Thread

G4 533MHz ZIF - $20

G4 533MHz ZIF - $20 Troubleshooting 84 posts Mar 19, 2008 — May 14, 2008

#61 Thu, 10 Apr 2008 - 05:50

I found a jumper table for speed settings on NewerTech ZIFs. I'm going to try and work in the PowerForce results I found above and see if they match.

[in B&W] set at 6 / 550 Mhz / set it at 7 it did not boot / set it at 5 / 650

For B/W G3: 6=550, 2=500for Beige G3: 7=500, C=533

Code:

 x      Bus speed      NewerTech    PowerForce
Mult    66    100      Jumpers      Dial
===============================================
3.0    200    300      | - - -      
3.5    233    350      | | | -      
4.0    266    400      | - | -      
4.5    300    450      - | | |      
5.0    333    500      | - | |      2
5.5    366    550      | - - |      6
6.0    400    600      | | - |      
6.5    433    650      - | - |      
7.0    466    700      - - | -      
7.5    500    750      - - - |      7
8.0    533    800      | | - -      C (=12)
10.0    666   1000      - | | -      
===============================================

                     | = jumper
                     - = no jumper

Heheh. Is anyone else seeing the pattern here?

#62 Thu, 10 Apr 2008 - 07:07

erm... no pattern :?:

only seeing that its the jumper settings are the same as the DIP switches on my G3/500

- afair.

mine too wont boot at what i guess is 600mhz

#63 Thu, 10 Apr 2008 - 07:45

There are 16 (4^2, hex10, 1111 binary) positions on the dial and four jumper positions. From that I guessed that the dial is a four pole switch, with the hex numbers corresponding to a four digit binary representation of the on-off state of four jumper switches.

So if you make the following changes to the table: read right-to-left rather than left-to-right, and make 0 = jumper and 1 = no jumper (which would make sense if they were normally-closed {NC} switches) ...

Code:

LSBMSB   MSBLSB    Invert     base10     Dial
========================================================
| - | |     | | - |     0 0 1 0    0+0+2+0  =   2
| - - |     | - - |     0 1 1 0    0+4+2+0  =   6
- - - |     | - - -     0 1 1 1    0+4+2+1  =   7
| | - -     - - | |     1 1 0 0    8+4+0+0  =  12 (=C)
========================================================

| = jumper              0 = jumper 
- = no jumper           1 = no jumper

Or does no-one do sense- and endian- reversed binary to hex conversions in their head anymore?

Bah, kids these days. Get off my lawn.

#64 Thu, 10 Apr 2008 - 08:07

... and so from there we can fill out the rest of the table:

Code:

NB THESE NUMBERS ARE UNTESTED except as noted above.  

=======================================================================
 x      Bus speed      NewerTech    PowerForce
Mult    66    100      Jumpers      Dial          Note:
=======================================================================
3.0    200    300      | - - -      E (14)
3.5    233    350      | | | -      8            
4.0    266    400      | - | -      A (10)
4.5    300    450      - | | |      1
5.0    333    500      | - | |      2
5.5    366    550      | - - |      6
6.0    400    600      | | - |      4
6.5    433    650      - | - |      5
7.0    466    700      - - | -      B (11)
7.5    500    750      - - - |      7
8.0    533    800      | | - -      C (12)
8.5    566    850
9.0    600    900      - - - -      F (15)        jumpers unconfirmed
9.5    633    950
10.0    666   1000      - | | -      9
=======================================================================

                     | = jumper
                     - = no jumper

" 10x ratio is not supported on all G3 CPUs,
  9x is the max ratio on G4 7400 CPUs in most G4 ZIF upgrades 
     (and all Newer Tech G4s I've seen)" *

I'll edit and add to this table as I fill out the rest. I'll also notate which settings have been tested. When it's all finished, I'll add it to the wiki.

* quote by Mike on xlr8yourmac

#65 Thu, 10 Apr 2008 - 08:32

ZIF / trim dial with numbers / speed setting / PowerForce by PowerLogix / positive match.

