E said:
I read, what some consider to be a rumor, that the next XBoX may have
up to three G5 CPUs. If so, looks like MS will be releasing a version of
the NT kernal (or something) to run on this. Maybe this could develop into
PPC systems of some sort running Windows Longhorn? Some say if the
XBox goes with PPC then the price will drop on the high end PPC chips.
That depends what PPC model the XBox2 uses. MS won't be using the
highest-end chips available because a single PPC970 chip costs more than the
entire console's target price. Thus, XBox2 won't have a strong effect on
the high-end PPC market, IMHO.
WinNT was originally ported to PPC but sales were nil and MS dropped the
port. I'm sure XBox2 will revive at least part of that port, but IMHO it
would take a miracle for them to release a complete retail Windows/PPC
again.
Anyone know how a G5 system running Linux or BSD compares to an
Opteron, Xeon or even an Itanium system running Linux or BSD?
Apple's OS/X is built on a FreeBSD derivative called Darwin. Performance of
a G5 is comparable to an Opteron, though as each side releases new chips the
performance lead will sway back and forth. There really isn't a significant
difference in MIPS per dollar between x86 and PPC today.
Virginia Tech has built a cluster of dual G5 Apples running OS X that
promises to be one of the fastest super computers in the world. I hear that
a dual G5 will blow the doors off of anything from Intel or AMD.
A dual G5 is on par with a dual Opteron. The only reason it may "blow the
doors off" the latter is because dual AMD64 systems aren't cheap and aren't
currently marketed as desktops.
Some seem to think that Intel will eventualy subside while PPC CPUs will
gain market share. Although I kind of root for AMD Athlong-64/Opteron,
it would be nice to see some G5's in the mix.
As someone recently said here, "hardware is cheap, software is expensive."
People buy hardware to run software, so what hardware is dominant at any
time is a reflection of what popular software runs on. Of course, this is a
vicious cycle, since developers target their software for whatever hardware
is popular.
Ironicaly, ARM could come out of no where and steal the show.
It could, but why build a high-end ARM chip when you can just use a PPC?
ARM is currently popular for low-power devices, but if you want a single
architecture with _existing chips_ scaling from embedded devices to desktops
and servers to supercomputers, PPC is the clear winner.
S