) on E4XXX in order to work, like:
http://img127.imageshack.us/img127/6796/intel01ack3.jpg

make a jumper between G30 <-> J30 and isolate H30 ;)
or simply isolate H30 to have 266MHz, an E6600 with half cache.





/// EDIT
reading Intel's Xeon 5100 Series Datasheet I've found that on page 19:
http://img71.imageshack.us/img71/3898/intel07ib1.jpg

So, the reserved is for 333.33MHz, and it is H-L-L ;) . The pins are the same, G30.

how did it boot just like that on the quad core? didnt it need more voltage at 1333?:eek:

Quote:

---

Originally Posted by GrimReaper

how did it boot just like that on the quad core? didnt it need more voltage at 1333?:eek:

---

on my CPU no.
mine will do 3.4 ghz at stock voltage, and 3.6 ghz with 1.38 volts.
thats the tops I was able to easily keep stable in my P5B Deluxe.

same CPU tested on RD600 and 680i, and in all cases, the CPU doesnt need any extra voltage to reach 3 ghz.
my CPU's just a good one, but I also said many times...

make sure your CPU can do the stock multiplier x 333 at stock voltage, otherwise this mod would be useless to you.

heres how to make a E4300 boot at 9 x 266.

http://www.ocforums.com/showpost.php...1&postcount=40

Quote:

---

Originally Posted by over@locker886

The mod works well..when computer power on, bios wrote New Cpu installed, after bios entered, I saw default clock is 9*333:banana:

But this mod isn't help cpu overclockable and more fsb:(

---

shame,i was getting all exited . have you tried benching? to compare and see if theres any loss or gain ?

nice work kunaak .

Quote:

---

Originally Posted by Kunaak

heres how to make a E4300 boot at 9 x 266.

http://www.ocforums.com/showpost.php...1&postcount=40

---

:eek: I need to grow some balls before I can try this on my 6600 :p:

Quote:

---

Originally Posted by MarlboroMan

so, its not possible an E4XXX (200MHz) become a 333MHz, because 200MHz are L-H-L and 333MHz are H-L-L

we have to mod more pins (BSEL1 and 2) on E4XXX in order to work, like:
http://img127.imageshack.us/img127/6796/intel01ack3.jpg

make a jumper between G30 <-> J30 and isolate H30 ;)
or simply isolate H30 to have 266MHz, an E6600 with half cache.





/// EDIT
reading Intel's Xeon 5100 Series Datasheet I've found that on page 19:
http://img71.imageshack.us/img71/3898/intel07ib1.jpg

So, the reserved is for 333.33MHz, and it is H-L-L ;) . The pins are the same, G30.

---

So far, the Pin-out for my Pentium 4 630 is absoluteley the same for 266mhz FSB. Does that mean i just have to isolate H30 to get the FSB 1066 working on my Processor?

I dont see though this oO...

Quote:

---

Originally Posted by Kunaak

on my CPU no.
mine will do 3.4 ghz at stock voltage, and 3.6 ghz with 1.38 volts.
thats the tops I was able to easily keep stable in my P5B Deluxe.

same CPU tested on RD600 and 680i, and in all cases, the CPU doesnt need any extra voltage to reach 3 ghz.
my CPU's just a good one, but I also said many times...

make sure your CPU can do the stock multiplier x 333 at stock voltage, otherwise this mod would be useless to you.

---

somebody has tried with a e6700 ?

333 * 10 = 3330 mhz instead of 2667 ....

Wait a second. I think I'm getting a little rusty, but I recall when I implemented the dothan mod, I had to make a switch from high resistance to low, and did so by connecting to vss (ground point). I've never worked the other way, but always remembered high resistance being flagged by pulling a pin back in pga days or something like white out with the newer lga's.

Looking at the spec, 266 is LLL and 333 is HLL. Was it determined that connecting to vcc (power) will flag it to high resistance?!?! Shouldn't we be whiting out the land for bsel2 to make it high? Sorry if I'm sounding a little stupid. I'm a tad bit rusty

Hey Kunaak,

I never thought of shorting a VCC pin to the BSEL2 pin, back in August I attempted the same thing by insulating the BSEL2 pin on an E6400 to force the 1333 strap but unfortunately my DFI at the time didn't support the 1333 strap so I got no POST :p:

Well done! Awesome to see that it works :) Now to order and E4300 and have a play around.

