kt7a and multis >13 with wiretrick?

Archived from groups: alt.comp.periphs.mainboard.abit (More info?)

On Wed, 26 May 2004 18:50:14 +0200, Hayno wrote:

> has anybody with a kt7a tried to do this wiretrick to get multipliers
> above 13?
> http://www.ocforums.com/showthread.php?s=c8f573747b2926eb4a5c6fa494249b02&threadid=279392
> i heard, every multi in bios is mapped to an multi above 12.5.
> i did the trick with pins al25 and am25 connected and got 10 boots 14,
> but there's no 14.5 or 15 (just 19 and 20).
> maybe Wes?
>
The link above isn't working for me right now, but I assume it pulls AJ 27
high, as that bit needs to be high on the cpu to access multipliers above
12.5. Once you pulled it high or set it high on the cpu then you can use
the Cross reference in the link below to see what the new multiplier
values are, I'm not sure what you did with AL25 and AM25, but see notes in
cross reference.

--
Abit KT7-Raid (KT133) Tbred B core CPU @2400MHz (24x100FSB)
http://mysite.verizon.net/res0exft/cpu.htm
 
Archived from groups: alt.comp.periphs.mainboard.abit (More info?)

thanks for your reply.
i know your page very good and for years! 😉

and sorry, i meant al25 and am24.
it's a pitty candjacs page isn't online anymore, i got the wiretrick
with these 2 socket holes from there.
http://home.tu-clausthal.de/~weeh/wiretrk.jpeg
and it was the only working trick - no solding resistors and cutting
bridges else helped.

the oc-forum is online again, could you have a look again please?
http://www.ocforums.com/showthread.php?threadid=279392
Wes Newell schrieb:
> On Wed, 26 May 2004 18:50:14 +0200, Hayno wrote:
>
>
>>has anybody with a kt7a tried to do this wiretrick to get multipliers
>>above 13?
>>http://www.ocforums.com/showthread.php?s=c8f573747b2926eb4a5c6fa494249b02&threadid=279392
>>i heard, every multi in bios is mapped to an multi above 12.5.
>>i did the trick with pins al25 and am25 connected and got 10 boots 14,
>>but there's no 14.5 or 15 (just 19 and 20).
>>maybe Wes?
>>
>
> The link above isn't working for me right now, but I assume it pulls AJ 27
> high, as that bit needs to be high on the cpu to access multipliers above
> 12.5. Once you pulled it high or set it high on the cpu then you can use
> the Cross reference in the link below to see what the new multiplier
> values are, I'm not sure what you did with AL25 and AM25, but see notes in
> cross reference.
>
 
Archived from groups: alt.comp.periphs.mainboard.abit (More info?)

Wes Newell schrieb:
> On Wed, 26 May 2004 18:50:14 +0200, Hayno wrote:
>
>
>>has anybody with a kt7a tried to do this wiretrick to get multipliers
>>above 13?
>>http://www.ocforums.com/showthread.php?s=c8f573747b2926eb4a5c6fa494249b02&threadid=279392
>>i heard, every multi in bios is mapped to an multi above 12.5.
>>i did the trick with pins al25 and am25 connected and got 10 boots 14,
>>but there's no 14.5 or 15 (just 19 and 20).
>>maybe Wes?
>>
>
> The link above isn't working for me right now, but I assume it pulls AJ 27
> high, as that bit needs to be high on the cpu to access multipliers above
> 12.5. Once you pulled it high or set it high on the cpu then you can use
> the Cross reference in the link below to see what the new multiplier
> values are, I'm not sure what you did with AL25 and AM25, but see notes in
> cross reference.
>

thanks for your reply.
i know your page very good and for years!

and sorry, i meant al25 and am24.
it's a pitty candjacs page isn't online anymore, i got the wiretrick
with these 2 socket holes from there.
http://home.tu-clausthal.de/~weeh/wiretrk.jpeg
and it was the only working trick - no soldering resistors and cutting
bridges else helped.

the oc-forum is online again, could you have a look again please?
http://www.ocforums.com/showthread.php?threadid=279392
obviously it is for unlocked mobiles, but maybe this wire may work for
any barton or thoroughbred?
 
