TbredB 2100+ OC tips...

[Floopyb]

Hi all,
I just ordered a 2100+ AIUHB (not sure what week yet) and I am going to put it in my A7V133. The mobo has the right clock generator that is supposed to work with XPs and jumperless mode, but the problem will be setting the multiplier (If indeed the chip works at all!). In the BIOS I can not get a multiplier over 14 (some lower ones are remapped to 13 and 14 in the bios) and I will probably want a multiplier of around 17 (FSB will be from 133 to 150) which will give me around the 2.3G mark which seems pretty standard. Would I be better off hard coding the chip by cutting the 4th L3 bridge to make the multiplier 17 (giving me an OC range of 2261Mhz@133FSB to 2550Mhz@150FSB) or should I enable the 8x bit so all my multipliers are +8 and set the multiplier in the bios to 9X (9+8=17). I think the 8x bit would be better but I'm not sure exactly what I need to do. Also, what is the best way to cut a bridge with out any specialist tools (does using a 9v battery work well?).

Hard coding tbred: http://www.ocinside.de/go_d.html?/html/workshop/socketa/tbred_painting_d.html
http://www.beachlink.com/candjac/TbredDecode0.htm#Table

setting 8x bit: http://www.beachlink.com/candjac/MultiCode.htm#Pic

Thanks in advance for any help

Floopyb

 

[ARGON]

I don't really too much about what you are trying to accomplish, but I can tell you the best way for most to cut the bridges. Just use the corner of a straight razor blade or an exacto. Cut lightly and retrace the cut if needed.

 

[Deathknight]

From what I have heard the 9v battery trick is the easiest way to cut bridges.

 

[Floopyb]

What would happen if I cut too deep with a blade?

 

[qwerty57]

What would happen if I cut too deep with a blade?

dead cpu, but seriouly get some money even 40 bucks can get you a referb kx7-333 abit why risk a 100 dollar cpu by doing all that too it?? if you get a good chip like mine that does 2600mhz thats a 3400+ I think that if out would cost you 500-700 dollars if you go by mhz per dollar.. not worth it at all IMO.. GL

 

[seer33]

qwerty57..what stepping is ur 2100 b...also what cooling do you use?

 

[Floopyb]

qwerty57- dont forget I need at least 512Megs of ram on top of that and I would like to keep using the promise raid controller on the A7V133.

 

[OC Detective]

Your 8X is already enabled so that is not an option - breaking the 4th L3 will however add 4X to your multi giving you the 17 multi that you desire.

 

[pctuner@amd]

Do not cut with a razor blade use a 5v source from your power supply or a nine volt battery and short the bridges they will blow like a fuse and cause no external damage and not risk losing your processor...

 

[Floopyb]

OC Detective-
Will this hard code my cpu at 17 or will I still be able to change the multiplier in the bios (but add 4 to what the bios says)?

 

[OC Detective]

I suspect you will get 17 and only 17 - though I am not 100% certain. I think if you tried it you maybe somewhat of a pioneer!

 

[John C.]

OC Detective-
Will this hard code my cpu at 17 or will I still be able to change the multiplier in the bios (but add 4 to what the bios says)?

As OCDetective says, you will hard code 17X, but no one knows how system/bios will behave, probably just like now....no change. If you want to be a pioneer, here's a suggestion.

Tell your system it has a Multiplier lower than 12.5X, but "don't change" your current default 13X Multiplier. Then the system might allow you to change the 1st 4 bits (5X-12.5X) in the bios...to which you'd add the 8X from your "open" 5th L3. Here's how we'd try this manuever.

Do the AJ27 wire trick, this will ground the 8X signal circuit on the system/mobo side to hopefully fool the system into letting you set 5X-12.5X. But then "open" the 5th L1 which will isolate the circuit on the CPU side from the wire trick ground and keep the 8X signal circuit on the CPU Multiplier module side HI. Written up as option #2 http://www.beachlink.com/candjac/index.htm link to Workarounds article.

Then get into bios asap on reboot and try 6X/14X, 7X/15X, 8X/16X, 8.5X/16.5X, 9X/17X, Other settings are Reserved and Remapped very high, might not post....
see Tbred article, settings Table, for those very high settings we've left out to avoid confusion and disappointment.

Seriously, the oc community needs this test.
John C.

