You know, somebody should........

Without quoting anything in particular, I have an XLR8 G4, it's clocked in at 500Mhz, I put a fan on it because it makes an uncomfortable amount of heat in a beige G3 tower case. I also have the BUS speed up to 83Mhz, which seems to matter more than increasing the CPU clock any higher. I'll try a higher PCI clock, but I'm not sure 3Mhz more is going to increase video performance enough to matter.
I looked at the Biege board I have some time ago and I think I spotted the oscillator that drives the PCI clock, and some other stuff.

Replacing the oscillator with a 33MHz part should run the PCI Bus at 66MHz, but then we need a 100MHz rated XPC106 to have any hope of running the systems bus at 133MHz as the 2x jumper setting would allow.

The real issue is Perch, the built-in ATI chip, and Mac-io.

Perch isn't really an issue unless you have installed PCI devices on it that will not work at 66MHz, the ATI chip can be removed if it's an issue, but Mac-io is the trouble, it don't not list as 66MHz capable, but that doesn't mean it won't function at 66MHz, but it's iffy and if we lose that, we lose all the built-in devices that sit behind it.

Some people have had luck with 66MHz PCI hacks on other old world Macs, but it's a crap shoot at best.

Whenever I get around to getting a 100Mhz XPC106 reworked onto a Beige board, I'll let everyone know if PCI 66Mhz will work or not, and what the drawbacks are.

The other issue is I've not found any PCI USB/FW cards that list as 66MHz capable, and we hate to lose that functionality. Maybe some of the PCI-E combo cards are, I'll have to check, as I do have a bridge that is 66Mhz capable.

All wonderfully worthless, I just want the fastest Old World Mac known to man!
 
somewhat offtopic but many raid/network cards were specifically compatible with 66mhz slots (although you do have to look out for the 3.3v-vs-5v keying with specific cards) .. even then since consumer kind of motherboards generally only bothered with 33mhz slots I have doubts that there were any other common type of cards for non-33mhz slots (such as eg usb) nevertheless?
 
somewhat offtopic but many raid/network cards were specifically compatible with 66mhz slots (although you do have to look out for the 3.3v-vs-5v keying with specific cards) .. even then since consumer kind of motherboards generally only bothered with 33mhz slots I have doubts that there were any other common type of cards for non-33mhz slots (such as eg usb) nevertheless?
To add to it, the GeForce 6200 PCI lists as 66MHz "Capable", but runs the PCI-X slots in my G5 Dual 2.0GHz in 33MHz mode.

So setting 0x20 in the PCI Status Register may just mean it will work @33 MHz when it detects a 66MHz slot?

So it's "Compatible" and doesn't seem to be "Capable", but I've got a Yikes Logic Board I keep meaning to setup, so let me see if it's just the G5 or the card.....
 
To add to it, the GeForce 6200 PCI lists as 66MHz "Capable", but runs the PCI-X slots in my G5 Dual 2.0GHz in 33MHz mode.

So setting 0x20 in the PCI Status Register may just mean it will work @33 MHz when it detects a 66MHz slot?

So it's "Compatible" and doesn't seem to be "Capable", but I've got a Yikes Logic Board I keep meaning to setup, so let me see if it's just the G5 or the card.....
Check continuity of M66EN with ground. If it's not grounded then it should support 66 MHz.
https://en.wikipedia.org/wiki/Peripheral_Component_Interconnect

I suppose it's possible for the 66MHZ_CAPABLE bit of the Status Register to not match the state of the M66EN pin (which can be held low by any 33MHz device on the same bus) or the card's implementation of that pin.

If a card has M66EN grounded but has 66MHZ_CAPABLE then maybe you can cut the trace on the card that is grounding M66EN (only if the card has 3.3V key which is required for 66 MHz).

There's probably a bit somewhere in the bus bridge or host controller that represents the status of all M66EN of all slots of their downstream bus.

Use lspci for Open Firmware to get all the config info.
 
I mean I could be wrong, but it looks like the Apple OEM 300/+MHz G3 ZIFs use KM736V689A and that is 2.5v, no?

The 400Mhz G3 ZIF seems to use mcm69p737tq3.0 and that is 3.3v.
Seems to me somebody already figured this out, as the 3.3v L2 Memory G3 ZIFs seem to sell for $20-25, even the 400MHz rated CPU's, the the ones with SEC L2 RAM sell for $55-$100.
This is a false assumption. Apple and their suppliers used several different models of cache SRAM, some of which, including the Samsung part number above, are compatible with both 3.3V and 2.5V I/O. However, all the ZIF boards run the cache chips' core and I/O supplies, VDD and VDDQ, at 3.3V.
 
Back
Top