Using Bartons, Tbreds and Palominos with the A7M266

[Senor Panadero]

Originally posted by The Tinkerer
Panadero and Trats- I agree. Trats, since you are the resident expert on the newest cpus, maybe you can start a new thread with just the relevant information for A7M266 owners for doing all types of cpus. We could delete my overdone explanation for the 2200+ A Tbred, since that one is probably almost non-existent. I believe that folks will only find Tbred Bs, Palominos, and Bartons?? IN that case, your wire trick is really all we need to know.

Anyway, if you want to do it, go for it. This thread is really too long now. We could ask the moderators to permanently close this thread, and have the last post include a link to the new thread and really clean things up. Are you up for it Trats???

We need a good title, something like "Using Bartons, Tbreds and Palominos with the A7M266".

This post continues the thread begun here and ending here. See the beginning post in that thread that contains the updates achieved there.

[The Tinkerer]

Our understanding is that all Tbred and Barton cpus must have bios 1008.02b. Bios 1007 and below are solely for Palomino cpus.

UPDATE: Here's a thread of a member who got an XP2600+ to run at XP2700 levels using the wire trick!!

http://www.amdforums.com/showthread....hreadid=265219

Special Note regarding XP2200+ Tbred Model A (680). Without the wire trick, you must set your voltage (VID 1-4) to 1.8 or 1.85 volts to get it to boot. The 2200 Tbred A (680 family, model, stepping) is a bit quirky and some members were unable to get it to work. We suspect the excess voltage is necessary to get the cpu id sequence to work. I do not recommend using that particular cpu due to the varying results. I am using it, however, in my A7M266 rig. In the alternative, you can try the wire trick shown below to set the necessary multiplier setting high and low, and leave your cpu voltage VID 1-4 at default. If you are able to do this, please post your results for confirmation.


Here's the sequence recommended for replacing the cpu SEE ALSO THE NEXT POST BY TRATS FOR THE "WIRE TRICK" DETAILS FOR TBRED "B" AND BARTON MODELS

1. flash bios to 1008.02b using your old CPU

2. restart the machine and change the bios setting to following:
Load Setup Defaults
Change S2K slew rate to 1
Save changes and exit bios

3. turn off the computer, pull the plug out, and change cpu
Make the appropriate wire trick change for the multiplier you need if the cpu default is not wanted or will not work with the mobo. See the next post by Trats (below) for the specifics on the wire trick. Remember, the mobo does not offer multiplier adjustments

4. You can leave the cpu jumper setting to default (VID 1-4) and the FSB speed to default to allow fsb increases in bios.

5. connect the power and start the computer.

6. If the computer does not boot, try clearing cmos by shorting the contacts on the mobo below the battery (clrtc)

[trats]

to run tbred A 2200, see Tinkerer's post above and here
here's the wire trick info for using tbred B and bartons with a7m266. follow steps 1-4, either 4a-4d depending on your cpu, and then step 5.

note that if your xp cpu is manufactured after week 39 or so in 2003, it is likely that the multi is locked, and such a cpu doesn't require any wire tricks to run at the correct multi on a7m . the ideal CPU for a7m in this case is a 133fsb locked xp 2400+ which runs at 15x.

all xp-m are unlocked so far, and require wire tricks (or software, see 4d below) to run at the correct speed and higher.

a bit of background. before using the wire trick, one needs to know the L3 and starting multi that the unlocked chip is running on the system and the target multi is obtained by connecting an27 (0.5x bit), al27 (1x), an25 (2x), al25 (4x), or aj27 (8x) to either vcc (am26, ak26 or ah26) or ground (am28, ak28, ah28, am24, ak24 or ah24).

john carcich has excellent infos on his site about multis etc.

i've listed different wire connections in steps 4a-4d below for unlocked chips; these connections are essentially variations on the same theme, where for a given multi, small changes (connections added or removed) were made based on L3s and/or default post multis that many a7m users have reported in this forum. for example, the 14x setting in 4d (an25-am26 2x high, and aj27-ah26 8x high) for the xp-m 2400 would give 13x with an unlocked xp 2500 on a7m, as shown in 4a, 13x. although both xp-m 2400 and xp 2500 post as 11x on a7m without wires, the 14x/13x discrepancy occurs because the xp-m has a 6x L3 vs. 11x L3 for the xp 2500. to run 13x on the xp-m 2400, an al27-ak28 (1x low) connection was added to 4d, 13x.

one could use the connections listed in 4a-4d for any unlock chip, with the end results of either no post, different multi, or correct multi or no change.

(1) flash bios to 1008.02b with aflash, and set slew rate 1 (thanks to uk_bloke for reminding)

(2) get a thin wire, e.g., from an 80-conductor ide cable or phone wire. the latter is best with three or more wire connections.