 

[Floopyb]

John C-
Let me see if I have this correct, once I get my new cpu I should reboot, set my bios to, lets say, 8x100 just too keep numbers low and hope for a post, reboot then power down and remove my old chip. B4 I put in the 2100 I should put a thin wire between pin sockets AJ27 and AH28 (I couldn't quite understand the explanation to what this will do on http://www.beachlink.com/candjac/Workarounds.htm , except that perhaps the mobo will think that 8x is set low but the chip sees 8x set to high?????) and make sure that the 5th L3 is open to allow the cpu to have the 8x bit set to high. Once I reboot I can set the multiplier in the bios to 6 and expect 14x or 7x=15x, 8x=16x etc except for reserved multipliers that are set to 22x or so. Does that sound about right.
I am very willing to pioneer this, but I need the help of everyone to make sure I am doing the right thing to begin with.
BTW, I should be getting the CPU tomorrow, so I will be ready to rock then.

 

[Floopyb]

BTW does this have anything to do with what you are talking about:
"4. This is just a theory. I make no claims as to doing it. Cut the trace to the .5x multiplier pin and move it to the 8x multiplier pin (AJ27) on the socket. This would do away with the half steps in the multiplier and should give you 5=5, 5.5=13, 6=6, 6.5=14, etc."
taken from http://mysite.verizon.net/res0exft/
or is that something slightly different?

 

[John C.]

John C-
Let me see if I have this correct, once I get my new cpu I should reboot, set my bios to, lets say, 8x100 just too keep numbers low and hope for a post, reboot then power down and remove my old chip. B4 I put in the 2100 I should put a thin wire between pin sockets AJ27 and AH28 (I couldn't quite understand the explanation to what this will do on http://www.beachlink.com/candjac/Workarounds.htm , except that perhaps the mobo will think that 8x is set low but the chip sees 8x set to high?????) and make sure that the 5th L3 is open to allow the cpu to have the 8x bit set to high. Once I reboot I can set the multiplier in the bios to 6 and expect 14x or 7x=15x, 8x=16x etc except for reserved multipliers that are set to 22x or so. Does that sound about right.
I am very willing to pioneer this, but I need the help of everyone to make sure I am doing the right thing to begin with.
BTW, I should be getting the CPU tomorrow, so I will be ready to rock then.

(2) mods....
#1 the AJ27 to AH 28 wire, this "positively" grounds the "trace" from 5th bit's circuit to the mobo. AJ27 connects to the trace, AH28 = Vss = ground.
#2 Open the 5th L1 bridge to "isolate" the 5th bit's HI on the CPU side of the circuit from the ground/LO you just set in #1...you forgot this one. Otherwise the ground/LO from #1 over-rides the default HI.

The 5th L3 will be Open on the 2100, as also the 3rd.
Those 2 open set 13X = 2100.

Otherwise you got the procedure and assumptions about what happens right. Suggest you print out the
Multiplier Table from the Tbred article. It has all the resrved and remapped settings, will help you understand what you get from what you set.

If you set your Host Clock to 100 MHz you could try to replicate some of those very high Remapped Multi's.
Good Luck,
John C.

PS Re your other post, it's a different way to handle the 5th bit.

 

[Floopyb]

Ok, I final got the cpu after some postage screw ups, it is a AIUHB 0302WPCW (brown). I dropped my FSB to 100 and my multiplier down to 8x, shut down and dropped the 2100+ in. BTW my mobo is a A7V133 1.04. with an ICS UK020477A 0052 94215AF. I wacked the power switch and I got post, it said 1600+. In the bios the multiplier was set to 12.5 anf FSB to 100 but once I booted into windows I noticed that MBM5 was saying my cpu was 2021MHz !!! I quickley fired up wcpuid only to get a reboot, DOH! I still had 3.0. Once i got 3.1 Wcpuid confirmed what MBM was telling me 2021, 20x100.9. Where the hell is the 20x comming from? 20x is remapped from 4.5x but that still doesnt explain what is going on. I tried to set the bios to 13x133 but then I couldnt post (I assume my bios is truing to stop me overclocking by setting it to the CPU default of 20x133!!!!! If only that would post)and had to put my old cpu in to get into the bios and set the FSB to 100. So at the moment I am running at 2120 (106x20) Vcore=1.71. I cant seem to up the Vcore past here, I just get reboots. My cooling is a 2 year old alpha and a 6800rpm fan on top and 4 2900rmp 80mm case fans. CPU temp is 51 under load (It is the middle of summer down here in Australia!). I am pretty happy with wat I have at the moment although it would be nice to be able to boot at 133FSB. I might push it a bit further in the cooler months.

 

[John C.]

I dropped my FSB to 100 and my multiplier down to 8x, shut down and dropped the 2100+ in. BTW my mobo is a A7V133 1.04. with an ICS UK020477A 0052 94215AF. I wacked the power switch and I got post, it said 1600+.

