Sidebar Sidebar
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

Wish me luck...

Well, things took an interesting turn. I'm typing this post from the Gigadesigns-equipped computer.

I finished doing a complete clean install and software updates(entirely in the computer I intended to install the upgrade in) and then flashed the ROM again.

I wanted to see how far I could get into a Leopard boot without installing the Kexts, and amazingly enough when I powered it up, it booted to the desktop.

Here's the interesting part-it's only reporting the speed as 1.33ghz despite using the DIP switch settings for 1.6ghz.

I'm going to install the Kexts now, and see what happens, but at least it's booting!
 

Attachments

  • Picture 1.png
    Picture 1.png
    1.3 MB · Views: 191
In addition, here are the Xbench results from the factory dual 1gzh and the Giga dual running at 1.33. The L3 cache on the factory 1gzh doesn't show up under the specs, as I was never able to get the computer to recognize it(supposedly, it was locked out on computers that didn't originally ship with L3-this was originally a single 800mhz QS 2002)
 

Attachments

  • Picture 2.png
    Picture 2.png
    778.1 KB · Views: 124
Alright, I'm getting somewhere, but I think that I'm already running into heat issues.

I inadvertently found a clock multiplier setting not listed in the documentation that came with the processor. For anyone interested, switches 1,3, and 5 "on" and 2 and 4 "off" give 10x for 1.33ghz in 133mhz FSB system. I'm running it now with this setting at 1.5, and it seems to be 100% stable.

I was unable to get anywhere at 1.6ghz...I think I may already be running into heat issues. Basically, at 1.5V, it will boot but lock up pretty much as soon as I try to do anything. Raising the voltage, even to 1.55V, causes a kernel panic on booting. Lowering the voltage to 1.45V will generally get it stuck either during booting or as soon as the blue screen(before the login screen appears). Lower the voltage too much-like to 1.35V, and it will generally kenel panic.

The instructions give the lowest setting as 12x, which corresponds to 1.6ghz in a 133mhz FSB system. Obviously, a 10x setting exists. I'd be interested to know if even an 11x(1.46ghz) exists so that I could try it and then get it stable before moving up to 12x(1.6ghz). I'm sure that the setting is there, it's just a problem of finding it...and I'm not about to go through a bunch of trial and error just to do so(I was lucky that I inadvertently found the 10x setting).

Even at 1.33ghz, though, it's noticeably faster, so I'm not totally disappointed. I probably am going to have to go "fans galore" to get 1.6, if my guesses about it getting too hot are correct.
 
bunnspecial said:
Alright, I'm getting somewhere, but I think that I'm already running into heat issues.

I inadvertently found a clock multiplier setting not listed in the documentation that came with the processor. For anyone interested, switches 1,3, and 5 "on" and 2 and 4 "off" give 10x for 1.33ghz in 133mhz FSB system. I'm running it now with this setting at 1.5, and it seems to be 100% stable.

I was unable to get anywhere at 1.6ghz...I think I may already be running into heat issues. Basically, at 1.5V, it will boot but lock up pretty much as soon as I try to do anything. Raising the voltage, even to 1.55V, causes a kernel panic on booting. Lowering the voltage to 1.45V will generally get it stuck either during booting or as soon as the blue screen(before the login screen appears). Lower the voltage too much-like to 1.35V, and it will generally kenel panic.

The instructions give the lowest setting as 12x, which corresponds to 1.6ghz in a 133mhz FSB system. Obviously, a 10x setting exists. I'd be interested to know if even an 11x(1.46ghz) exists so that I could try it and then get it stable before moving up to 12x(1.6ghz). I'm sure that the setting is there, it's just a problem of finding it...and I'm not about to go through a bunch of trial and error just to do so(I was lucky that I inadvertently found the 10x setting).

Even at 1.33ghz, though, it's noticeably faster, so I'm not totally disappointed. I probably am going to have to go "fans galore" to get 1.6, if my guesses about it getting too hot are correct.
Click to expand...

1.33 is a good speed. the Xserve G4 and iBook G4 were available with 1.33 processors. Try adding a high CFM (95+) to the CPU 80mm SHOULD be big enough.
 
bunnspecial said:
Alright, I'm getting somewhere, but I think that I'm already running into heat issues.

