Nothing past 12.5x on 8rda with xp2600

[candjac]

Originally posted by bowman1964
well best i can do at the moment.i have dig pics of my cpu at home .after i made this L3 cut the bios defaulted to 100 x 20 setting.i dont know how but it did.it identifies it as a xp2400

WHOA!!! You say you have a 2600/333, which has to be a Tbred which has ONLY 5 L3 Multiplier setting bridges...but you show a pic of a "Palomino" which has completely different Multiplier bridges, ie, 4 L3s, 4 L4s, and 2 L10s. What's going on??
Is that really what your 2600 looks like, or did you really just cut the "rightmost"/5th L3 of your Tbred...and just used the wrong picture from oc-inside.de to show us what you cut??...(the rightmost L3 in the pic)

Must clear this up before anything else. For reference,
http://www.beachlink.com/candjac/index.htm link to Multiplier Code article which has pic of Multiplier bridges for ALL socket A CPUs from Duron/Tbird thru Tbreds.
John C.

[muzz]

Thats not what it looks like ( I have the same chip) candjac..... that is the interactive guide they have available at ocinside.
There are 5 l3's on this chip ( as you know) and when looking at his pic there from top ( under where it says "L3") to bottom my stock chip is :
open
open
closed
closed
closed

I guess he just opened the "Bottom"(closest to the core), as the top ( closest to where it says "L3") is already open.

My guess is thats all he had to actually show what trace he cut.

[bowman1964]

yea i just pulled that off of ocinside.closest thing i could see

if you guys are nice.......i might take the chip back out one last time tonight and take a pic before i go back to benching.and take you a nice picture.because i dont know what all it did to the inside but i couldnt get this chip this high of speed before.heck if i know but it made this chip come to life.

[candjac]

Originally posted by muzz
Thats not what it looks like ( I have the same chip) candjac..... that is the interactive guide they have available at ocinside.
There are 5 l3's on this chip ( as you know) and when looking at his pic there from top ( under where it says "L3") to bottom my stock chip is :
open
open
closed
closed
closed

I guess he just opened the "Bottom"(closest to the core), as the top ( closest to where it says "L3") is already open.

My guess is thats all he had to actually show what trace he cut.

Muzz, Bowman,
OK, let's assume Bowman did as Muzz says...then...
First let's rotate the above to get it L to R.
L3 O-O-C-C-C = 12.5X = default for 2600 12.5X166
L3 O-O-C-C-O = above plus 5th L3 cut by Bowman = a "Mobile" 4.5X setting listed = "Reserved" by AMD. And we can now analyze "this".

O-O-C-C-O = 4.5X is one of (4) 3.0X thru 4.5X settings originally designed for very low frequency operation, along with reduced Vcore, when system goes into the battery saving mode of the "Power Now" function. 3X and/or 4X are actually used, but 3.5X and 4.5X are listed as 'Reserved".

But just recently "Soulkeeper" posted at Anand's that he was able to get his 1700 Tbred to run at 20X100 by setting the 1st 4 bits to 12.5X from the mobo "with the 5th L3 OPEN"...thats O-O-C-C-O same as what we're all assuming Bowman did above. Now do you see why we kept insisting on "detail"?? Bowman may have just "confirmed" Soulkeeper's result!!! There may no longer be a mystery about Bowman's 20X.

In fact Soulkeeper was able to get 5 of the 6 "Reserved" settings in AMD's Multiplier Code to boot at 19X, 20X, 22X, 23X, 24X. Bet the missing one is designed to boot at 21X. Which says that AMD must have"Remapped" these "high" Multipliers to "Reserved" settings, but as usual did NOT document them...to discourage discovery and oc'g??

We do have some questions...How was the 100MHz attained??...automatically by system or did Bowman set 100Mhz FSB in bios. Also, was Multiplier option in bios on "Auto" which would not interfere with the O-O-C-C-O setting...or was it in user-defined option set to 12.5X which would send HI-HI-LO-LO 1st 4 bits/signals to the Multiplier, which would "not change" what the L3 bridges had set.

We can post all 5 of Soulkeepers discoveries re Multipliers > 18X if you want. You can also see a Table of bridge settings from 3X up at our beachlink.com site, link to Tbred article. Use it to verify all the listings above.

We're so groggy with the twists and turns here, not sure if all issues/questions covered...ask away if not.

And btw, Jens at oc-inside.de has another "Tbred Guide" with just the 5 L3s...there's a link to it, use it and not the Palomino.
John C.

PS:- Bowman you could verify everything we and Muzz "assumed" you did.

[bowman1964]

this is my actual chip i found it in a pic i took of my motherboard the other night.tell me if this is enought to see what you need to candjac. let me know soon as i will be going home in a hour or so.if not clear enought i will pull it out for you tonight

P.S. the 100mhz x 20 was set a default of the bios.clear cmos and thats what i get xp2400 running at 100x20 .one small thing though,,, i always after i clear my cmos i go to load optimised defaults and save exit and then start my bios tweaking.when i tried that last night after cutting the chip i get the no go alarm from the bios.have to clear cmos and just go to tweaking.