Archived from groups: alt.comp.periphs.mainboard.abit (More info?)

On Thu, 27 May 2004 18:58:02 +0200, Hayno wrote:

> Wes Newell schrieb:
>> On Wed, 26 May 2004 18:50:14 +0200, Hayno wrote:
>>
>>
>>>has anybody with a kt7a tried to do this wiretrick to get multipliers
>>>above 13?
>>>http://www.ocforums.com/showthread.php?s=c8f573747b2926eb4a5c6fa494249b02&threadid=279392
>>>i heard, every multi in bios is mapped to an multi above 12.5.
>>>i did the trick with pins al25 and am25 connected and got 10 boots 14,
>>>but there's no 14.5 or 15 (just 19 and 20).
>>>maybe Wes?
>>>
>>
>> The link above isn't working for me right now, but I assume it pulls AJ 27
>> high, as that bit needs to be high on the cpu to access multipliers above
>> 12.5. Once you pulled it high or set it high on the cpu then you can use
>> the Cross reference in the link below to see what the new multiplier
>> values are, I'm not sure what you did with AL25 and AM25, but see notes in
>> cross reference.
>>
>
> thanks for your reply.
> i know your page very good and for years!
>
> and sorry, i meant al25 and am24.
> it's a pitty candjacs page isn't online anymore, i got the wiretrick
> with these 2 socket holes from there.
> http://home.tu-clausthal.de/~weeh/wiretrk.jpeg
> and it was the only working trick - no soldering resistors and cutting
> bridges else helped.
>
> the oc-forum is online again, could you have a look again please?
> http://www.ocforums.com/showthread.php?threadid=279392
> obviously it is for unlocked mobiles, but maybe this wire may work for
> any barton or thoroughbred?

OCforums is currently down for maintenance. 🙂

I guess I don't know what you are trying to do. Pulling AL25 low to AM24
basically disables the 4x multiplier bit thus changing the cross reference
up completely, assuming you had the 8x multiplier bit high.


--
Abit KT7-Raid (KT133) Tbred B core CPU @2400MHz (24x100FSB)
http://mysite.verizon.net/res0exft/cpu.htm
 
Archived from groups: alt.comp.periphs.mainboard.abit (More info?)

Wes Newell schrieb:
> On Thu, 27 May 2004 18:58:02 +0200, Hayno wrote:
>
>
>>Wes Newell schrieb:
>>
>>>On Wed, 26 May 2004 18:50:14 +0200, Hayno wrote:
>>>
>>>
>>>
>>>>has anybody with a kt7a tried to do this wiretrick to get multipliers
>>>>above 13?
>>>>http://www.ocforums.com/showthread.php?s=c8f573747b2926eb4a5c6fa494249b02&threadid=279392
>>>>i heard, every multi in bios is mapped to an multi above 12.5.
>>>>i did the trick with pins al25 and am25 connected and got 10 boots 14,
>>>>but there's no 14.5 or 15 (just 19 and 20).
>>>>maybe Wes?
>>>>
>>>
>>>The link above isn't working for me right now, but I assume it pulls AJ 27
>>>high, as that bit needs to be high on the cpu to access multipliers above
>>>12.5. Once you pulled it high or set it high on the cpu then you can use
>>>the Cross reference in the link below to see what the new multiplier
>>>values are, I'm not sure what you did with AL25 and AM25, but see notes in
>>>cross reference.
>>>
>>
>>thanks for your reply.
>>i know your page very good and for years!
>>
>>and sorry, i meant al25 and am24.
>>it's a pitty candjacs page isn't online anymore, i got the wiretrick
>>with these 2 socket holes from there.
>>http://home.tu-clausthal.de/~weeh/wiretrk.jpeg
>>and it was the only working trick - no soldering resistors and cutting
>>bridges else helped.
>>
>>the oc-forum is online again, could you have a look again please?
>>http://www.ocforums.com/showthread.php?threadid=279392
>>obviously it is for unlocked mobiles, but maybe this wire may work for
>>any barton or thoroughbred?
>
>
> OCforums is currently down for maintenance. 🙂
>
> I guess I don't know what you are trying to do. Pulling AL25 low to AM24
> basically disables the 4x multiplier bit thus changing the cross reference
> up completely, assuming you had the 8x multiplier bit high.
>
>