I inadvertently found a clock multiplier setting not listed in the documentation that came with the processor. For anyone interested, switches 1,3, and 5 "on" and 2 and 4 "off" give 10x for 1.33ghz in 133mhz FSB system. I'm running it now with this setting at 1.5, and it seems to be 100% stable.

I was unable to get anywhere at 1.6ghz...I think I may already be running into heat issues. Basically, at 1.5V, it will boot but lock up pretty much as soon as I try to do anything. Raising the voltage, even to 1.55V, causes a kernel panic on booting. Lowering the voltage to 1.45V will generally get it stuck either during booting or as soon as the blue screen(before the login screen appears). Lower the voltage too much-like to 1.35V, and it will generally kenel panic.

The instructions give the lowest setting as 12x, which corresponds to 1.6ghz in a 133mhz FSB system. Obviously, a 10x setting exists. I'd be interested to know if even an 11x(1.46ghz) exists so that I could try it and then get it stable before moving up to 12x(1.6ghz). I'm sure that the setting is there, it's just a problem of finding it...and I'm not about to go through a bunch of trial and error just to do so(I was lucky that I inadvertently found the 10x setting).

Even at 1.33ghz, though, it's noticeably faster, so I'm not totally disappointed. I probably am going to have to go "fans galore" to get 1.6, if my guesses about it getting too hot are correct.
Click to expand...

I was worried you'd get it working at all. Glad to see you found something though. I'd have never figured that out!

I believe the Gigameter install is mainly to have the system report the correct speed. But it installs kext files and without it the Mac never booted for me. I guess maybe because I didn't hit the right combination whereas you did.

In any case, yes, heat. It will function only for a short time and then freeze. The suggestion to add an additional fan to it is a good one. I had an 80mm fan strapped on to the top as well as those two smaller fans, so maybe something you might want to try as well, IDK.

But this is something I don't understand about this design. Why provide faster speeds without proper cooling? The Sonnet I have has no dip switches. There is no way to change the speed (short of a soldering iron I guess) and it keeps perfectly cool.

Anyway, I'm glad it's now working for you!
 
Thanks-I too am glad that I have at least something out of it. Even at 1.33ghz, it's no slouch and still faster than any stock Quicksilver.

We're snowed in(or more correctly iced in) with no church this morning, so I'm going to spend some more time playing with it see if I can get anything faster. Unfortunately, I doubt I can even make it to Radio Shack to get fans if that's what it takes.
 
MatthewLTL said:
1.33 is a good speed. the Xserve G4 and iBook G4 were available with 1.33 processors. Try adding a high CFM (95+) to the CPU 80mm SHOULD be big enough.
Click to expand...

My G4 iBook has 1.33 and it runs extremely smooth, even with Leopard Webkit and Minecraft.
 
The 7447s are supposed to have an integrated thermal diode. Does anyone know if there's any possible way to read the temperature off them?

Temperature Monitor can't seem to see them.
 
bunnspecial said:
Alright, I'm getting somewhere, but I think that I'm already running into heat issues.

I inadvertently found a clock multiplier setting not listed in the documentation that came with the processor. For anyone interested, switches 1,3, and 5 "on" and 2 and 4 "off" give 10x for 1.33ghz in 133mhz FSB system. I'm running it now with this setting at 1.5, and it seems to be 100% stable.

I was unable to get anywhere at 1.6ghz...I think I may already be running into heat issues. Basically, at 1.5V, it will boot but lock up pretty much as soon as I try to do anything. Raising the voltage, even to 1.55V, causes a kernel panic on booting. Lowering the voltage to 1.45V will generally get it stuck either during booting or as soon as the blue screen(before the login screen appears). Lower the voltage too much-like to 1.35V, and it will generally kenel panic.

The instructions give the lowest setting as 12x, which corresponds to 1.6ghz in a 133mhz FSB system. Obviously, a 10x setting exists. I'd be interested to know if even an 11x(1.46ghz) exists so that I could try it and then get it stable before moving up to 12x(1.6ghz). I'm sure that the setting is there, it's just a problem of finding it...and I'm not about to go through a bunch of trial and error just to do so(I was lucky that I inadvertently found the 10x setting).