Click to expand...

I have a 500MHz G4 that is just like that. I just kept turning the dial until the system profiler reported a speed of 500MHz. It's running in my 8500 with a ZIF Carrier card.

Temetka: any chance you could pop the top and check what the dial is set to?

#66 Thu, 10 Apr 2008 - 09:15

I've just never seen one of those pad things before and the heatsink wobbles without it. I tried putting a little bit of extra bend in the clamp so it would hold it more firmly, but it still seems to be pushing down in the wrong spot... It holds the heatsink at a slight angle at the contact point of the processor.

You might have the clip on backwards. If you look closely you'll see that the lowest point is not exactly in the centre. That point is meant to sit exactly above the CPU die. Try reversing it.

If it doesn't seem to be applying enough pressure after your bending, I'd suggest shimming it up with something metallic between the clip and the heatsink. Whatever you put in there should be conductive and firmly held down by the clip so it doesn't move. The heatsink is meant to be pretty firmly held down by the clip; ie it should be difficult but not impossible to get the clip off and on.

#67 Sun, 4 May 2008 - 19:48

hmmm...

is there a way to set these ZIFs to a 9.0 or 9.5 cpu/bus ratio?

i need these because i want to play around with the ZIF in my 9600 on a ZIF carrier.

the setting of the ZIF itself isnt overidden by the setting of the ZIF carrier.

#68 Sat, 10 May 2008 - 13:15

hey, this is funkymonkeypie, I have one of those PowerLogix G4 ZIF CPUs with the dial on it, the fastest i could get it go in a Mac G3 blue and white is 650 Mhz with the full 100 Mhz bus speed, i tried all combos of different dial settings and bus speeds. I think that the limit is the cache speed on the cpu, I remember getting it to boot a couple of times at a higher speed than 650Mhz by dropping the cache speed with CPU director but it was totaly unstable.

Still running great after a year, I put a 80mm fan on the heatsink to be safe, offset slightly towards the case hinge to clear the optical drive housing, powered off the same leads that run the case fan.

I picked this upgrade because there was no need to install any cache enablers for it, i hate having tu rely on manufacturers to provide legacy drivers, or updates for new OS's.

#69 Sat, 10 May 2008 - 20:43

Hey great! That's one of the unconfirmed predicted settings from my list. If I'm not insane, your dial should be on 5. Is that right?

And if my calculations are correct, Temetka's should be on 9, for a bus multiplier of 10x

#70 Sat, 10 May 2008 - 23:47

Just thought I'd let you know, the G4/400 (7400) I bought from IC-China is pretty well confirmed DOA. I tried it in the ZIF card and it didn't work but assumed it was the ZIF card being fussy. I tried it in the B&W today prior to donating it to a friend and it wouldn't boot in that either. I suspect it's fragged. It looks like it was originally fitted by someone with a rather too generous hand with the Arctic Silver. It was COATED in it. Either way it doesn't work

#71 Sat, 10 May 2008 - 23:54

Darn shame there Mark. At least you're not out a huge wad of cash for it.

hmmm...is there a way to set these ZIFs to a 9.0 or 9.5 cpu/bus ratio?

It's not documented anywhere I've been able to find, but from the table above, there's only /edit/ four undocumented settings: Powerforce dial 0, 3, D (13) and F (15). They would correspond to:

Code:

NB THESE NUMBERS ARE UNTESTED except as noted above.  

===============================================
 x      Bus speed      NewerTech    PowerForce
Mult    66    100      Jumpers      Dial
===============================================
??                     | | | |      0
??                     - - | |      3
??                     - | - -      D (13)
??                     - - - -      F (15)  /edit/ now documented above
===============================================

/edit/ where ?? is either 8.5, 9, 9.5 or some number below 3 or above 10, or a non-functioning setting.

So far I haven't been deduced any pattern between the jumper settings and the resulting bus multiplier.

