At 21:10 -0700 07/09/2002, Kennedy Brandt wrote: >With a G3, the maximum processor speed is 8x the system bus speed, or 320MHz >in an Alchemy system and 400MHz in a Tsunami system. G4's may apply an even >higher multiplier. > >Of course, if anyone's actually running a G3 at faster than 400MHz in an >SuperMac Tsunami system, then there are some holes in the plot.
Kennedy, IBM came out with a later revision of the G3 that has a 10X multiplier. They dropped the lowest multiplier (1X, 2X?) to make room for it in the PID. It jumps straight from 8X to 10X though with no intervening multipliers. So there are now 500 MHz G3 upgrades that run 50 MHz bus, 10X multiplier and 500 MHz CPU speed. The early G4 processors (I think 7400 and 7410) have up to a 9X multiplier. So 450 MHz was the maximum for them with a 50 MHz bus speed. The later G4s (7450 and something else?) go up to 16X but they have a different cache scheme, wherein they have a 256K (?) L2 cache on the chip and a controller and pins on the chip for an external L3 cache of up to 2 or 4 MB. I'm a little hazy on the details. Apparently this additional layer of caching made it more difficult to adapt the newer G4 chips to our older machines. My expectation was that no-one ever would. The 16X multiplier times a 50 MHz bus gives the 800 MHz CPU speed they're advertising. I used 50 MHz for the bus speed here, because that is the nominal maximum for our machines, but many folks have found that higher bus speeds work pretty well. It seems to depend a lot on a person's CPU card. I have an old XLR8 PowerBoost Pro (adjustable 604e card) which runs at least 61.667 MHz bus in every S900 I've tried it in (eight and counting) and also in three PowerTower Pros. My ancient 220/110 G3 card won't do better than 49 MHz no matter what machine I install it in. More modern G3 cards seem to go up about 55 MHz pretty commonly. Personally, I'm kind of interested in the new IBM PPC750FX chip which has a built-in 512K L2 cache (no pins for external cache, so no complications) and up to a 20X bus multiplier. It is supposed to ship in speeds from 700 to 1000 MHz, so one should be able to get up to 1 GHz on our machines with a 50 MHz bus. I don't know what kind of real world performance one would get with such a high bus multiplier and only 512K (as opposed to 1 MB) of L2 cache, but the L2 cache does run at the full core CPU speed because it is on the CPU chip. You wouldn't have a source of information about the design of PCBs with BGAs on them? And a how-to on soldering BGA chips? If I knew what was required when laying out a BGA on a PCB design I could get a ZIF for the 750FX made. But I would still need to attach the chip to the board...Oh, and I would need a supplier for the pin array that goes on teh bottom of the module. The basic design and pinouts for the ZIF module are available on Motorola's site as MPCPCMEC.pdf and the pinout and such for the IBM 750FX is available on IBM's site. So in theory it wouldn't be that tough if one could get the chips from IBM and whip out a design for the PCB in something like Osmond PCB. Jeff Walther -- SuperMacs is sponsored by <http://lowendmac.com/> and... Small Dog Electronics http://www.smalldog.com | Refurbished Drives | Service & Replacement Parts [EMAIL PROTECTED] | & CDRWs on Sale! | Support Low End Mac <http://lowendmac.com/lists/support.html> SuperMacs list info: <http://lowendmac.com/supermacs/list.shtml> --> AOL users, remove "mailto:"; Send list messages to: <mailto:[EMAIL PROTECTED]> To unsubscribe, email: <mailto:[EMAIL PROTECTED]> For digest mode, email: <mailto:[EMAIL PROTECTED]> Subscription questions: <mailto:[EMAIL PROTECTED]> Archive: <http://www.mail-archive.com/supermacs%40mail.maclaunch.com/> Using a Mac? Free email & more at Applelinks! http://www.applelinks.com