Even at 1.33ghz, though, it's noticeably faster, so I'm not totally disappointed. I probably am going to have to go "fans galore" to get 1.6, if my guesses about it getting too hot are correct.
Click to expand...

Fans galore :D I love fans galore that is the best part of overclocking!!! :D
 
bunnspecial said:
The 7447s are supposed to have an integrated thermal diode. Does anyone know if there's any possible way to read the temperature off them?

Temperature Monitor can't seem to see them.
Click to expand...

Only the 7447A has a thermal diode. I don't think any third party upgrade uses a 7447A.
 
I didn't run Geekbench with the factory processor, but from past experience I know that the score with the dual 1ghz should be around 850-900. Here it is with the Gigadesigns at 1.33.
 

Attachments

  • Picture 4.png
    Picture 4.png
    1.1 MB · Views: 92
bunnspecial said:
I didn't run Geekbench with the factory processor, but from past experience I know that the score with the dual 1ghz should be around 850-900. Here it is with the Gigadesigns at 1.33.
Click to expand...
Pretty good!

I am glad you have this working!
 
So far, I've had absolutely zero issues with this running at 1.33ghz-as I said, I'm just glad that I discovered this setting by accident(and didn't inadvertently fry something). It's as reliable and stable as a factory processor.

Until I get some more cooling on it, 1.6ghz just isn't going to happen. I've gone through and cleaned and oiled all three of the factory fans to get the most I can out of them, which means that I need to add more and/or replace the stock fans with better ones(I'm inclined to do both, at least for the main case fan and PSU fan). I'll try the 80mm fan on top of the two CPU fans and see if that helps any.

I tried 1.6 again today. I know that AS5 specifies some "cure time" for maximum efficiency. From pulling heatsinks, I think there is some merit to this, as AS5 that's been in place for a while looks a LOT different from freshly applied AS5. I've run benchmarks and other things today to get the CPUs good and toasty(and spent some time with them pegged at 100%) along with some idle periods so had hoped that a few thermal cycles over 20 or so hours might help things enough that I could at least get somewhere at 1.6ghz...apparently I though wrong, although admittedly I think that the cure time is given as somewhere around 200 hours of use with several hot/cold cycles in between. Since it seems to work so well at 1.33, I'll probably shut it down tonight at least for an hour or so to let things get back to room temperature and hopefully help the process along.
 
Along the same lines of thought as my previous post-does anyone know of a good stress test program that will work on PPC?

I know Prime95 is a defacto standard for x86, but I haven't been able to find anything like that for PPC/OS X.
 
You are correct. I'm quite sure I'm the previous owner of that cpu.

Sadly, the powerlogix one I have now has died. I think I'm cursed.
 
DZ/015 said:
You are correct. I'm quite sure I'm the previous owner of that cpu.

Sadly, the powerlogix one I have now has died. I think I'm cursed.
Click to expand...
Yep, you are the one I traded with. Apparently, we've both had the same conclusions about it though at different times.

Bunns saved it from being used for target practice. :D
 
eyoungren said:
Bunns saved it from being used for target practice. :D
Click to expand...

Don't count that out just yet :)

In all serious, though, now that I've found the 1.33 setting, I know I can always have that to fall back to since it seems to cause zero problems at least over the past day of on-and-off use. 1.33ghz(even without L3) is still faster than any factory Quicksilver.

Plus, ammo is too expensive these days :) (admittedly I'm less effected by it since I reload, but still powder, bullets, and primer have all gone up and powder in particular can be hard to find). I'll save my target practice for dead hard drives.
 
I went "shopping" today in Surplus for fans, and picked up a whole range of sizes from 40mm to 120mm.

The big blower-type fan should be interesting-I'm thinking about sticking it in the ZIP bay, and it should draw a lot of air out of the case.
 

Attachments

  • image.jpeg
    image.jpeg
    1.4 MB · Views: 128
bunnspecial said:
i went "shopping" today in surplus for fans, and picked up a whole range of sizes from 40mm to 120mm.

The big blower-type fan should be interesting-i'm thinking about sticking it in the zip bay, and it should draw a lot of air out of the case.
Click to expand...

FANS FANS YAY FANS!!!! :D :D :D
qqqqqqqq
 
You must log in or register to reply here.
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.
Top Bottom
Back