Quad core G4 upgrade possible?

[Quadraman]

http://www.freescale.com/webapp/sps/site/prod_summary.jsp?code=MPC8641D&webpageId=M98653&nodeId=0162468rH3bTdG8653&fromPage=tax

So, how fast do you think a G4 would run with two of these dual core CPU's at 1.5 ghz??

 

[MacJunky]

Looks like fun! But the existing PC100/133 RAM would totally kill it.

You are talking about plopping some of those into an Apple machine right?

 

[Flash!]

I'd expect the same results as trying to swap an '040 for a '060. Compatible....close...kinda...might boot....

 

[madmax_2069]

the slow system bus and memory of the Mac G4 systems would be the bottle neck keeping a quad G4 upgrade from performing like it could, but it would still be awesome to have and see what its limits can be. cause i would think even on a slow system /memory bus it would still result in a nice performance gain.

but that is IF this was ever made for a Mac system. it would cost a arm and a leg to buy one if a upgrade was ever made for G4 Mac's .

i would love to see what a G4 CPU would do with 128k L1 cache (64 instruction/64k data) , 1MB L2 cache , and 2MB L3 cache. i would love to see if it would increase the performance or decrease it in any way. wouldn't you think that by lowering the L1 cache to 32k (16K I/16K D) that it would drop in performance vs a cpu with 64k L1 ( 32K I/32K D). the link shows that CPU has only 32k total L1 cache, or did i get the wrong ideal on how its stated.

i would love to see a quad G4 cpu upgrade available for Mac. but the price for one would be insanely high. and see who would fork over that much to buy one.

 

[Franklinstein]

With the MPC8XXX chips being highly integrated SOC-type devices, I think it's pretty unlikely that they would scale well to being used in an environment with highly obsolete components, if they could even be rigged to use external memory controllers and bus controllers at all. These chips are designed for use in switches and routers and other high-performance network applications, not desktop computing. I'd bet that it would be easier to hack OS X to run on a POWER6 workstation than it would to engineer an MPC8XXX to work at all, let alone well, in a G4 tower. Hacking OS X for a POWER experience would be the more worthwhile endeavor, I'd say - the POWER6 is INSANELY fast, faster than any upgraded (or even brand-new) Mac could ever hope to be.

Unfortunately, Freescale doesn't seem to care about anything besides the embedded market, and IBM only cares about POWER and whatever project gets them the most money (such as their Sony-Toshiba-IBM partnership for the Cell BE, and their dealings with Microsoft and Nintendo for their chips). I'm afraid that PowerPC is yet another fantastic platform marginalized by poor marketing choices and indifference on the part of their proprietors.