okay, let's see:
i have got a 1700+ jiucb and opened the 5th L3-bridge (8x bit low>hi),
to acces higher multipliers above 13.
the cpu than bootet with 20*100, which is not acceptable (known problem).
i than did the trick with al25 and am24 to get mapped multipliers.
what i got was:
7.5 boots 19
8.5 boots 20
9.0 boots 13
9.5 boots 13.5
10.0 boots 14
10.5 boots 21
all others don't boot because of mobile or reserved settings or other
strange behavior.
what i want to get is a multiplier remap to 14.5 and 15 to take the cpu
to the 2ghz-limit and beyond.
atm the processor runs @1.55v and 14*133 (1866mhz).

the socket-mod at ocforums (online again 😉) is quite different from
what i used.
do you know that mod or do you know(!) how to get multipliers between 14
and 19 to work?

thanks again...
 
Archived from groups: alt.comp.periphs.mainboard.abit (More info?)

On Fri, 28 May 2004 17:16:56 +0200, Hayno wrote:

> okay, let's see:
> i have got a 1700+ jiucb and opened the 5th L3-bridge (8x bit low>hi),
> to acces higher multipliers above 13.
> the cpu than bootet with 20*100, which is not acceptable (known problem).
> i than did the trick with al25 and am24 to get mapped multipliers.
> what i got was:
> 7.5 boots 19
> 8.5 boots 20
> 9.0 boots 13
> 9.5 boots 13.5
> 10.0 boots 14
> 10.5 boots 21
> all others don't boot because of mobile or reserved settings or other
> strange behavior.
> what i want to get is a multiplier remap to 14.5 and 15 to take the cpu
> to the 2ghz-limit and beyond.
> atm the processor runs @1.55v and 14*133 (1866mhz).
>
> the socket-mod at ocforums (online again 😉) is quite different from
> what i used.
> do you know that mod or do you know(!) how to get multipliers between 14
> and 19 to work?
>
If you cut L3-5 and L3-4 (setting default to 15x) that will give you
access to multipliers from 15-24 without a wire mod. 9 (17x) will work
with a 133MHz FSb, but that's the only one. That's the standard cross
reference. Most can get it to run with the 9 setting at 2266MHz with a
little more vcore. If that doesn't work for you, you can either follow
the notes in the cross reference guide to get other multipliers that will
work with 133MHz FSB, OR, you can set the cpu multiplier manually using
the L3 bridges to whatever multiplier you want and then lock it by cutting
the L1 bridges. Locking the cpu allows any FSB setting up to whatever your
board will do. Since you're probably limited to under 145MHz for the FSB,
I'd use either the default 15 or 16 as a locked setting. You can determine
how fast your cpu will run using the default unlocked setting before
choosing a permanant locked setting. A 16 would probably be the best. You
can get 17 by default, and you probably won't have much luck with anything
higher than that with a 133Mhz FSB that you couldn't get with the default
17 and raising the FSB a little (17x145.45=2472MHz). AFAIK, there is no
way to get 14-19 to work at the same time with a 133MHz FSB without
changing the mod around. Hope this helps. It's about all I can think of
right now.

--
Abit KT7-Raid (KT133) Tbred B core CPU @2400MHz (24x100FSB)
http://mysite.verizon.net/res0exft/cpu.htm
 
Archived from groups: alt.comp.periphs.mainboard.abit (More info?)