#72 Sun, 11 May 2008 - 06:53

Just thought I'd let you know, the G4/400 (7400) I bought from IC-China is pretty well confirmed DOA. I tried it in the ZIF card and it didn't work but assumed it was the ZIF card being fussy. I tried it in the B&W today prior to donating it to a friend and it wouldn't boot in that either. I suspect it's fragged. It looks like it was originally fitted by someone with a rather too generous hand with the Arctic Silver. It was COATED in it. Either way it doesn't work

It may still be alive. Apple patched the ROM in the blue and white when the Yikes! was released so you couldn't just plug in a G4. Best machine to try would be a Yikes! since that is probably what it came out of originally.

#73 Sun, 11 May 2008 - 07:55

You need a firmware patch installed before you put the G4 into a B&W for it to work.

#74 Sun, 11 May 2008 - 10:26

F on the dial of my powerlogix G4 zif boots into both osx and os9 at 600 Mhz and a 66Mhz bus speed

#75 Sun, 11 May 2008 - 16:17

Wooo! Awesome. That's one of the undocumented settings. 9x - which answers this:

is there a way to set these ZIFs to a 9.0 or 9.5 cpu/bus ratio?

I'll update above with the F dial setting and the deduced, untested jumper settings.

/edit/ Welcome to the forums Mr Pie

and thanks for putting your previous results up somewhere they could be found. Couldn't have got this far without you.

How'd you find us?

#76 Sun, 11 May 2008 - 18:15

My B&W is already running a G4/500, it's been patched. I was testing the 400 to post on to a friend with another B&W.

#77 Mon, 12 May 2008 - 01:32

I'm ashamed to say i googled "funkymonkyPiE" and it brought me here. I was in the middle of trying to overclock two Wallstreets that I use to run Opcode Vision sequencing software on some Studio5 MIDI interfaces that I have.

Mannaged to get a 266 to 300 and a 233 (512 cache) to 266 with charts from here:

http://www.phrax.ca/projects/OverclockPowerbookG3Wallstreet/

Overclocking an Apple G3 Series (Wallstreet) Laptop

There are links at the bottom of that page where he got his info, it is a page in french, the french guy got his charts from a japanese site that is no longer available.

My main project mac is a G3 Blue and White with;

Powerlogix G4 at 650Mhz

4 x 256 Ram (CL-2)

ATI 7000 Video card (not flashed)

Sonnet tempo Trio ATA-133 / USB 2.0 / Firewire

M-Audio Delta 1010LT Audio Card (10 channel)

Belkin F5D 7001 Wireless G+ card (airport extreme)

Griffin Modem to serial Adapter

Pioneer 111 DVD/RW Drive

21" Studio Display

I have an Nvidia 5500 PCI card in a PC that is on a deathwatch that I'll flash over to mac to complete the project.

Not sure if this is useful but i found a site with ALL the Apple service manuals from the 8500 and newer.

http://node123.cit.geneseo.edu/~stratton/ServiceManuals/

Index of /~stratton/ServiceManuals

My project this week is to brute force an install of Leopard on the BandW, I know its not possible with the official release but i have seen reports that the orriginal developers release had the drivers for and installed on a Yikes with an ATI 7000. So i am going to first try installing that then Pacifisizing the official 10.5 release on top of that in the hopes that the developer drivers stick.

#78 Mon, 12 May 2008 - 08:08

I'm ashamed to say i googled "funkymonkyPiE" and it brought me here.

No shame in it sir [

)] ]'>

/ overclock two Wallstreets / Opcode Vision / Studio5 MIDI interfaces / M-Audio Delta 1010LT Audio Card (10 channel)

Ah... your circle of interests overlaps mine.

/ charts from here / links / a page in french / charts from a japanese site that is no longer available.

Don't ya love the net [

] ]'>

My main project mac

Remind me to send you a Tshirt when I get around to making them up.