[candjac]

Originally posted by bowman1964
this is my actual chip i found it in a pic i took of my motherboard the other night.tell me if this is enought to see what you need to candjac. let me know soon as i will be going home in a hour or so.if not clear enought i will pull it out for you tonight

P.S. the 100mhz x 20 was set a default of the bios.clear cmos and thats what i get xp2400 running at 100x20 .one small thing though,,, i always after i clear my cmos i go to load optimised defaults and save exit and then start my bios tweaking.when i tried that last night after cutting the chip i get the no go alarm from the bios.have to clear cmos and just go to tweaking.

Don't pull it out any more...you cut the 5th L3 as Muzz said and as we hoped. So you have "confirmed" that 19X and up Multipliers "can be had" (on Tbreds) by executing settings which are listed as "Reserved" in AMD's Tbable of Settings in Mobile datasheet, the only such listing we know of...but should apply to all socket A CPUs. How does it feel to be like John Cabot, who came here very early but not 1st like Columbus.;-)

OK so 100Mhz FSB was set in bios...but did bios also have the 20X setting option???...cause Soulkeeper got his 20X by setting 12.5X with the 5th L3 open. Can you confirm what was actually displayed in bios??

And why did you also cut the 2nd L12?? Did someone explain why you should have done that??
We're all still learning from these various experiments.
Thanks, John C

[bowman1964]

Originally posted by candjac

And why did you also cut the 2nd L12?? Did someone explain why you should have done that??
We're all still learning from these various experiments.
Thanks, John C

well i cut the l12 to allow me to set the cpu to a 133 bus instead of a 166 bus.if or for some reason at lower setting(133) you can clock higher on the fsb with the nforce chipset set.i think the chipset sets how aggressive the timmings are by if it is running a defaut 166 bus or a 133 bus.i couldnt go over 200 fsb untill i cut that bridge. once i did that it opened up for me.the board has a jumper that do alot off nothing for me on my board with this xp2600 chip.

i will look tonight to see if 20x is available for sure

[bowman1964]

i tell you i have gotten so confused back and forth over this chip i am going to pull it off tonight for sure guys so just wait before you cut your chips.i will confirm with a pic.i am getting old enought to start second guessing my self.let me get that pic i will be back on tonight ok later guys

[bowman1964]

well here you go, i cannt do any better than this.and i did point you to the wrong L12 look at the pic you can see where i cut it.i told you i better look.but L3 is like i said.
this took me a bunch of work to get posted i hope it helps.
i like to be sure.no other way to be surer than this

[muzz]

I didn't see your previous pics Bowman.. I've been working over my new RDA+........
Yes I noticed scrolling through your pics that there was a mistake in the L12 pic... def. the 3rd one over(L-R).

Thanks for the pics, confirmation is always a good thing.

BTW I believe I have the EXACT same chip as that 1 in your pic.
I know its a "45" and that it is DKV3D . oooops maybe a "C" at the end though... I'll have to look later.

[PimpJack]

bowman: I ran into trouble running 210fsb at 11.5 (still waiting for pump and pelt so a backup volcano+ is it) I took it down to 10 and then 10.5 and ofcourse it became stable. DO any of your mods help out this problem. And u did get 13x correct? I won't need 20x at 200+fsb LOL. Since I will be pulling my system apart I might aswell shred my athlon to pieces. What is the best tool to cut those bridges? nF is my first AMD since I started overclocking (learned on p4's hehe) so I have no experiance with cutting bridges.
Did the jumper make ur chip become a 2400+. A while back i recall boot up saying it was 2400+ but i forget the settings.

Thanx. Muzz told me about this mod a while back, but i didn't have it in me to trust that guy that claimed it.

[muzz]

I just did it to my 2600+...... it works MINTY!!!!!!!!!!
I have only tried 13 +15, but they both work great, and I see no problems with the lower multis AFAICT ...

Boots up as a 2400 10x200

PJ there are alot of ways to remove the jumper ( trace) some say to use a battery or your 5 volt rail to ground. I don't like the idea of pumping voltage through my chip. so I took a razor knife and just ran it over the trace a couple times carefully with hardly any pressure and it was gone. ( can hardly tell- but I have a magnifying light so that helps).

Just be careful, especially digging down ( layers) and of course adjoining bridges........
Piece of cake, that took less than 1 minute, and the good thing is it is an outside trace.... the easiest to redo if ya ever want to.... which is nice...

[muzz]

BTW Thanks for being the guinea pig Bowman..........

[PimpJack]

Muzz: Do u simple select 13x from the bios ore is there a different "formula" such as setting multiplier y will actually give u multi x ?

Good to hear its worth it. How about the L12? I think my board jumper is doing the trick.

