Please Help on Overclocking a 1 GHZ MDD Single

[dopemann]

Hallo,

I have been looking a long time in the internet for a guide on how to overclock you MDD. I found many, but all of them speak of the dual cpu models none for single. I'm trying to get a 1 Ghz Single CPU, w/ 1Mb L3 Cache, its a 7455 and it has the LTC1709EG voltage regulator. So voltage is not the problem, I can also manage the mainboard bus clock from 133 to 167 Mhz, and now here it comes I cant find the PLL0-3 +EXT Jumpers or bridges on the freakin Module. So I also thought that ok dont touch the CPU module, just clock the bus, right. OK in therori I might get what 1.25 Ghz, if I did my calculation corectly then the Module runs 7.5 multiplier. So is it possible because I read on these overclocking guides that the mac wont run if the bus is 167 Mhz and the CPU requirements are 133 Mhz?. Can anyone help with these single models. Thanks

[robotnz1000]

cheers,

If it won't run, you will soon know!

I know its awhile since you posted and also I am a new member - how did you get on here?

I may be a bit presumptious in answering this post of yours I found - but since no other answers I will have a go.

For my halfpenny worth I would suggest the pretty obvious - ie, study the macbidoulle pix and compare your single cpu pcb daughter card. Surely its pretty similar as mac not gonna waste too much time in conversion from one to two cpu!

So the 7455 is the same and so will be plenty of other circuitry and ic's on the pcb. If sizing is a problem or you want to track tracks then you could scan the daughter card and enlarge (or ideally interpolate pix up via pixelsmartscale or genuine fractals) then post pix on the forums for comment - or you can compare those scans to macbidoulle pix and see how apple eliminated the 2nd cpu.

Overall I would guess the daughter cards are not too different and that the jumpers for the voltage/fsb speed ic's are the same and same brand and that the mods therefore should be pretty easy to convert. Which is likely why this post has had no replies.

Keep in mind - its really not that easy to totally fry your CPU, given good paste and proper heat sink-fitting and fans - since you can leave case open and chuck on several extra fans - right! Then, you start cautious with bus clocked to 167 first, that is pretty easy if your daughter card is representative of the dual pix.

Then boot and check stability. Start CPU temp monitor first. If PC freezes on multi task - then check temp on frozen screen! If temp is OK ie under 60c, then maybe your PC CPU cannot handle 167, the default multiplier, and so you need to drop the multiplier, ie 6.5 down to 6.0, etc untill you get stability. I found that voltage increase is not helpful - default is good. Research your cpu model at motorola (google) and read the .pdf specs (datasheet) there if you feel unhappy to move forward anytime. Or find its data sheet on internet.

Again the jumpers will be relevant if the fsb speed/voltage Ic's controllers are the same brands as macbidoulle. Its pretty easy to find the ic's pdf's on google, via search of the ic number, then read the pdfs and trace tracks from ic's relevant pins to find jumper points you need. Not much pcb level skill is really required - as any cheap meter set on continuity can do little damage, due to micro voltage, and if device has audio signal on finding conduction you are done.

:-)

[blacksheep]

Hi all,
Now (instead of 1.67 DP) i'm workin on FW800 1.0 GHz Single. I have OC'ed system bus to 167 MHz with no issues, and i got bumped CPU to 1.25 GHz by only one resistor. So far so good... no troubles. L3 activated correctly, of course.
Next i decided to boost Core Voltage from 1.55 V to 1.65 V and go little higher with multipliers... I get 1.58 GHz on 167 Mhz bus ... but L3 is not activated
I want to stay with stock cooling (and no higher voltage) – this machine is for sale, i don't want to spend money for it; so i started to lower multiplier. I stopped so far at the 1.42 GHz and still no L3 Current CV is 1.65V.
I can't find datasheet for L3 chips, which are: Samsung K7D403671B - HC30, to know their max operating frequency.
Maybe with next 1 GHz module (i got 2 more) i will try higher voltages... 58% OC with 7455A it's a promise for future... who knows

Update:
I tried even 1.75 V with no luck. Now i'm back at 1.25 GHz. Bad news: L3 sometimes active, sometimes not. I didn't notice that before... I think, that L3 chips are near of their frequency limit. So work still in progress

Regards
black

[blacksheep]

This is max frequency for this CPU with properly functioning L3.
Max CPU multiplier is x7 (with 167 MHz bus) - tested and stable.
Max frequency of L3 chips installed on 1.0 GHz daughtercard is 300 MHz, CPU/L3 ratio is 1:4. So with 133 MHz bus, max CPU multiplier will be x9 (1.2 GHz) - not tested yet.
Max CPU clock i have reached (with no activated L3) is 1.67 GHz (unstable), max stable was 1.50 GHz.
Here are screens:


1.17 GHz works excellent with stock voltage – 1.55 V

P.S.
I know, HDD used for test is crappy

regards
black

[ginz111]

Cool. You guys are making me want to go pick up a single for fun. They can be had for cheap. What were your xbench scores at 1.5?

[blacksheep]

At 1.5 i didn't run xbench test. I played mp3, video and net surfing about 30 mins, then i tested deep slee and wake up.
Performance without L3 no matters to me: goal was max with L3 active. Maybe with next CPU i will test it at 1.50 GHz.

[bizzle]

How did you get the CPU module's bus speed to 167mhz? I've yet to find any information on what resistor to change on this module. I've compared to the DP 867 and I can't find an R22 on my module. My logic board is already set up for 167, all that's left is the CPU.

[blacksheep]

bus freq FS2 FS1
133 MHz 0 0
167 MHz 0 1

(0 - resistor, 1 - no resistor)

Like on this pic above (original 1.25 GHz CPU, 167 bus).

When you bump CPU FSB up, you got to lower it's multiplier from 7.5x to 7x or lower (if you leave it as is, CPU clock will be 1.25 GHz - too much for L3 chips).

[bizzle]

Great. I have it working now at 1.25 with no L3 just to see if it's even stable. Any luck increasing voltage to activate L3?

[bizzle]

Apparently I can't read, because that's already been discussed. So far I've had minimal success. I ran default multiplier and 167mhz bus and it booted and ran stable at 1.25 with no L3. Boosted voltage to 1.65v (00010) and it KP'd instantly.

[blacksheep]

Couldn't get active L3 above 1.17 GHz with different voltage settings. As i wrote before, no luck with those chips above 300 MHz. IMO better to have L3 than higher clock freq, so i ended with 1.17 GHz.

[bizzle]

Can you please post a chart for the multipliers?

[blacksheep]

Chart is on bitsandpieces.info, PLL location is described in my 1.25 single thread, here