Wes Newell schrieb:
> On Fri, 28 May 2004 17:16:56 +0200, Hayno wrote:
>
>
>>okay, let's see:
>>i have got a 1700+ jiucb and opened the 5th L3-bridge (8x bit low>hi),
>>to acces higher multipliers above 13.
>>the cpu than bootet with 20*100, which is not acceptable (known problem).
>>i than did the trick with al25 and am24 to get mapped multipliers.
>>what i got was:
>>7.5 boots 19
>>8.5 boots 20
>>9.0 boots 13
>>9.5 boots 13.5
>>10.0 boots 14
>>10.5 boots 21
>>all others don't boot because of mobile or reserved settings or other
>>strange behavior.
>>what i want to get is a multiplier remap to 14.5 and 15 to take the cpu
>>to the 2ghz-limit and beyond.
>>atm the processor runs @1.55v and 14*133 (1866mhz).
>>
>>the socket-mod at ocforums (online again 😉) is quite different from
>>what i used.
>>do you know that mod or do you know(!) how to get multipliers between 14
>>and 19 to work?
>>
>
> If you cut L3-5 and L3-4 (setting default to 15x) that will give you
> access to multipliers from 15-24 without a wire mod. 9 (17x) will work
> with a 133MHz FSb, but that's the only one. That's the standard cross
> reference. Most can get it to run with the 9 setting at 2266MHz with a
> little more vcore. If that doesn't work for you, you can either follow
> the notes in the cross reference guide to get other multipliers that will
> work with 133MHz FSB, OR, you can set the cpu multiplier manually using
> the L3 bridges to whatever multiplier you want and then lock it by cutting
> the L1 bridges. Locking the cpu allows any FSB setting up to whatever your
> board will do. Since you're probably limited to under 145MHz for the FSB,
> I'd use either the default 15 or 16 as a locked setting. You can determine
> how fast your cpu will run using the default unlocked setting before
> choosing a permanant locked setting. A 16 would probably be the best. You
> can get 17 by default, and you probably won't have much luck with anything
> higher than that with a 133Mhz FSB that you couldn't get with the default
> 17 and raising the FSB a little (17x145.45=2472MHz). AFAIK, there is no
> way to get 14-19 to work at the same time with a 133MHz FSB without
> changing the mod around. Hope this helps. It's about all I can think of
> right now.
>

okay, thanks again.
so all i can get is a changeable multiplier of 17 with fsb 133?
everything with fsb 100 ist not worth a thought...
i don't want a fixed multiplier because of the fixed power consumption.
if there is no way to get easily changeable multipliers of (let's say)
14 AND 15 AND 16 with fsb 133 i am not willing to take the risk of
cutting any bridges again.
at the moment i'm very happy with the stability and if there's no proven
workaround to go with changeable multipliers i will stay at 1866mhz,
which is quite good for the old infrastructure and way better than
anything like 20*100.
thank you very much for your effort and thoughts!
 
Archived from groups: alt.comp.periphs.mainboard.abit (More info?)

On Sat, 29 May 2004 09:17:29 +0200, Hayno wrote:

> Wes Newell schrieb:
>> On Fri, 28 May 2004 17:16:56 +0200, Hayno wrote:
>>
>>
>>>okay, let's see:
>>>i have got a 1700+ jiucb and opened the 5th L3-bridge (8x bit low>hi),
>>>to acces higher multipliers above 13.
>>>the cpu than bootet with 20*100, which is not acceptable (known problem).
>>>i than did the trick with al25 and am24 to get mapped multipliers.
>>>what i got was:
>>>7.5 boots 19
>>>8.5 boots 20
>>>9.0 boots 13
>>>9.5 boots 13.5
>>>10.0 boots 14
>>>10.5 boots 21
>>>all others don't boot because of mobile or reserved settings or other
>>>strange behavior.
>>>what i want to get is a multiplier remap to 14.5 and 15 to take the cpu
>>>to the 2ghz-limit and beyond.
>>>atm the processor runs @1.55v and 14*133 (1866mhz).
>>>
>>>the socket-mod at ocforums (online again 😉) is quite different from
>>>what i used.
>>>do you know that mod or do you know(!) how to get multipliers between 14
>>>and 19 to work?
>>>
>>
>> If you cut L3-5 and L3-4 (setting default to 15x) that will give you
>> access to multipliers from 15-24 without a wire mod. 9 (17x) will work
>> with a 133MHz FSb, but that's the only one. That's the standard cross
>> reference. Most can get it to run with the 9 setting at 2266MHz with a
>> little more vcore. If that doesn't work for you, you can either follow
>> the notes in the cross reference guide to get other multipliers that will
>> work with 133MHz FSB, OR, you can set the cpu multiplier manually using
>> the L3 bridges to whatever multiplier you want and then lock it by cutting
>> the L1 bridges. Locking the cpu allows any FSB setting up to whatever your
>> board will do. Since you're probably limited to under 145MHz for the FSB,
>> I'd use either the default 15 or 16 as a locked setting. You can determine
>> how fast your cpu will run using the default unlocked setting before
>> choosing a permanant locked setting. A 16 would probably be the best. You
>> can get 17 by default, and you probably won't have much luck with anything
>> higher than that with a 133Mhz FSB that you couldn't get with the default
>> 17 and raising the FSB a little (17x145.45=2472MHz). AFAIK, there is no
>> way to get 14-19 to work at the same time with a 133MHz FSB without
>> changing the mod around. Hope this helps. It's about all I can think of
>> right now.
>>
>
> okay, thanks again.
> so all i can get is a changeable multiplier of 17 with fsb 133?

