Forums | MacLife

That's the same pic as from the latest macaddict.

MMM MMMM Good

"At up to 1.8 GHz"

"up to 900-MHz processor interface bus"

"data at a rate of up to 6.4 GBps"

...sounds awesome, but that "up to" that keeps popping up sounds rather suspicious.

Not at all. They are saying "up to" because they won't know the exact yields they'll get until the manufacturing is in place.

Keep in mind they are stating the possible performance of the chip when it ships. Over time they'll refine the manufacturing process. At this point though, without the manufacturing in place they can't say exactly what they'll be able to produce. Saying "up to" such and such etc. means that the initial design is capable of reaching these numbers. They just won't know if the first gen hardware will be that fast.

-=Jax=-

The main thing I'm looking for is a faster memory interface. There's an interesting thread over at the xlr8yourmac forums about the G4 being a fast processor hobbled by a slow memory bus. I have seen this also (with Altivec), so a bus that can run potentially 900MHz RAM is going to massively increase app speeds, even if we're seeing 1.8GHz G4s by then from Motorola.

http://www-3.ibm.com/chips/products/powerpc/newsletter/dec2002/newproductfocus2.html

Go there. That's the official news bulletin that IBM gave out. I get excited everytime I read it.

Probably old news for some of you. But I just found the page tonight.

thugmoni

Anyone know off hand how the SPEC tests are optomized? i.e do they favor one instruction set over another?

http://www.aceshardware.com/read.jsp?id=50000336

IBM disclosed that the 1.8 GHz chip will reach 937 in SPECint2000 and 1051 in SPECfp2000. One look at our SPEC Datamine, and you see that this is at the level of a 2.8 GHz Pentium 4...

Anyone know off hand how the SPEC tests are optomized? i.e do they favor one instruction set over another?

http://www.aceshardware.com/read.jsp?id=50000336

IBM disclosed that the 1.8 GHz chip will reach 937 in SPECint2000 and 1051 in SPECfp2000. One look at our SPEC Datamine, and you see that this is at the level of a 2.8 GHz Pentium 4...

Excellent...

We'll start of behind the P4, but I'd expect a rapid catch-up...

The ball will be in our corner, and we'll take it up to the top of the board, king it and bam! Full house!

There's an awful lot of emphasis on the 64 bit factor.  I hope that it runs 32 bit apps at good speed too.

There's an interesting article here: http://www.newsfactor.com/

It all sounds great, until one realises what X11 could mean for Apple.

Something that most Macolytes agree on is that if Apple ported to Intel, or AMD without proprietary locks, it would mean the end of Apple, because people could just get cheaper hardware elswhere and run Mac OS.
In a way, that's what Apple risks with X11. Linux users (like myself) could run our *nix servers and apps on Apple hardware with X11, but why on earth would we? Why pay a hardware premium, and then be saddled with proprietary cases and video cards? Maybe to run all our Mac apps; but we generally don't have any Mac apps (we're the great unwashed  mass of Linux users, right?).

So who's Apple going after with X11? Are they seeking the 'anti-market' of Linux home users? (Apple is the uber-market). Are they going after the corporate market of Linux servers that IBM covets (can Apple compete?) Or are they going for the desktop workstation market where Linux faces it's toughest challenge?

you're missing the point here,

Mac OS X is not just yet-another-unix-with-X11, it's years ahead of any other Unix OS in terms of user interface (Aqua) thanks to a great compositing engine (Quartz). And it now also has the ability to run X11 applications on top of that...

Will Linux and other GNU users be satisfied with Apple hardware, when there are so many less expensive and less proprietary alternatives? How would the 970 fit into this? Is it going to be so far ahead of everyone else that it's not even a contest anymore? Intel's 32 bit will be running on an 800MHz bus soon, and AMD is shipping Opterons already. It's not a slam dunk by any stretch, and if it is shown ,and I think it will be, that competeing 64 bit and 32 bit workstations and servers can outperform the fledgling IBM offering, then it could be disasterous for Apple's workstation and server plans.

that's because Apple's machines are in a completely other league... I too use linux on x86 as a web/file/application server, but I'm really glad I have my Mac to do my daily work. It might not even come close to the speed of my linux-box, but I sure get the job done faster on Mac OS X than on Linux.

Ah, I should clarify.  I understand that the 970 will run 32 bit apps without a problem, but will it be able to achieve the speeds they are saying with 32 bit apps?  That is, will there be a speed difference if I run say, iMovie on a 1.25 GHz G4 vs. on a 1.25 GHz 970?  Or would iMovie need to be optimized for the 64 bit architecture to be faster?

I'm talking on chips with the clock speed.  I have little doubt that a 1.8 GHz 970 will be faster than the current G4 line up/

AltiVec is a trademark of Motorola, Inc., used under license. Additional information can be found at www.ibm.com/powerpc.

So IBM has to pay Motorola to use AltiVec (or at least the word). I doubt IBM would suddenly say, "Hey, lets start using AltiVec even though we poo pooed it early on!" This chip has got to be for someone who uses the tech extensively, or Apple.

AltiVec is a trademark of Motorola, Inc., used under license. Additional information can be found at www.ibm.com/powerpc.

So IBM has to pay Motorola to use AltiVec (or at least the word). I doubt IBM would suddenly say, "Hey, lets start using AltiVec even though we poo pooed it early on!" This chip has got to be for someone who uses the tech extensively, or Apple.

Actually, the latest Linux libraries coming out have Altivec support.  IBM, Moto and Apple all hand their hands in the Altivec design, Motorola just copyrighted the name "Altivec", and Apple uses "Velocity Engine".  IBM just calls it something boring.

As far as the "up to 900Mhz bus".  The 970 uses a bus speed that is run half of what the processor runs, so a 1.4Ghz can have a 700Mhz bus, while the 1.8Ghz can have the 900Mhz.  That is all it means.  I imagine this design is so that the DDR memory will run completely in sync with the processor, optimizing performance and being able to keep the processor fed.

-mark

this chip is going to be upgraded more then the motorola one because apple wont be the main user of the chip. UNIX servers will. because of this it has to be developed faster to match the larger demand. SWEEEET!!

Imagine the memory bus running at the full processor speed, wouldn't that just be the coolest thing? Something to look forward to anyway.