In the bios the multiplier was set to 12.5 anf FSB to 100 but once I booted into windows I noticed that MBM5 was saying my cpu was 2021MHz !!! 3.1 Wcpuid confirmed what MBM was telling me 2021, 20x100.9. Where the hell is the 20x comming from? 20x is remapped from 4.5x but that still doesnt explain what is going on.

I tried to set the bios to 13x133 but then I couldnt post (I assume my bios is truing to stop me overclocking by setting it to the CPU default of 20x133!!!!! If only that would post)and had to put my old cpu in to get into the bios and set the FSB to 100. So at the moment I am running at 2120 (106x20) Vcore=1.71.

Congratulations Floppyb "Pioneer"!! By the numbers...

1:- "Post" said 1600 because you set the "1st 4 bits" to 8X, but because the 5th bit on the 2100 is open by default the 5 bit code of 8X from 1st 4 bits + 8X from the 5th bit = legitimate code for 16X...times 100 =1600.
Nothing wrong with that, just what should happen if 4 bit mobo sets 1st 4, Multiplier reads all 5, and 4 bit mobo does not interfere with 5th bit in any way...as we wrote 4 bit mobos should/might do in Workarounds article.

2:- 12.5X getting 20X is also expected once the 16X in #1 worked...same theory...bios reset 1st 4 bits to 12.5X + 5th bit HI. But here's how you got the 4.5X.
O-O-C-C-C = 12.5X, reset by bios but it CAN"T RESET the 5th bit, so what you get is...
O-O-C-C-0 = 4.5X from the 1st 4 by bios and 5th from the open 5th L3... which 4 bit mobo "leaves alone". And since 4.5X remaps to 20X, you got it. That's what's going on....Success Pioneer...Success!!! You just gotta "understand the Multiplier Code.

3:- 13X133. Not sure why no post with 13X.
C-C-O-C-O = 13X. Maybe it's something to do with timing/setup when system sees 13X. BUT BUT BUT, looking at #1 and #2, shouldn't you be setting 5X in the bios, instead of 13X?? 5X 1st 4 bits + 8X from open 5th L3 = 13X.
Just remember that 14.5X and 15.5X are Reserved, 14.5X might not boot or remap to 21X and 15.5X should remap to 22X...so expect those if/when you try 6.5X and 7.5X. Maybe try "them" at 100MHz. You've got the Table of settings and the Workarounds article, so you should be on your way to more optimizing of your system.

So you've just verified what we've theorized about "Workarounds" and what Soulkeeper and others have discovered. Congratulations again, and keep posting your results, but now YOU should be able to explain them...right??
John C.

BTW, the A7V series of mobos was used by Soulkeeper and another, so they seem to be "excellent" older mobos to apply these workarounds as they ALL seem to keep their mitts off the 5th bit and don't interfere in other ways as well.

 

[UnLoadeD]

Wow this is very interesting thread. I've got a 1700 juihb coming tomorrow. I planned on using it in my epox 8k5a2+ board. I've got an A7V sitting around with a 1ghz tbird in it. I might haveta try it out in there too. I'd planned on setting the old box up as HTPC with my 8500 AIW. Also I'm one pump away from having 2 extra water cooling setups. If this new chip does well in there I'll get another and a pump for that setup. It would be nice being watercooled and quick enough for good capturing. I've got decent ram in that box that did 150mhz nicely, now I am getting excited. I had planned on it being a slower but still useful box, now it sounds like I might get a decent setup for pretty cheap. Thanks for all the info and pioneering guys. I'll post any info I can gather once I get a chance to tinker with it.

peace.
unloaded

 

[The Coolest]

That is really good to hear it works for you, kinda, I think that what John C. suggests should help ya, right now you're kinda getting x8 + *whatever_you_select_in_the_BIOS*, so just sub a 8 from the multi Uri shooting for and it should be fine, just avoid the ones that John C. mentions. I'm hopefully gonna get my tbredB soon enough, and gonna take it on a test drive. Also you're pretty lucky, as I heard many people have reported that A7V133 v1.04 and down, don't like XPs much. I've got v1.05 tho, so there should be no problems for me . Thanks for being the pioneer, as I thought I was gonna be the first one there. I always loved this ASUS mobo, and now, reading your post I love it even more.

 

[Aggieeng]

Whoa...whoa.....whoa......

So is there a way to get a 14X multi WITHOUT cutting any bridges? Like set one of the lower multi's and it'll reset it or something? I'm having trouble following the posts up there...they confuse me....