I'm not sure what you mean here, but there's 3 choice of mods for
multipliers that work with a 133MHz FSB. Copied from website:

Just ground one of the multiplier bits from the socket (3rd =2x, 4th=4x or
5th=8x). This way I was able to boot with fsb 133 and up.

3rd pulled LO at least multipliers 15-16.5 works with fsb 133

4th pulled LO multis 13-14 works with fsb 133

5th pulled LO multis 5-12.5 works with fsb 133

The bit you want to ground depends how far your memory can take the fsb. If
you cannot complete the boot sequence with 17x133 then try lower multi with
higher fsb (not 2.26Ghz but still faster).

> everything with fsb 100 ist not worth a thought...

I don't know why not. There's only about a 3% performance hit between 133
and 100.

> i don't want a fixed multiplier because of the fixed power consumption.

Huh? Multiplier doesn't control power consumption, Vcore and CPU speed
do. With a 17 multiplier at 133MHz as default, you can just lower vcore
and FSB to 100 to conserve power.

> if there is no way to get easily changeable multipliers of (let's say)
> 14 AND 15 AND 16 with fsb 133

See above.

> i will stay at 1866mhz, which is quite good for the old
> infrastructure and way better than anything like 20*100.

2000Mhz on a 100MHz FSB will outperform 1866MHz on a 133MHz FSB in
everything except memory benchmark test maybe, which are of no real value
since 90% of memory fetches are serviced by the cpu cache.

--
Abit KT7-Raid (KT133) Tbred B core CPU @2400MHz (24x100FSB)
http://mysite.verizon.net/res0exft/cpu.htm
 
Archived from groups: alt.comp.periphs.mainboard.abit (More info?)

Wes Newell schrieb:
> I'm not sure what you mean here, but there's 3 choice of mods for
> multipliers that work with a 133MHz FSB. Copied from website:
>
> Just ground one of the multiplier bits from the socket (3rd =2x, 4th=4x or
> 5th=8x). This way I was able to boot with fsb 133 and up.
>
> 3rd pulled LO at least multipliers 15-16.5 works with fsb 133
>
> 4th pulled LO multis 13-14 works with fsb 133
>
> 5th pulled LO multis 5-12.5 works with fsb 133
>
> There's only about a 3% performance hit between 133
> and 100.
>
> Multiplier doesn't control power consumption, Vcore and CPU speed
> do. With a 17 multiplier at 133MHz as default, you can just lower vcore
> and FSB to 100 to conserve power.

thanks again!
maybe i'm a little bit more clever now.
to use your words: what i did was to set the 4th multiplier bit to LO,
because i get 13 ... 14 with fsb 133. and i did it by inserting a wire
into al25 and nearby ground am24. right?
if i want a multiplier of 15 and 16 with fsb 133 i have to set the 3rd
multiplier bit to LO. what kind of wire u-loop i have to insert into the
socket, which holes i have to use? do you know that? maybe there was a
discription on candjacs page that is gone.

the difference between fsb 100 and 133 i notice with every click i do -
the whole behavior icluding editing grafics, handling videos and work
with text or tables. i don't want to have fsb 100 back again (had it
long enough with my elder kt7).

in fact multipliers don't control any power consumption, but when i need
most available power i set the highest multiplier with elevated vcore -
when i don't need high power i go back to low multi and lowered vcore
(plus can stay at a acceptable memory bandwidth 😉 )
best thing would be to have multiplier 14 (with vcore 1.55v) and 15
(with at least 1.65v), because there is silence in the case with little
air cooling i want to stay with.
but in the next available free time i'm gonna pull the 3rd multiplier
bit to LO, if you (or anybody else) can exactly describe where to insert
the wire.
very curious...
 
