InvestorsHub Logo
icon url

edgarcayce

06/08/03 3:18 AM

#5906 RE: KeithDust2000 #5905

Keith-re:AMD Hypertransport in Apples.
I had heard from a salesperson that the new G5's were going to use Hypertransport but I thought he was getting his technology confused. Guess it's true. I find the 64 bit addressing by HT rather interesting with Athlon 64 getting ready to release.
e
icon url

j3pflynn

06/08/03 7:11 AM

#5909 RE: KeithDust2000 #5905

Keith, that's one more bit of good news!
icon url

yourbankruptcy

06/08/03 9:42 AM

#5911 RE: KeithDust2000 #5905

This is related to the fact that AMD is moving its chip design team to IBM headquarters, right?

icon url

sgolds

06/08/03 12:41 PM

#5917 RE: KeithDust2000 #5905

Keith, Mac G5 & Hypertransport: In the past, Apple has been known to put x86 coprocessor cards in the Mac so that the same machine can run Windows applications concurrently. Don't know if they have any such product right now.

With a Hypertransport motherboard, it would be real easy to implement an MP system that has an AMD processor to run Windows and a G5 for Mac OS. Especially since each processor gets its own local memory. You would not enable the ability to read/write remote memory (which means a low cost Athlon64 would be sufficient), instead loosely coupling the processors so that they can share data through a DMA operation over aHT, probably in the form of IP packets. One processor would act as the I/O service processor (in addition to its other duties) so you would need a simple mailbox protocol to service I/O requests.

No significant work needs to be done to the two OSs because they each have full control over their environment. You just need some driver work to manage the I/O mailbox protocol and one OS needs to host the window for the display of the other OS (or they can each be given the own monitor on a dual-head video card). The keyboard and mouse input goes to whichever OS has the cursor focus.

That product just begs to be developed! :)
icon url

j3pflynn

06/08/03 12:46 PM

#5918 RE: KeithDust2000 #5905

yb, I doubt it. That would be Apple's decision to change their motherboard architecture, not IBM's. There isn't any reason to believe IBM has changed their FSB interface, is there? I'm assuming we're talking chipset connectivity to the rest of the system, not chipset to CPU.
Paul