Here's the thread I created on the subject:
http://www.xtremesystems.org/forums/...d.php?t=111510

Quote:

---

Originally Posted by ziddey

Shouldn't we be whiting out the land for bsel2 to make it high? Sorry if I'm sounding a little stupid. I'm a tad bit rusty

---

Theoretically yes, but I believe connecting the VCC pin means that current will flow into the bsel pad on the CPU, instead of from the bsel pad into the lga array, thereby insulating it. The bsel pad must be lower resistance than the corresponding pin then and what we're doing is making the pad high resistance.

I think :rolleyes: :confused:

Hmm, interesting. Here's a more complete list I've compiled:

BSEL2 BSEL1 BSEL0 FSB
L L L 266
L L H 133
L H H 166
L H L 200
H H L
H H H
H L H
H L L 333

Not sure if the intermediate bsel flags would still work since it's still largely bios dependent as to whether it can interpret it properly or not

Quote:

---

Originally Posted by cryptocat

So far, the Pin-out for my Pentium 4 630 is absoluteley the same for 266mhz FSB. Does that mean i just have to isolate H30 to get the FSB 1066 working on my Processor?

I dont see though this oO...

---

no, this was my mistake.. You can try isolating it, but i don't think it will work.
You have to make a jumper between H30 (BSEL 1) and L30 (VSS - ground)

make sure you are doing right, download and check Pentiums datasheet on intel's website ;)

Erh... why L30? thats like 3 Pins away from the BSEL ones....

And by jumper you mean, connecting those 2 pins with silverpaint?

Quote:

---

Originally Posted by Kunaak

Asus 1333 FSB Bios and Board list.

---

I cannot seem to find this 1901 bios for the P5W-DH anywhere.
I'll give it a go with my 6700.

Anyone?

Nice find and good work Kunaak

Quote:

---

Originally Posted by cryptocat

Erh... why L30? thats like 3 Pins away from the BSEL ones....

And by jumper you mean, connecting those 2 pins with silverpaint?

---

if you don't know what are you doing, don't do it

my advise for ya is not to do it or learn more b4 doing it ;)

Does anybody know a decent place to get an explanation of BSEL and all this other stuff you guys are talking about?

Quote:

---

Originally Posted by MotF Bane

Does anybody know a decent place to get an explanation of BSEL and all this other stuff you guys are talking about?

---

Go to intels website and look at the white sheets for the processors.

Motf bane, google up xeon pinmodding. Most of what you'd find would be about old stuff back when they were pga but there should be some good explanations. I wrote a decent article three years ago but the server went mia unfortunately. Google still has a cache of the writeup although the pictures (useful) are gone. You can find it at http://209.85.165.104/search?q=cache...lnk&cd=3&gl=us
It takes a long time to load for some reason, but I actually had to reference the cache today myself since I got a little rusty. Hope it helps.

Basically, bsel is a set of flags defined on the cpu that the motherboard reads to know what fsb to set as default. Input is measured in terms of high or low resistance. Thus, in the traditional sense, to achieve a high resistance from something that's originally low, you can use white out or nail polish or tape to cover up a land. In the pga days, it's be nail polish to insulate or just yanking a pin.

To make a high resistance into one of low resistance, one can either link it to a bsel of low resistance, or link it to a ground (vss).

It looks like from this technique as posted by op, it's been determined that you can link up a bsel that's originally flagged as low to a vcc (power) to bring it to high. I'm still thinking of the phenomena behind this myself, but it looks like it'd essentially be causing a relative resistance as seen by the motherboard to be high and thus be the same as isolation of the land/pin.

Hope this helps :)

how about testing this on one of the mATX boards where the pci-e lock breaks past 300.

would LOVE!!!!!! to see this work!

The problem is that the board itself has to have the bios to support the bsel flags. Otherwise, it'll either still boot at 266 or worse, not post at all.

Quote:

---

Originally Posted by ziddey

The problem is that the board itself has to have the bios to support the bsel flags. Otherwise, it'll either still boot at 266 or worse, not post at all.

---

I tried e6400 with p5b dlx, it says bios 1004 supports it, but no post.