Archived from groups: alt.comp.periphs.mainboard.abit (More info?)

On Sat, 29 May 2004 22:34:44 +0200, Hayno wrote:

> to use your words: what i did was to set the 4th multiplier bit to LO,
> because i get 13 ... 14 with fsb 133. and i did it by inserting a wire
> into al25 and nearby ground am24. right?

yes.

> if i want a multiplier of 15 and 16 with fsb 133 i have to set the 3rd
> multiplier bit to LO. what kind of wire u-loop i have to insert into the
> socket, which holes i have to use? do you know that? maybe there was a
> discription on candjacs page that is gone.
>
Try this. It shows it pretty clear.

http://fab51.fc2web.com/pc/barton/athlon-e20.html

> but in the next available free time i'm gonna pull the 3rd multiplier
> bit to LO, if you (or anybody else) can exactly describe where to insert
> the wire.

Between ground and AN25. If you really want easy access to 13-16.5 then
wire AN25, AL25, and Ground to a SPDT switch and select which you want
with the toggle switch. You'd solder the wires to the cpu socket pads on
the back of the MB. Ground to center of switch of course. Use a center off
switch if you want it back to normal. Use your imagination. Many things
are possible.

--
Abit KT7-Raid (KT133) Tbred B core CPU @2400MHz (24x100FSB)
http://mysite.verizon.net/res0exft/cpu.htm
 
Archived from groups: alt.comp.periphs.mainboard.abit (More info?)

Wes Newell schrieb:

> Try this. It shows it pretty clear.
>
> http://fab51.fc2web.com/pc/barton/athlon-e20.html
>
> Between ground and AN25. If you really want easy access to 13-16.5 then
> wire AN25, AL25, and Ground to a SPDT switch and select which you want
> with the toggle switch. You'd solder the wires to the cpu socket pads on
> the back of the MB. Ground to center of switch of course. Use a center off
> switch if you want it back to normal. Use your imagination. Many things
> are possible.

word! it's getting brighter!
i saw the picture of the socket at that page you linked above and i
mirrored it to see the pin assignment in top view.
what is interesting now is the bp_fid (at the moment grounded al25,
which is the green pin numbered 3 in the picture).
what i should do to get multiplier 15 and 16 is to ground an25, which is
the green pin numbered 2 in the picture (at the edge of the socket and
in the same column with al25). right? hopefully i'm not confusing you.
i think i'm right until here.
one last question: what multiplier is mapped to what in that case?
sorry, if this is a little bit stupid, but my brain is in chaos now.
if that works i will think over a kind of switch.
i thank you again for bringing me forward...
 
Archived from groups: alt.comp.periphs.mainboard.abit (More info?)

On Mon, 31 May 2004 16:06:22 +0200, Hayno wrote:

> tried it and found out:
> grounded an25 gave 9 boots 15 (2000mhz with fsb 133) and 10 boots 16
> (2133mhz). bios was stable with good temperature.
> but windows didn't start - no matter what vcore (up to 1.75v) or cpu
> drive strength (0 to 5).
> in the end i'm back to 1866mhz @1.55v (grounded al25). took me a total
> of 4 hours, thanks to the alpha8045 and disassambling the whole pc 2 times.
> thanks again for replying.
>
> abit kt7a rev1.0
> thoroughbred b 1700+ jiucb
> 2*256mb infineon pc133 cl2

B cores with JIUCB steppings weren't known as real good overclockers, so
that may be the problem.

--
Abit KT7-Raid (KT133) Tbred B core CPU @2400MHz (24x100FSB)
http://mysite.verizon.net/res0exft/cpu.htm