#79 Mon, 12 May 2008 - 14:58

hmmm... thx bunsen for figureing out the 9.0 and 9.5 settings.

Im running the 9600 @ 530/53 with 10x now. its solid as a rock.

#80 Mon, 12 May 2008 - 16:51

Well, I haven't figured out the 9.5 setting yet, and the 9.0 is just what's been reported by the funky monkey.

What machine are you running, and is it a Newer ZIF with jumpers, or a PowerForce with dial? Is 10x the only one you've tested, or can you help us confirm any of the other settings?

When I say confirm, I don't mean for stable long-term performance; just long enough to verify the CPU speed corresponds (or doesn't) to the predictions.

#81 Mon, 12 May 2008 - 23:24

yea sure...

hmm testing would mean to rip apart my nicely done setup. My ZIF has DIP switches but the settings correspond to teh newer tech settings. its just inverted.

no jumper means for me to set the switch to the "on" position.

Im running it in my PM 9600 as i said using a daystar ZIF carrier.

i may test the 9x setting if i can find some time.

#82 Tue, 13 May 2008 - 06:49

My 200 watt power supply in the B&W has just now bit the byte, and i personally blame everyone here.

For weeks I have been getting freezes everytime my Optical drive started ramping up to top speed or i attempted to do multiple gigabyte transfers over LAN. I should have known better than to load up all the slots in the mac and not check the wattage on the PS, which it turns out is a paltry 200w. So a new PC-ATX 420w is going in tomorrow, and I'm going for the magic 700Mhz @ 100 bus mark (B on the dial), as it would boot to at least the mouse and rounded corners at that setting. Just something to consider when attempting the higher speeds, those olde PS units can probably power the CPU but as soon as something else drains the current, (like a hard drive) they crash, i suspect that it is the on-chip cache that is the real culprit. I find that heat is not really the issue as there is no increase in voltage as you can do with PC boards when overclocking.

On another note, I noticed that there are only 6 pins coming out of the dial on the G4 ZIF not the 8 or 5 that i would have expected for 16 different settings.

I am curious what y'all use for stress testing for stability, as there are heaps of apps on the WIN-PC side but not much for running the cpu at full tilt on the mac. For quick tests i drop in a cd and dump to mp3 in iTunes, but there is no error checking on that, just full cpu usage.

#83 Tue, 13 May 2008 - 07:49

I find that hard to believe personally, given that anytime I've over-clocked anything it's crashed under some sort of heavy load as a result, and disk access or network access is a prime cuplrit as that puts a lot of traffic across the CPU. If it was true then my B&W running a stock G4/500, a Radeon 9200 and 3 hard drives (2 of which were 10krpm SCSI) would also have struggled, and it doesn't. Faster CPUs have a bearing on it, but also consider that people run 1GHz G4s and G3s in B&W PowerMacs without issues. Although only rated at 200W, the PSUs are very good ones and the 200W output rating is a pretty accurate number, rather than these cheap crappy PC PSUs that say they are 500W and are actually substantially (and in some cases scarily) less under heavy load.

I think you may find the reason it's locking up is the CPU is running too fast, and can't cope with the memory swapping that that sort of ethernet traffic generates, and it's crashing. I suggest you get hold of the Altivec Fractal Generator and run a few repeated runs on it and see if that locks it up. If it does it's time to hang up the over clockers tools and step it down a few notches. It's a typical over-clockers reaction to blame everything else rather than the decidedly risky over-clocking they are messing around with [

] ]'>

You'll either suffer lockups for the rest of eternity, or blow the CPU completely (much less likely, but more so the higher you go). Even if you get it stable at a higher speed it'll last a lot less long than it would at it's factory setting, and, given it's probably second hand anyway, that would have me worried about it's life span.

#84 Wed, 14 May 2008 - 21:04

On another note, I noticed that there are only 6 pins coming out of the dial on the G4 ZIF not the 8 or 5 that i would have expected for 16 different settings.

Two grounds?