(3) bend wire to form a U-shape, 6mm long, 1.5mm wide.

(4) insert one end of the wire into one of these holes:

an27 (0.5x bit value), or
al27 (1x), or,
an25 (2x), or
al25 (4x), or,
aj27 (8x),

and insert the other end into either Vcc/vcore (ah26, ak26, am26) or vss/ground (ah24, ak24, am24, ah28, ak28, am28). the pin locations described here.

depending on your cpu, follow either step 4a, 4b, 4c or 4d.

(4a) tbred 1700, 1800 and barton 2500, to obtain multipliers 11-17x, connect the following holes:

11.0x: no connection required (default)
11.5x: connect an27 to am26 (reset 0.5x bit value high to vcc)
12.0x: connect al27-am26 (1x high)
12.5x: connect an27-am26 (0.5x high), and al27-ak26 (1x high)

13.0x: connect an25-am26 (2x high) and aj27-ah26 (8x high)
14.0x: connect al27-ak26 (1x high), an25-am26 (2x high), and aj27-ah26 (8x high)
15.0x: connect al25-ak26 (4x high), and aj27-ah26 (8x high)
16.0x: connect al27-am26 (1x high), al25-ak26 (4x high), and aj27-ah26 (8x high)
17.0x: connect an25-am26 (2x high), al25-ak26 (4x high), and aj27-ah26 (8x high)

4(b) tbred 2200, to obtain multipliers 13-18x, connect the following:

13.0x: connect an27-am28 (0.5x low), and an25-am26 (2x high) [default]
13.5x: connect an25-am26 (2x high)
14.0x: connect an27-am28 (0.5x low), al27-ak26 (1x high), and an25-am26 (2x high)
15.0x: connect an27-am28 (0.5x low), and al25-ak26 (4x high)
16.0x: connect an27-am28 (0.5x low), al27-am26 (1x high), and al25-ak26 (4x high)
16.5x: connect al27-am26 (1x high), and al25-ak26 (4x high)
17.0x: connect an27-am28 (0.5x low), an25-am26 (2x high), and al25-ak26 (4x high)
18.0x: connect an25-am26 (2x high), and al25-ak26 (4x high)

(4C) for tbred B 2100, 2400, 2700, barton 3000 (333fsb), to obtain multipliers 13-17x, connect the following:

13.0x: connect an25-am26 (2x high) [default, 2100/2700/barton 3000]
14.0x: connect al27-ak26 (1x high), and an25-am26 (2x high)
15.0x: connect al25-ak26 (4x high) [default, 2400]
16.0x: connect al27-am26 (1x high), and al25-ak26 (4x high)
17.0x: connect an25-am26 (2x high), and al25-ak26 (4x high)

(4d) XP-M 2400+

13.0x: connect al27-ak28 (1x low), an25-am26 (2x high), and aj27-ah26 (8x high)
13.5x: connect an27-am26 (0.5x high), al27-ak28 (1x low), an25-am26 (2x high), and and aj27-ah26 (8x high)
14.0x: connect an25-am26 (2x high), and aj27-ah26 (8x high)
15.0x: connect al27-ak28 (1x low), al25-ak26 (4x high), and aj27-ah26 (8x high)
16.0x: connect al25-ak26 (4x high), and aj27-ah26 (8x high)
16.5x: connect an27-am26 (0.5x high), al25-ak26 (4x high), and aj27-ah26 (8x high)
17.0x: connect al27-ak28 (1x low), an25-am26 (2x high), al25-ak26 (4x high), and aj27-ah26 (8x high)
18.0x: connect an27-am26 (0.5x high), al27-ak28 (1x low), an25-am26 (2x high), al25-ak26 (4x high), and aj27-ah26 (8x high) using thinner phone wires

here's a neat one . without any mod (wires, 5-fid or altering bridges),the xp-m multipliers can be changed on-the-fly on a7m with the cpumsr software. register 44 bit 0 was first be set to 1 with wpcredit. alternatively, one could also use wpcrset, as shown here. fsb was 152 in bios 1008.02b, and vcore was set to default by jumpers to far right, and can't be altered with cpumsr. the only multiplier that didn't work was 3x (where the system crashed), others up to 13.5x were ok. according to petr koc's guide, the max available multiplier can be set by opening some L6 bridges, and the max multiplier (24x) will be available when all L6 bridges are opened.

(5) insert your cpu into the socket, and make sure that the cpu sits flat on socket in spite of the wires beneath.

if you're running the cpu at the default speed, the vcore setting can be left at default (VID1-VID4, far right), or increased up to 1.85v max (VID1-VID4, far left) if you're overclocking.

the wire trick has worked for many members; see poll here.

