• Hello MLAers! We've re-enabled auto-approval for accounts. If you are still waiting on account approval, please check this thread for more information.

750FX to 750FL Interposer

MP G3s are not a thing - the 750 family does not have the hardware to implement SMP cache coherency.
Other than those weird PCI cards that don't really count, because it is just several G3s that happen to be on speaking terms in the same neighborhood.
 
I've been on vacation for the past few weeks, so haven't worked on it a whole lot since, but should start back up again soon. I will say it has been a great learning experience using KiCAD. I've made a lot of it, then started over twice. But I think it's helping me find the easiest and most efficient ways to do things. I have designed PCBs in the past, but they were analog only, with the highest pin count chip being like 8 pins. It was very forgiving when I did things inefficiently on those.

I am hoping to get the Broadway CPU used in the Nintendo Wii to work; anyone see any additional issues there compared to the 750FX? One thing I am concerned about is the Power Mac firmware not wanting to work with the CPUs. Is this an issue I am likely to encounter, and if so, might some firmware upgrade or enabler software from Sonnet/XLR8/etc work?
 
MP G3s are not a thing - the 750 family does not have the hardware to implement SMP cache coherency.

Well, all except one, the Espresso. Nintendo really wanted that G3 backwards compatibility, yet multiprocessing to boost performance, so IBM made SMP happen with the 750 architecture on that one tri-core chip. Unsurprisingly, it was a nightmare to program for.

I am hoping to get the Broadway CPU used in the Nintendo Wii to work; anyone see any additional issues there compared to the 750FX? One thing I am concerned about is the Power Mac firmware not wanting to work with the CPUs. Is this an issue I am likely to encounter, and if so, might some firmware upgrade or enabler software from Sonnet/XLR8/etc work?

Last time I checked, I think I came to the conclusion that it should work. It has some stuff on the same pins that the 750FX uses for dynamic PLL, but if its in the place of a non-dynamic PLL chip, that doesn't matter. I personally actually have a Broadway CPU, in the hopes of one day doing this myself, but I haven't put the time into CAD yet to get there.
 
Well, all except one, the Espresso.
That's not a 750 - it's 750-based much in the same way the 750 is 603e-based. Similar, but not the same.
I am hoping to get the Broadway CPU used in the Nintendo Wii to work
Reading and understanding datasheets, and how CHRP(-like) machines boot would be more productive than random guessing about whether a custom CPU would hypothetically come up on a yet-to-be-designed, ridiculously complex interposer. Heck, maybe even just looking at what the patch does.
 
hoping != guessing
Several similar interposers have been designed and successfully implemented, not sure what issue you see with this specific one is. I am fairly far along with the design, and it does not seem "ridiculously complex" so far.
 
Hoping is the same as guessing in that both are eliminated by proper design. An engineer doesn't design a locomotive and either hopes or guesses it won't go off the rails. That's my sermon for today.

In any case, what are you using for a stackup and rules?
 
Reading and understanding datasheets, and how CHRP(-like) machines boot would be more productive than random guessing about whether a custom CPU would hypothetically come up on a yet-to-be-designed, ridiculously complex interposer. Heck, maybe even just looking at what the patch does.
The Broadway was also sold in different clockspeeds as the 750CL. Of which datasheets do exist. My comments on that are based on my reading of the datasheets a long time back.
 
Back
Top