[bowman1964]

Originally posted by muzz
BTW Thanks for being the guinea pig Bowman..........

ha ha guys yea i was the guinea pig.but i dont turn back.
and did you notice how i whent out of my way to get you a perfect picture to look at to go by.
but being the nice guy i am lol/i didnt mind sharing in the wealth....if it works as well as mine i am benching right now with prime at 2750 or so.....ha damm..............
by the way this is my best so far.absolutly no overclocking on the radeon while i test the rest of the system but just look at this benchfor a stock 325/320 radeon.

[muzz]

Looking good bowman.. get that 9700 cranked up to 470/360+ and your right up there.

I knew what trace you cut.... the other end is already cut.

Thanks,

muzz

[candjac]

Ok Bowman, that last pic showing the 3rd L12 cut calmed us down...now you did what everyone else did...including the 5th L3.

You wrote "i will look tonight to see if 20x is available for sure"...so we're still expecting to hear "where" you were able to "actually set" 20X.

And also curious as to "why" you went so high with the Multiplier which forced you so low with FSB. Understand the need to cut the 5th L3 in order to be able to get 13x and up, in fact we may have been the first to promote that, esp for older mobos with low FSB. But for best performance as you well know high FSB is best...and you mention that you cut the L12 to get over 200MHz FSB...and we know you can't run that cpu at 20X200. So what are you actually running at, and why??
Why not something like 12.5X200+??
Thanks, John C.

[muzz]

Originally posted by PimpJack
Muzz: Do u simple select 13x from the bios ore is there a different "formula" such as setting multiplier y will actually give u multi x ?

Good to hear its worth it. How about the L12? I think my board jumper is doing the trick.

I just set it to 13 in the bios. and fired it up. """""presto read all about it"""""".... heheheh not sure why I actually did it tbo......... at this point in time there really is no need for over 12.5 .... I guess I just wanted it is all.

As far as the L12...... I never did it with my last board....... as there was no need..... this one needs a little more help, it's not as good a clocker stock....... this one NEEDS mods, the other one didn't really. So I never did cut the L12.

[bowman1964]

Originally posted by candjac
Ok Bowman, that last pic showing the 3rd L12 cut calmed us down...now you did what everyone else did...including the 5th L3.

You wrote "i will look tonight to see if 20x is available for sure"...so we're still expecting to hear "where" you were able to "actually set" 20X.

And also curious as to "why" you went so high with the Multiplier which forced you so low with FSB. Understand the need to cut the 5th L3 in order to be able to get 13x and up, in fact we may have been the first to promote that, esp for older mobos with low FSB. But for best performance as you well know high FSB is best...and you mention that you cut the L12 to get over 200MHz FSB...and we know you can't run that cpu at 20X200. So what are you actually running at, and why??
Why not something like 12.5X200+??
Thanks, John C.

well i didnt have time to try and set the chip to 20x last night i did look and it is now avalible.but i will check it tonight.
took me a while last night to remove the chip and clean all the dielectric grease off of it and do the pics and all. and the only reason i needed the 13 x is i needed to test the chip's overclockablity.there is no way to isolate any memory problem or chip clocking problems if i cannt test the 2 seperate from each other.i hope you understand what i am trying to say?
and i dodnt know what it excatly it did to the chipset timing by going to the L12 and setting it to 133 but it does make a differance in how far you can push the memory.
my current setting is 12.5 x 219= 2750mhz or so.memory is benching around 3379 or so in sandra.i can run at 225 dual but running 225x 12 =2700 with 3540 sandra scores ,and my 3d is about the same as 219x12.5.but 219 x 12.5 is more stable.
have a ways to go this weekend.i need to burn in some this week and maybe this weekend crank up the video some

[candjac]

Originally posted by bowman1964
well i didnt have time to try and set the chip to 20x last night i did look and it is now avalible.but i will check it tonight.
The only reason i needed the 13 x is i needed to test the chip's overclockablity.there is no way to isolate any memory problem or chip clocking problems if i cannt test the 2 seperate from each other.i hope you understand what i am trying to say?

and i dodnt know what it excatly it did to the chipset timing by going to the L12 and setting it to 133 but it does make a differance in how far you can push the memory.
my current setting is 12.5 x 219= 2750mhz or so.

Yes, we understand "exactly" what you're trying to say, and we're impressed that you would go to such lengths (20X) to "isolate" the memory and CPU from each other. We just want to be sure of all the details and motivations, 'cause we're thinking of writing a new page at our site relating only to the various ways to "workaround" these darn mobos' limitations in accessing low and high Multiplier ranges, and esp the 19X and up which seem to Remap from settings listed By AMD as "Reserved". And your 20X100 experience should be a part of that. Just in case we can't find your mobo model later in this thread, what is it??

As for the L12 FSB_SENSE autodetect function, no one knows exactly why memory can be pushed more with that code set at 133, but as you say it does. Most people think that somewhere in the boot process some memory timings/wait states?? get set more agressively, but doesn't seem to make sense since you would think those timings would be set as a function of "actual" FSB. But maybe that's not the case.

Nice OC, and it all makes "sense" now. Thanks for clarifying.
John C.