Senor Panadero & Tinkerer, thanks for your help with this thread. thanks to stormyandcold, monge, apa, P@N, sexyjw, cbunn ,dario , Nigel2003, and many others for their work on tbreds and bartons. thanks also to uk bloke for important infos on cpu identification/vcore settings, pci slot sharing, bios bible, slew rate below

Originally posted by UK_Bloke

How to identify your CPU (condensed here)
Colour Pic of voltage jumpers. (don't you just hate that manual)
Shared slots on the A7M266.

I know stormyandcold originally wrote a bible for the A7M266, but a copy can be found here. Some of it's getting dated, but the below is important. (some Creative SBLive tips that worked for me as well)

-=CHIP CONFIGURATION=-
ONBOARD PCI IDE ENABLE: [BOTH]
S2K SLEW RATE CONTROL: [3] for Athlon Thunderbird & Duron, [1] for Athlon XP (Only BIOS 1005 and up)

If you use USB in the A7M266, read this if you find your system seems slow.
HTH

here's another useful bit that uk_bloke put together on how to read barton, thorton, tbred and palomino OPNs.

some people might be wondering about wire tricks on palominos. very risky because the L1s need to be manually closed first, and thus can be very very risky when the ground isn't insulated properly, i.e., can lead to short circuit (vcc-vss/ground connection). also palominos are generally not as overclockable as tbreds or barton. as it is, bios 1007 supports all palominos up to 2100, without any mod

[trats]

ok...i'm back made some changes above.

thanks uk bloke, those infos are really useful m8, so i've copied and pasted it above

[The Tinkerer]

Bump for those who might not have seen this yet.

[Gopher]

I have been reading over the information available about the A7M266 available online, and I am still left with a question.

The 'wire trick' that has been discussed on this website seems to be a solution for using contemporary processors with our antiquated boards, however the beta Asus bios that is floating around leaves me wondering if the 'wire trick' will still necessary once the next version of the BIOS is out.

Or is there a physical incompatibility that makes the 'wire trick' absolutely necessary or has Asus simply abandoned the A7M266?

I have been out of work for two years and I am obviously **** broke, so getting a little more life out of my A7M would be very welcome.

I would also be curious as it the success that anyone has had overclocking the Barton 2500+ on the A7M.

Thanks,

[trats]

bios 1008.02b is the last one for this board. higher end tbred B and barton multiplier codes are messed up on a7m, and that's why they can't run without the wire trick.

cbunn's got the 2500 running here. the 2500 aquca, aqxda, aqxea steppings are very overclockable up to 2.4ghz or more.

[Gopher]

will the beta bios ID the processor correctly?

I flashed mine earlier and it is detecting my 1ghz Duron as Unknown @ 0mhz

what is the highest I might expect of a 2500+ using the retail HSF and stock voltage?

[trats]

that's "normal" for 1008.02b; my tbird 1.33ghz was reported as "0mhz". without any wire trick, the bios will recognise the 2500 as a "xp 1700+" (11x multiplier), and as a 2700 when set to 16.5x. at >17x multiplier, bios will report "0mhz", but that doesn't affect the actual speed the cpu is running.

at stock voltage, 2.2ghz was the highest i've heard for the 2500. 2ghz is likely ok with the retail hsf.

[Gopher]

what would happen if I tried to use the wire multiplyer trick on my Duron?

The dealer I bought it from put nail polish over the bridges.

[trats]

depending on which duron you have, the pencil trick would probably be easier for duron. wire trick on duron is more complicated because one needs to reset up to 4 BP_FID (L3 + L4) + 4 FID (L6). in any case, you may also need to use the pencil trick to close the 4 L1s on the duron.

as an alternative, tbred B 1700 (jiuhb stepping) is v. cheap these days, and also v. overclockable, probably to around 2ghz or more at the default vcore.

[Gopher]

I already bought a 2500+, it will be Wednesday or Thursday before it comes in.

Unfortunetly I can't try the pencil trick because to prevent people from returning processors that had been overclocked they put nail polish over the L1 bridges.

[trats]

nail polish remover/ethyl acetate/acetone might remove it. but since your barton is coming soon, probably not worth the bother

[Monge]

Thanks for this great thread trats, The Tinkerer, Senor Panadero,
UK_Bloke and everybody else who have made this posible

Very easy to find the information needed.

I just want to mention that I noticed a decrease in memory performance around 5-7% when going from 15X to 16,5X multiplier with my Tbred.
I don't know if others have noticed such a thing, but it could be a general thing with the 0,5 bit factors maybe?

I can maybe delete this post after, if this is not a general thing.
Just to keep this thread clear.

[Grazehell]

what is the default speed of the barton xp 2500+ isn't it 1.82 GHZ so shouldn't the multiplier be 13.7 on our 133fsb motherboards